Glossary
Plain-language definitions of the terms our guides and manuals reference, from customer data platforms and marketing automation to the specifics of Bloomreach Marketing (Engagement), Klaviyo, and Trifft.
Each entry gives the definition first, then the practitioner context that makes it useful. Search by keyword or browse the terms alphabetically below.
A
A/B Test
An A/B test is a controlled experiment that serves two versions of an experience to a randomly split audience and compares a chosen outcome to see which performs better.
Abandoned Cart
An abandoned cart is a session where a customer adds items to a cart but does not complete the purchase within a defined window.
Abandoned Cart Flow
An abandoned cart flow is a triggered sequence of messages, across email, SMS, and web, sent after a customer leaves items in their cart without purchasing. Where an abandoned cart is the event, the flow is the automated response that tries to bring them back.
Action Node
An action node is a step inside a scenario that executes something for the customer who reaches it: sending an email, SMS, mobile push, or webhook, or writing to an ad audience.
Activation
Activation is pushing CDP audiences and signals into the downstream tools that act on them: email platforms, ad networks, on-site personalization, and support systems.
Agent Harness
An agent harness is the software around a language model that prompts it, runs the tools it calls, feeds results back, and stops when the task is done.
Agent Memory
Agent memory is any mechanism that lets an AI agent carry information between sessions or steps, since the model itself forgets everything once a call ends.
Agentic CDP
An agentic CDP is an emerging concept where AI agents act on customer data and campaign goals inside or on top of the CDP, deciding and executing rather than only reporting.
Aggregate (Bloomreach)
An aggregate in Bloomreach Engagement is a number the platform calculates from a customer's events, such as total orders, lifetime spend, or the date of the last purchase.
AI Agent
An AI agent is an AI system that can reason, decide, and take actions through tools, rather than only generating text.
AI Assistant
An AI assistant is a conversational AI product, such as ChatGPT or Claude, that answers questions and does tasks in natural language, not from a fixed script.
AI Decisioning (Bloomreach)
AI decisioning is the part of Bloomreach's Loomi AI that picks the best variant or next action for each customer, instead of one winner for the whole audience.
AI Governance
AI governance is the set of rules for safe, compliant, explainable, and controlled use of AI, especially when AI uses customer data to make decisions.
AI in Marketing
AI in marketing is the use of machine learning and generative models to predict customer behavior, personalize content, and automate decisions across channels.
AI Observability
AI observability is logging what an AI system was asked, what it retrieved, which tools it called, and what it answered, so any result can be explained later.
AI Overviews
AI Overviews are Google's AI-generated summaries at the top of search results; they answer the query directly and cite a few sources instead of listing links.
AI Slop
AI slop is low-effort, mass-produced AI-generated content published without review, recognizable by generic phrasing, padding, and an absence of specific facts.
AI-Ready Data
AI-ready data is customer data that an AI feature can use without a person cleaning it up first: every event has one agreed name, every person has one profile, consent is stored on the profile by purpose, and the product catalog matches the events.
analytics_storage
analytics_storage is the Google Consent Mode v2 signal that controls whether analytics cookies and measurement can run for a given user.
Anonymization
Anonymization removes identifiers so that an individual cannot reasonably be re-identified from the data.
Anti-Flicker
Anti-flicker is a pre-render guard that briefly hides page content until experiment variants are applied, so visitors do not see the original version flash before the test version loads.
API Trigger (Bloomreach Marketing)
An API trigger starts a Bloomreach Engagement scenario when an outside system calls a URL with news about a real-world event, such as a canceled flight, a weather alert, or a goal scored.
App Inbox (Bloomreach Marketing)
App Inbox is a Bloomreach Engagement mobile channel that keeps messages in an inbox inside your app, so a customer can read them after dismissing a push notification, or without ever opting in to push.
App Link
An App Link is an Android HTTPS URL, verified to belong to a specific app, so that tapping it opens that app directly to a specific screen instead of falling back to the browser.
Approval Workflow (Bloomreach Marketing)
Approval workflow is a Bloomreach Engagement feature that sends a scenario through up to 20 approval steps, each assigned to a user group, before it can go live.
Async Mode
Async mode is the Bloomreach experiment mode that applies variants after the DOM is ready, hiding the affected content with an async-hide class until the variant is in place.
Attribution
Attribution is the method of assigning credit for a conversion to one or more of the marketing touchpoints that preceded it.
Audience Sync
An audience sync sends a segment from a customer data platform to an advertising platform, matching hashed emails or IDs so the segment can be targeted or excluded in ads. Google calls it Customer Match; Meta calls it Custom Audiences. The sync usually runs on a schedule and only the profiles the ad platform can match become targetable.
Audit Log (Bloomreach Marketing)
The audit log is the Bloomreach Engagement record of who did what, when and where, kept as append-only, gzipped JSON Lines files in a Google Cloud Storage bucket in the region where the activity happened. Each record names the actor (a user, an API key or public access), the method and path called, and for some actions the resource ID, such as a customer, scenario, banner or experiment; records share a `requestID` per user action and a `sessionID` per login. Files stay downloadable for at least 60 days, then move to archive storage where they can no longer be fetched. There is no viewer in the app: the files are built for import into a SIEM such as Splunk or Elastic.
Autosegments (Bloomreach Marketing)
Autosegments is a premium Bloomreach Engagement feature, powered by Loomi, that tries combinations of customer properties and proposes micro-segments, each with its performance against the project average.
Average Order Value (AOV)
Average order value (AOV) is the average revenue per order over a defined period: total revenue divided by the number of orders.
B
Back-in-Stock Flow
A back-in-stock flow messages contacts who asked to hear when a sold-out product returns, triggered by the inventory change rather than by their behavior. It needs two things most flows don’t: a way to record the request per product, and an inventory signal (a catalog or feed update) that can fire the send.
Backfill
A backfill is a one-time load of historical data into a platform so that profiles, segments and models start with history instead of an empty slate. Without it, a new CDP sees every customer as new, lifetime value starts at zero and a churn model has nothing to learn from.
Batch Webhook (Bloomreach Marketing)
A batch webhook is a Bloomreach Engagement webhook mode that bundles up to 1,000 customers into one HTTP request instead of sending one request per customer. High-volume campaigns need it when the receiving API accepts bulk operations.
Behavioral Data
Behavioral data is data about what people actually do across channels, as opposed to static attributes about who they are.
Benchmark
A benchmark is a reference value a campaign result is compared against: the program’s own prior result, a holdout group, or an external industry figure, in that order of usefulness. Industry benchmarks describe other companies’ lists and other companies’ offers.
Bloomreach Content
Bloomreach Content is Bloomreach’s headless content management system (CMS). On bloomreach.com it now sells as the Enterprise CMS under the Loomi AI platform, though practitioners and the documentation still call it Content.
Bloomreach Discovery
Bloomreach Discovery is Bloomreach’s AI-powered search and merchandising product. On bloomreach.com it now sells as Ecommerce Search under the Loomi AI platform. In 2026 Bloomreach renamed it to Search across its documentation, though many practitioners and contracts still call it Discovery.
Bloomreach Marketing
Bloomreach Marketing is the name Bloomreach’s documentation has used since 2026 for Bloomreach Engagement, its marketing automation and customer data product. Same product, new label: Bloomreach states the change does not alter the services, their functionality, or any terms of an agreement. The full entry, history, and how the platform works live under Bloomreach Marketing (Engagement); this entry exists so the new name has a home.
Bloomreach Marketing (Engagement)
Bloomreach Engagement is Bloomreach's customer data platform and marketing automation product. It combines tracking, customer profiles with identity resolution, segmentation, analytics, and campaign execution across email, SMS, push, weblayers, and ads. It is the product people usually mean when they say "we have Bloomreach."
Bloomreach SDK
The Bloomreach SDK is the JavaScript (and mobile) library that collects events and identifies customers on your site or app, then sends that data to Bloomreach.
Bounce Rate (Email)
Email bounce rate is the share of sent messages that could not be delivered.
Brevo
Brevo, formerly Sendinblue, is a marketing and CRM platform for email, SMS, WhatsApp, and chat, aimed at small and mid-sized businesses and priced mainly on send volume. It combines campaigns, automations, transactional email, and a sales CRM in one account.
Browse Abandonment Flow
A browse abandonment flow is a triggered message to a known contact who viewed a product or category and left without adding anything to the cart. It sits one step earlier in the funnel than the abandoned cart flow and depends on the visitor being identified, since an anonymous view has no address to send to.
Business Case (CDP)
A CDP business case is the document that justifies the platform to finance. It lists a few measurable campaigns, each with a baseline, a holdout group, and a conservative estimate, sets them against the full cost of running the platform, and ends with the month the platform pays for itself.
C
Campaign Calendar (Bloomreach Marketing)
The campaign calendar is a Bloomreach Engagement view that shows scheduled scenarios and email campaigns on a month or week grid, with status, message previews, subject lines, and live metrics. It updates each time a campaign is saved.
Campaign Evaluation
Campaign evaluation measures a campaign's effect by comparing the customers who received it with a control group that did not, on revenue, conversions or another goal. The difference between the two groups is the campaign's contribution; the revenue of the recipients alone is not.
Campaign Event Status (Bloomreach Marketing)
A campaign event status in Bloomreach Engagement records what happened to one message for one customer, from enqueued through delivered to opened, clicked, bounced or complained. Each step is a separate `campaign` event on the profile, so one email can produce several statuses over its life.
Catalog (Bloomreach)
A catalog in Bloomreach is a structured dataset of items you import into the platform, such as products, articles, or stores, and reference in campaigns. Each item has an ID and a set of fields, such as name, price, image URL, stock status, or category. You look an item up in Jinja and read its fields, for example catalogs['Products'].item_by_id(product_id) followed by item['display_name'].
Churn Prediction
Churn prediction is forecasting which customers are likely to stop buying, stop engaging, or cancel.
Churn Rate
Churn rate is the share of customers who stop buying or cancel within a period; it is the mirror image of retention rate. In a subscription business churn is an event. Everywhere else it is an absence, and the business has to decide how long an absence has to be before it counts.
Clicked (Campaign Status)
Clicked is the Bloomreach Engagement campaign status recorded when a link inside the message was clicked, including the unsubscribe link. The event carries the link, so clicks can be reported per button. Links routed through the campaign link shortener are tracked the same way.
Clicked Honeypot (Campaign Status)
Clicked honeypot is the Bloomreach Engagement campaign status for a click on a hidden link that no person would see. The click identifies a bot or a security scanner that opens every link. Corporate mail filters do this to check links for malware. Without the honeypot, those clicks would count as engagement. With bot detection on, the platform filters bot clicks out of reporting and marks them `bot = true`.
Client-Side Tracking
Client-side tracking collects event and identity data directly in the browser or app and sends it straight to each vendor, instead of routing it through a server you control.
Cohort Analysis
Cohort analysis groups customers by when they shared a starting event, such as the month of their first purchase, and follows each group over time so you can compare them fairly. A cohort is one such group.
Commerce Media
Commerce media is advertising powered by purchase intent and transaction data, spanning retailers, marketplaces, and commerce platforms.
Complained (Campaign Status)
Complained is the Bloomreach Engagement campaign status recorded when the recipient marked the email as spam and the mailbox provider reported it back through a feedback loop. It is the status mailbox providers weigh most heavily against a sender. Google's email sender guidelines ask bulk senders to keep the spam rate in Postmaster Tools below 0.3%, and ideally below 0.1%.
Complaint Rate
Complaint rate is the share of delivered messages recipients marked as spam. Mailbox providers use it to decide sender reputation for every message after. Google's sender guidelines ask bulk senders to keep it under 0.1% and never reach 0.3%, as measured in Postmaster Tools.
Composable CDP
A composable CDP is a CDP assembled from modular, best-of-breed tools rather than one packaged, all-in-one platform.
Consent Data
Consent data is the record of what a customer has allowed or denied, expressed by purpose, that every downstream tool must honor before acting.
Consent Group (Brevo)
A Brevo consent group is a topic a contact opts into, such as newsletter, offers, or product updates. Campaigns and automations target a group instead of a list. Once groups are active, the one-click unsubscribe in the email header removes the contact from the targeted group only, not from everything. Brevo introduced consent groups in its spring 2026 release.
Consent Management Platform (CMP)
A consent management platform (CMP) is the tool that collects and enforces privacy consent, typically starting with the cookie banner.
Contact Policy
A contact policy is the one-page rule set that limits how many messages one customer receives across all channels and campaigns. It also sets when nothing is sent, and which message wins when two qualify on the same day. It lists what counts as transactional and who is suppressed from what.
Content Block
A content block is a reusable piece of a message template, such as a header, footer or product grid, maintained in one place and inserted into many campaigns. Change the block and every template that uses it changes with it.
Context Window
A context window is the maximum text, in tokens, a language model can take in at once, including the instructions, the data, and its own answer.
Contextual Personalization (Bloomreach Marketing)
Contextual personalization is a Bloomreach Engagement feature, powered by Loomi, that picks the best variant for each customer from their context, instead of one winning variant for the whole audience. At decision time it reads the customer's device, location, time of day, and browsing and purchase history. It serves the variant most likely to reach the campaign goal. The outcome feeds back into a model that retrains in the background. It supports up to 10 variants across weblayers, email campaigns and scenario channels. Webhooks and ad audiences cannot be optimization targets. It is a Premium tier feature. It only pays off on always-on campaigns, because a one-off send gives the model nothing to learn from.
Conversational Channel
A conversational channel is a chat, messaging, or voice interface, staffed by a person or a bot, treated in a customer data stack like any other channel. It needs the customer's identity, consent, and history flowing in. It needs the conversation's events, handoffs, and suppression signals flowing out to the profile.
Conversions API
A conversions API sends conversion events from your server directly to an ad platform instead of relying on a browser pixel, so the events survive ad blockers and cookie limits. Meta's Conversions API and Google's enhanced conversions are the common ones. The same purchase is typically sent twice, once from the browser and once from the server.
Cookie (Bloomreach Marketing)
The Bloomreach Engagement cookie is a soft ID the platform creates for each browser or device, so anonymous visits are tracked to a profile until a hard ID such as an email is known.
Cookie Deprecation
Cookie deprecation is the set of browser and privacy changes that make third-party cookies less useful for tracking people across sites. The marketing environment this creates is often called the post-cookie era, or cookieless marketing.
Cookie Tracking Modes (Bloomreach Marketing)
Cookie tracking modes are the four ways the Bloomreach Engagement JS SDK stores the anonymous visitor identity: default, shared cookies, custom tracking domain and private tracking domain. Every mode keeps the first-party `__exponea_etc__` cookie set by the SDK. Each mode adds a backend cookie that differs by mode. In default mode it is `xnpe_<project-id>`, set as a third-party cookie on the API domain. In shared cookies mode it is `__EX__<group-id>__`. In custom tracking domain mode it is `xnpe_<project-id>`, set on your own domain. On each page load the SDK sends a `system/ids` command with every known ID. The backend then resets the cookie so its expiration always matches the SDK setting, three years by default. The documentation discourages shared cookies mode because browser privacy rules block it widely.
Copilot
A copilot is an AI assistant built into a tool you already use; it drafts, suggests, or explains inside that tool while a person stays in control.
Cost of Error
Cost of error is the test for whether a task can be handed to an AI agent unsupervised. It asks two questions: how expensive a wrong answer is (a wasted hour, a wrong report, a sent message, a legal exposure), and whether the error is visible before it costs anything or only after. Cheap and visible goes to the machine; expensive and invisible stays with a person.
Cross-Sell and Upsell Flow
A cross-sell or upsell flow is an automated journey that offers a related product or a higher tier after a purchase. It is timed to when the customer is likely to need it. Cross-sell adds a complementary product; upsell moves the customer to a better version of what they already bought.
Cross-Site Scripting (XSS)
Cross-site scripting (XSS) is a security vulnerability where an attacker injects malicious script into a page that other users' browsers then execute.
CTOR
CTOR, click-to-open rate, is clicks divided by opens: how well the content of an email performed among the people who saw it. Click rate measures the whole send, including subject line and deliverability; CTOR isolates what happened after the open.
Custom Events (Bloomreach Marketing)
Custom events are the events a business chooses to track in Bloomreach Engagement, with attributes the business defines itself. Typical examples are `purchase`, `purchase_item`, `view_item`, `view_category`, `cart_update`, `checkout` and `consent`. Unlike system events, their names and attributes vary from project to project because each integration decides what to send. The SDK adds `location`, `os`, `device` and `browser` to every tracked event by default. The common failure is a `purchase` event whose `total_price` or `purchase_status` is named differently from the documented convention. Dashboards and predictions then cannot find the field until data mapping points them to it.
Custom Prediction (Bloomreach)
A Bloomreach custom prediction is a model you define yourself from four parts. A target filter describes the customers who did the thing you want to predict. An eligible-customer filter decides who gets scored. Then come the features the model may learn from, and the algorithm settings. The output per customer is a probability (yes or no), a number (regression), or a best-fit class. Custom predictions sit alongside the prebuilt templates for purchase, churn, email open, and in-session intent.
Custom Tracking Domain
A Custom Tracking Domain (CTD) is a Bloomreach subdomain you own that serves Bloomreach’s tracking API, so the tracking cookie is set through an HTTP response header instead of by JavaScript. On Safari, that can avoid the older JavaScript-set cookie cap, though Safari 16.4 and later can still cap server-set cookies when the tracking host resolves through third-party CNAME or IP patterns (covered below).
Customer Attribute
A customer attribute is a stored property on a customer profile, such as name, email, country, or a computed value like lifetime value.
Customer Data & Experience Platform (CDXP)
A customer data and experience platform (CDXP) is a CDP plus the activation layer: a unified customer-data foundation combined with journeys, personalization, and campaign execution built on top.
Customer Data Architecture
Customer data architecture is the design of how customer data is modeled, stored, and connected across systems, so it can be trusted and used for marketing and analytics.
Customer Data Infrastructure (CDI)
Customer data infrastructure (CDI) is the technical foundation for collecting, transforming, and routing customer data across systems.
Customer Data Platform (CDP)
A customer data platform (CDP) is packaged software that unifies customer data from every source into one persistent, real-time profile that other tools can use. It solves the problem of scattered data and identity across channels: web, app, email, ads, point of sale, call center, loyalty. Where a CRM organizes work around known contacts, a CDP unifies behavior and identity across everyone, known and anonymous, and makes the result available for segmentation, personalization, and measurement.
Customer Filter (Bloomreach Marketing)
A customer filter is a Bloomreach Engagement set of conditions on customer attributes or event history that decides which customers an analysis, campaign audience or scenario node includes. Conditions combine with AND and OR. A set of conditions can be saved as a template that pre-fills the same logic in scenarios, email campaigns and the customer list. Two details catch people out. A report's global date filter does not reach the event conditions inside a customer filter. And a template is a copy, not a live reference, so editing it leaves earlier uses unchanged.
Customer Intelligence Loop
A customer intelligence loop is the cycle where a system reads customer data, decides on an action, acts, measures the result, learns, and repeats.
Customer Journey
A customer journey is the sequence of stages and touchpoints a customer goes through: anonymous visitor, lead, buyer, repeat customer, advocate.
Customer Lifetime Value (CLV / LTV)
Customer lifetime value (CLV or LTV) is the estimated long-term value of a customer to the business.
Customer Merging
Customer merging combines two profiles into one when they are found to belong to the same person. The usual trigger is a shared hard ID such as an email or customer number. The common case is an anonymous profile tracked under a cookie that logs in: its browsing history joins the known profile. In most platforms a merge cannot be undone.
Customer Relationship Management (CRM)
A customer relationship management (CRM) system organizes the work and relationships around known contacts: sales pipeline, service cases, and interaction history.
Customer Retention
Customer retention is the practice of keeping existing customers active and purchasing over time, usually measured as the share of customers who remain across a given period.
Customer Segmentation
Customer segmentation is the practice of dividing customers into groups by shared traits or behavior, so campaigns and journeys can target each group differently.
Customer Token (Bloomreach Marketing)
A customer token is a signed JSON Web Token that proves which customer IDs a device may act for. Bloomreach Engagement APIs and SDKs accept it in place of raw customer IDs for features such as App Inbox. Your backend generates it after verifying the login. The `sub` claim holds the customer ID types and values. The `kid` claim names the private API key. The signature is HS256, keyed with the SHA-256 checksum of that key's API secret. The `exp` claim sets an optional expiry. Bloomreach offers no token-generation API because only the website or app backend can confirm who is logged in. The API secret must never ship to a browser or app, and it cannot be recovered once the key is created.
Cyber Resilience Act (CRA)
The Cyber Resilience Act (CRA) is the EU regulation setting security rules for hardware and software sold in the EU; reporting duties apply from September 2026.
D
Data Act
The Data Act is the EU regulation, applying since September 2025, that gives users access to connected-product data and eases switching between cloud services.
Data Audit (CDP)
A CDP data audit, also called a tracking audit, is a short, structured check of what data arrives, whether each customer is one profile, what can legally be sent, and what runs. It ends in a ranked fix list.
Data Clean Room
A data clean room is a privacy-safe environment where two or more parties analyze combined data without either exposing raw, user-level records.
Data Contract
A data contract is a written agreement between the team that produces data and the teams that consume it, fixing the schema, meaning, freshness and owner of each field. It turns 'the purchase event' from something everyone assumes into something someone is responsible for.
Data Controller
A data controller is the organization that decides why and how personal data is processed under the GDPR, and so carries the main legal responsibility for it.
Data Exfiltration
Data exfiltration is the unauthorized movement of data out of a system; with AI it includes an agent being tricked into sending data where it shouldn't go.
Data Flushing (Bloomreach Marketing)
Data flushing is how the Bloomreach Engagement iOS and Android SDKs upload cached events, sessions and customer properties to the API. The flush mode sets when that upload happens. The default mode sends each item as soon as it is tracked. The other modes send when the app leaves the foreground, at a set interval (and on app close), or only when the app calls the flush method itself. The SDK flushes only on a stable network connection, so after a connection error the data stays cached on the device until a later flush succeeds. A tracked event therefore has no fixed arrival time in the platform: a periodic or manual mode can delay it by minutes or until the next app open.
Data Governance
Data governance is the set of rules for data quality, access, privacy, ownership, and usage across an organization.
Data Ingestion
Data ingestion is the way records enter a customer data platform: real-time event streams from tracking, API calls from other systems, or scheduled batch imports from files and databases. Each method has a different delay, cost and failure mode, and a typical setup uses all three.
Data Lake
A data lake is storage for data in its raw, original form: event streams, logs, exports, and files, stored cheaply and structured later, if at all. Data engineers and data scientists work here; the lake is where data lands first, not where campaigns run from. Raw data usually has to be cleaned and structured before marketing can act on it. A lakehouse is a lake with warehouse-style querying layered on top.
Data Lakehouse
A data lakehouse is an architecture that combines the low-cost scale of a data lake with the structure and querying of a data warehouse.
Data Latency
Data latency is the delay between an event happening in a source system and the same event being usable on the customer profile. A website tag is near real-time; an ERP export is nightly; a loyalty file is weekly. The slowest source a campaign depends on sets that campaign’s latency.
Data Layer
A data layer is a JavaScript object on a web page, usually window.dataLayer, where the site pushes events and values for a tag manager to read. Google Tag Manager reads from it; Consent Mode updates, page context, and ecommerce events all arrive as pushes into the same array.
Data Lineage
Data lineage is visibility into where data came from, how it was transformed, and where it moved across systems.
Data Location (Bloomreach Marketing)
Data location is the Google Cloud region where a Bloomreach Engagement instance stores its customer data. It is chosen at setup from the sites the documentation lists: Belgium or London for EMEA, Montréal or Iowa for North America, and Sydney for APAC. The region is picked to fit the customer's business needs when the instance is set up. Google Cloud replicates the data in real time to at least two geographically separate data centers, and an automated system runs full-system backups daily. One exception matters for compliance. Files in Asset manager sit outside this infrastructure and their location is not guaranteed, so the documentation advises against storing sensitive data there.
Data Management Platform (DMP)
A data management platform (DMP) is adtech software for collecting and activating anonymous or pseudonymous audiences, mostly for advertising. It dealt largely in third-party data.
Data Mapping (Bloomreach Marketing)
Data mapping is the Bloomreach Engagement setting in Data manager that links your own event names, customer properties and catalog fields to the platform's predefined formats. Built-in features can then read your data without you renaming it. It covers events such as `Purchase`, `Purchase item`, `Add to cart`, `View item` and `Search`, properties such as `first_name`, and the main product catalog with its reserved fields. Performance dashboards, recommendations, predictions and use-case templates all depend on it. The platform asks you to map the `Purchase (completed)` event and its `total_price`, `currency` and `total_quantity` attributes in advance. That mapping decides how the out-of-the-box dashboards process data.
Data Mart
A data mart is a purpose-built subset of the data warehouse prepared for one team’s questions. The marketing mart holds customer, order, and campaign tables shaped for marketing use, without the risk tables and logistics feeds the rest of the warehouse carries. Faster queries, simpler models, and a smaller blast radius when something upstream changes.
Data Minimization
Data minimization is the practice of collecting only the data necessary for a defined purpose.
Data Model (CDP)
A CDP data model is the agreed structure of customers, events, attributes, and catalogs in the platform: what exists, what it is called, what it carries, and how it joins. It is written in the tracking plan and it is the first thing a migration should produce, with the engineers who will send the data.
Data Ownership
Data ownership names the person or team responsible for a data source or field: that it arrives, is correct, and changes to it are announced before they ship. In a CDP program every event source has an owner on the engineering side and the platform as a whole has an owner on the marketing side.
Data Processing Agreement (DPA)
A data processing agreement (DPA) is the contract the GDPR requires between a controller and each processor, setting what the processor may do with the data.
Data Processor
A data processor is an organization that handles personal data on a controller's behalf and only on its instructions, such as a CDP, email platform, or agency.
Data Quality
Data quality is the accuracy, completeness, consistency, and freshness of data.
Data Residency
Data residency is the requirement that certain data be stored and processed in a specific country or region, which decides where a platform's instance and its backups may run. It comes from law, from a client contract or from a regulator, and it is settled before the platform is bought, not after.
Data Retention
Data retention is the rule for how long a platform keeps each kind of customer data before deleting or anonymizing it. Privacy law asks for a limit that fits the purpose; platforms charge for what they store; models need enough history to learn. The retention policy is where those three meet.
Data Warehouse
A data warehouse is storage for data that has been cleaned, deduplicated, and modeled into tables with agreed meanings, ready for SQL queries and reporting. This is where “one version of the truth” lives and where BI dashboards point. When the marketing platform’s numbers and finance’s numbers disagree, the warehouse is the referee. Common platforms include BigQuery, Snowflake, and Redshift.
Decision Engine
A decision engine is a system that ranks what each customer should be offered next: which product, message, channel, and moment. Decisions are computed centrally, from a shared data platform, so every channel makes the same promise; the alternative is a call center offering a discount that the email journey knows nothing about. The engine can be a vendor product or an in-house model stack.
Deliverability
Deliverability is whether your email actually reaches the inbox rather than the spam folder or a block, unlike delivery, which only means the receiving server accepted it.
Delivered (Campaign Status)
Delivered is the Bloomreach Engagement campaign status recorded when the email provider confirmed the receiving mail server accepted the message. The receiving server defines it, so delivered does not distinguish the inbox from the spam folder.
Demo Project
A demo project is a separate platform project configured to match production, with the same event names, ID types, consent categories, and catalog schema. It is loaded with a pseudonymized slice of real history, not a vendor sample. Scenarios, templates, and weblayers are built and tested there and promoted to production from the same spec.
Destination
A destination is any system a customer data platform sends audiences or events to: an ad platform, an email tool, a warehouse, a support desk, a webhook. Sources bring data in; destinations are where the data does something.
Deterministic Matching
Deterministic matching links records using confirmed identifiers, such as a login email or account ID. The match is treated as certain.
Digital Experience Platform (DXP)
A digital experience platform (DXP) manages and delivers websites, content, portals, and other digital experiences across channels.
Digital Markets Act (DMA)
The Digital Markets Act (DMA) is the EU regulation, in effect since March 2024, imposing conduct rules on designated gatekeepers such as Google, Meta, or Apple.
Digital Services Act (DSA)
The Digital Services Act (DSA) is the EU regulation, fully in effect since February 2024, setting content, advertising, and design rules for online platforms.
DKIM
DKIM is a cryptographic signature added to each outgoing email and checked by the receiver against a public key in your DNS. A match proves the message came from your domain and wasn't altered on the way. Each sending platform signs with its own key, published in DNS under its own name, called a selector.
DMARC
DMARC is a DNS-published email policy that tells receiving servers what to do with messages that fail authentication: monitor them (p=none), send them to spam or quarantine (p=quarantine), or reject them (p=reject). It is published at _dmarc.yourdomain.com. A missing DMARC record makes your domain easier to spoof and your marketing email is more likely to land in spam, and a p=none record only monitors. It does not enforce anything.
DORA
DORA is the EU's Digital Operational Resilience Act, applying since January 2025, setting ICT risk, incident, and vendor oversight rules for financial firms.
Double Opt-In
Double opt-in is a signup process in which the subscriber must click a confirmation link in an email before being added to the list. The click proves the address is real and belongs to them. Single opt-in adds the address the moment the form is submitted.
DPIA
A DPIA is a data protection impact assessment, the written risk analysis the GDPR requires before high-risk processing such as large-scale profiling.
DPO
A DPO is a data protection officer, the person the GDPR requires some organizations to appoint to oversee compliance and deal with regulators and the public.
DSAR
A DSAR, data subject access request, is a person's request under the GDPR to see, correct, export or delete the personal data an organization holds about them. The organization has a month to respond. The answer has to cover every system that holds the person's data, including the ones the request didn't arrive through.
Dynamic Content
Dynamic content is content that changes based on customer data or context, so different people see different versions of the same placement.
Dynamic Wait Time (Bloomreach Marketing)
Dynamic wait time is a Bloomreach Engagement wait node setting where a Jinja expression (the platform's templating language) computes the delay separately for each customer who reaches the node, instead of one fixed interval for everyone. The expression can read metrics, aggregates, trigger event attributes or any personalization value. The usual job is spreading a large send over time with a random number such as `{{ range(0,60) | random }}`.
E
Earn and Burn
Earn and burn describes the two sides of a points program: how members accumulate points and how they redeem them. The ratio between the two shows whether the program is used. Points earned and never redeemed are a cost on the balance sheet and a member who hasn't found a reason to come back.
Email Click Heatmap (Bloomreach Marketing)
The email click heatmap is a Bloomreach Engagement evaluation view that overlays the share of recipients who clicked each link on a preview of the email. You find it under `Click map` in an email campaign or scenario. It can show click rate (`unique_clicks` divided by `unique_deliveries`), unique clicks or total clicks. It counts clicks by the link's position in the email structure, not by URL, so personalized product slots still add up.
Email Deliverability Dashboard (Bloomreach)
The email deliverability dashboard is a Bloomreach Engagement channel performance view that shows delivery rate, bounce rates, complaint rate, open rate and click-through rate from your own sending. Next to those it shows reputation, spam rate and authentication signals pulled from Google Postmaster Tools and Microsoft SNDS, the two providers' free sender-reputation services. It sits under Overview > Channel performance > Email. It splits into four sections: deliverability overview, failed sends and bounces, audience health, and external deliverability data. The external section needs a connected Postmaster Tools or SNDS integration before it shows anything.
Email Domain Report (Bloomreach Marketing)
An email domain report is a Bloomreach Engagement report that breaks sent, delivered, bounced, open and click rates down by the recipient's mailbox provider. A reputation problem at one provider then stands out from a general content problem. It isn't a built-in view. A scenario first writes an `email_domain` attribute to each profile with `{{ customer.email.split('@') | last }}`. It runs once with a `Now` trigger, then daily with a `Repeat` trigger to cover new addresses. A report then groups campaign metrics by that attribute.
Email Editors (Bloomreach Marketing)
Email editors in Bloomreach Engagement are the two ways to build an email. The visual builder is a drag-and-drop editor based on BEE, with Settings, Rows and Content tabs. The HTML builder gives full control over HTML and CSS, with code, edit and view modes. Both support Jinja personalization, language versions and HTML blocks. HTML blocks are reusable snippets such as headers and footers, stored in the Asset Manager and inserted as dynamic content. A block can carry typed parameters that the editor fills in per campaign. Rows in the visual builder can carry a display condition built from a segment, which the editor turns into a Jinja `if` around the row.
Email List Hygiene Filter (Bloomreach Marketing)
The email list hygiene filter is a Bloomreach Engagement project setting that excludes unengaged or inactive subscribers from every marketing email at once, without adding an exclusion to each campaign. You switch it on under project settings for the email channel. It ships with a default filter, which you can edit. The default drops three groups. The first is subscribers who last engaged over 180 days ago and were mailed in that period. The second is subscribers who engaged 90 to 180 days ago and were mailed in the last 7 days. The third is subscribers who signed up over 90 days ago and never engaged. Transactional and general-consent email still goes out. Excluded customers don't appear in the recipient count on the `Test` tab, which surprises people the first time.
Email List Validation (Bloomreach Marketing)
Email list validation is a Bloomreach Engagement setup that passes each new address to Mailgun's validation API before it enters a mailing program. Mailgun checks syntax, provider-specific rules, typos such as `.vom` and spambot patterns. The result is one of four values: `Deliverable`, `Undeliverable`, `Do not contact` or `Unknown`. A scenario that listens for subscribe events, or runs weekly, writes it to the customer profile. Deliverable addresses go straight to the welcome flow. Undeliverable and do-not-contact addresses are never added. Unknown ones are sent through double opt-in. Your Bloomreach contact enables the integration, and Bloomreach offers a ready-made use case for the scenario.
Email Suppression List (Bloomreach Marketing)
Email suppression lists are Bloomreach Engagement blocklists of addresses and domains that the email node checks at send time. It checks the exact address first, then the domain, and skips a match with a `suppressed` campaign event. There are two lists. The global suppression list is maintained by Bloomreach for spam traps and dead domains. It is shared by every Bloomreach project and blocks everything, including transactional and general-consent email. The custom suppression list is per project, empty by default, and holds up to 5,000 entries. It blocks normal campaign email only; messages sent under general consent or through the transactional API still go out.
Email Warm-Up
Email warm-up is the gradual increase of sending volume from a new IP address or domain so that mailbox providers can build a reputation for it before full volume arrives. A sender with no history is treated as suspect; sudden volume from it looks like a spam run and gets throttled or blocked. Warm-up ramps the daily volume over weeks, starting with the recipients most likely to open, and watches each provider's response before stepping up.
Embedding
An embedding is a list of numbers representing the meaning of a text, image, or product, so similar items sit close together and can be compared mathematically.
Enhanced Web Targeting (Bloomreach Marketing)
Enhanced web targeting is a Bloomreach Engagement feature that shows weblayers and experiments based on page variables such as a dataLayer value, a URL parameter, a global JavaScript variable or the return value of custom JavaScript. A technical user defines the variables once under project settings and gives each a friendly name. Marketers then build rules in the `Show on pages with variables` field using operators like equals, contains or in between. The JS SDK evaluates the rule at display time, so the variable must already exist on the page. A missing value compares as undefined. URL conditions are checked first, and page variable conditions are combined with AND. For single-page apps, an experiment can also wait for a named `dataLayer.push` event instead of page load. That is how a cart-value banner reacts to a cart that changes without a reload.
Enqueue Failed (Campaign Status)
Enqueue failed is the Bloomreach Engagement campaign status recorded when a message couldn't be handed to the email or messaging provider. The cause is a connection error or an error on the platform side, such as a broken Jinja template. The event's `message` attribute holds the error. A customer with no email address on the profile doesn't produce this status.
Enqueued (Campaign Status)
Enqueued is the Bloomreach Engagement campaign status recorded when the platform has handed a message to the email or messaging provider for sending. It is the first status a message gets and the closest thing to a send count: everything that was supposed to go out and was not suppressed.
ePrivacy Directive
The ePrivacy Directive is the EU law behind cookie consent: storing or reading anything on a user's device needs consent unless strictly necessary.
EU AI Act
The EU AI Act is the European Union's law on AI, in force since August 2024, that sorts AI uses by risk and phases in duties for each tier through 2028.
EU-US Data Privacy Framework
The EU-US Data Privacy Framework is the 2023 arrangement under which personal data may flow from the EU to US companies that self-certify to its rules.
European Accessibility Act (EAA)
The European Accessibility Act (EAA) is the EU directive, applying since June 2025, requiring e-commerce, banking, and other consumer services to be accessible.
Evals
Evals are repeatable tests that score an AI system's outputs against expected results, so a change to the model, prompt, or data is measured, not eyeballed.
Event Data
Event data is data generated by actions, such as clicks, page views, purchases, logins, and searches.
Event Segmentation (Bloomreach Marketing)
An event segmentation is a Bloomreach Engagement definition that sorts the occurrences of one event type into named buckets based on their attributes. For example, it can group `session_start` by `device` into mobile, desktop and tablet. Once saved, it appears as an extra attribute of that event, so a report can select the bucket instead of listing every raw value. Each event lands in exactly one bucket. Where conditions overlap, the first segment from the left wins. Conditions inside one segment combine with AND. A segmentation covers a single event type and cannot be read in Jinja.
Event Taxonomy
An event taxonomy, also called an event schema, is the agreed list of events a platform tracks, with the name and the properties each one must carry. Every system that sends data shares it: web tracking, the mobile app, the order system, and the loyalty tool all write to the same profile, and the taxonomy keeps them writing the same thing.
Event-Based Usage (Bloomreach Marketing)
Event-based usage is the Bloomreach Engagement billing model metered on two numbers. Monthly processed events (MPE) is every event tracked, imported or generated in a calendar month. Maximum stored events (MES) is the peak number of events held in the account at any point. Campaign sends, web tracking, ad audience syncs, weblayer shows and frequency-policy suppressions all generate events that count. Customer attribute updates, catalog imports, exports and predictions don't. Duplicates count each time they arrive. Deleting or expiring events lowers MES only, never MPE. The dashboard under Overview > Bloomreach usage updates every 48 hours. Accounts on profile-based pricing, introduced in May 2026, use a separate Profile usage dashboard instead.
Experiment (Bloomreach)
An experiment in Bloomreach Marketing (Engagement) is the feature for running on-site A/B tests and personalization: serving different variants of a page or element, then measuring which performs better.
Exponea
Exponea is the former name of [Bloomreach Engagement](/platforms/bloomreach). It began as a Slovak-founded customer data platform and marketing automation product, originally launched as Infinario. It took the Exponea name in 2016, and Bloomreach acquired it in January 2021. The product lives on as Bloomreach Marketing, renamed from Bloomreach Engagement in 2026. The old name still shows up in live implementations. The `api.exponea.com` tracking endpoint, `__exponea` cookie names and `exponea.start()` snippets all predate the rebrand and keep working.
Expression (Bloomreach)
An expression in Bloomreach is a formula-based derived attribute: it combines customer attributes, aggregates, functions, and operators into a single computed value.
External ID (Bloomreach Marketing)
An external ID is a customer identifier you define in Bloomreach Engagement, such as `registered` (a hard ID) or `cookie` (a soft ID). It maps records from your own systems to a customer profile. Every profile must carry at least one, and a project can define any number of them, each under its own name. Identifiers are case-sensitive, so `Email@example.com` and `email@example.com` create two profiles unless the lowercase transformation is on. Once set, an identifier cannot be renamed or removed, and a soft ID cannot be promoted to a hard ID.
F
Failed (Campaign Status)
Failed is the Bloomreach Engagement campaign status recorded when communication with the sending provider broke. It is also recorded when a scenario node failed while running for a customer, for example on a Jinja evaluation error. The second case logs a failed event although nothing was sent, which is why failed can appear on a campaign that never reached a provider.
Feature Leakage
Feature leakage is when a model learns from a variable that is a consequence of the target or of the treatment. For example, a “received retention email” feature predicts churn because the retention email was sent to at-risk customers. The model then predicts your past campaigns instead of your customers, and its accuracy score looks high.
Fine-Tuning
Fine-tuning is further training of an existing AI model on a specific dataset so it adopts a style, format, or domain behavior without being built from scratch.
First-Party Data
First-party data is data you collect directly from your own customers and channels: site and app behavior, purchases, and consented profile information.
Flow (Klaviyo)
A flow in Klaviyo is an automated, triggered sequence of messages and logic, the equivalent of what other platforms call a scenario or journey.
Flow Filter (Klaviyo)
A Klaviyo flow filter is a condition that Klaviyo re-evaluates at every step of a flow. A contact who no longer qualifies, for example one who has purchased since entering, is dropped before the next send. It differs from a trigger filter, which is evaluated once, at entry.
Flows (Bloomreach Marketing)
Flows is the Bloomreach Engagement analysis that charts the sequences of events customers take before or after a chosen event. You pick either a starting event ("Starts with") or a target event ("Leads to"). It shows the most common paths and where people drop off. You can drill into any step by an attribute, for example splitting `page_visit` by page, hide a step, or extend the chart one more step. A toggle switches between counting every event and counting only the first occurrence per customer per day. A checkbox limits the analysis to each customer's first-ever occurrence. By default every event type appears; restricting the chart to an included-events list keeps the paths readable.
Framer
Framer is a hosted website builder with a visual editor, a CMS for collections, and its own cookie banner. Teams use it for marketing sites shipped without a developer. Because it hosts the site, it also sets the limits. It decides what can sit at the domain root, how custom code loads, and how its consent banner talks to Google Consent Mode.
Frequency Capping
Frequency capping limits how many messages one contact can receive in a period, across campaigns and channels, so automations don’t stack up on one person. Without it, a contact who qualifies for a welcome flow, a browse-abandonment flow, and a weekly newsletter on the same day gets all three.
Funnel Analysis
Funnel analysis measures how many people move through each step of a sequence, such as view product, add to cart, checkout and purchase. It also shows where they drop off. The biggest drop is where the next campaign or the next site fix belongs.
G
GA4 Client ID
The GA4 client ID is the browser-scoped identifier GA4 assigns to a device and stores in the _ga cookie. In Bloomreach it’s captured and written to a soft ID, so the same browser visitor can be matched across GA4 and Bloomreach for analysis and warehouse joins.
GDPR
The GDPR is the European Union's data protection law. It sets the rules for collecting, storing and using personal data, including lawful basis, consent and the rights people hold over their data. It applies to any company processing data about people in the EU, wherever that company sits, if it offers them goods or services or monitors their behavior.
Generative Engine Optimization (GEO)
Generative engine optimization (GEO) is the practice of making content likely to be cited, and summarized accurately, by AI answer engines such as AI Overviews.
Ghost Sessions (Bloomreach Marketing)
Ghost sessions are short, empty sessions the Bloomreach Engagement JS SDK records when a page left open in a background tab wakes along with a sleeping computer. The page sends a session ping although nobody is active. The fix is the `ping.activity` option, which is off by default. With it on, the SDK pings only when the user has clicked, typed, scrolled or moved the mouse since the last ping. Otherwise it stops pinging. Turning it on changes the session definition. The backend closes a session with a `session_end` event after 20 minutes without pings, so session counts and durations shift.
Global URL List (Bloomreach Marketing)
A global URL list is a project-level set of URL rules in Bloomreach Engagement that weblayers, experiments and tags share. Each rule is one of `Page with URL`, `Page URL containing` or `Page matching regexp`. Campaigns reference the list in their `Show on` or `Do not show on` settings instead of each carrying its own copy. Lists are created under project settings and hold up to 50 rules each. You can import one from CSV, with the match type in column one and the URL in column two. Because campaigns reference the list and do not copy it, an edit applies to every live weblayer at once. A list in use cannot be deleted until each campaign drops it.
Golden Record
A golden record is the single most trusted version of a record after duplicates are merged and conflicting fields are resolved.
Google Analytics 4 (GA4)
Google Analytics 4 (GA4) is Google’s analytics platform, built around an event-based data model rather than Universal Analytics' session-and-pageview model.
Google Consent Mode
Google Consent Mode is a Google framework that adjusts how Google tags, such as GA4 and Ads, behave based on a user’s consent choices. Version 2 uses four signals: analytics_storage, ad_storage, ad_user_data, and ad_personalization.
Google Tag Manager (GTM)
Google Tag Manager (GTM) is a tag management system that lets you deploy and update tracking tags, such as GA4, Bloomreach, and ad pixels, through a web interface instead of editing site code for each change.
Guardrails
Guardrails are checks around an AI model's inputs and outputs, such as blocked topics, format validation, or approval steps, that keep it inside agreed limits.
H
Hallucination
A hallucination is a confident, fluent answer from a language model that is false or unsupported; the model predicts plausible text instead of checking facts.
Hard Bounced (Campaign Status)
Hard bounced is the Bloomreach Engagement campaign status recorded when the receiving server rejected a message permanently, typically because the address does not exist. The platform counts bounces per address, and repeated bounces lead to future sends being suppressed.
Hard ID vs Soft ID
A hard ID is a stable, explicit identifier that reliably points to one person or account, such as an email address, phone number, customer ID, or login. A soft ID is a weaker signal that suggests identity but can be shared, reset, or expire, such as a cookie, device ID, or IP address.
Holdout Group
A holdout group is a share of an audience deliberately excluded from a campaign or automation so its behavior can be compared with the contacts who received it. It’s how a program measures what a message caused rather than what happened around it. An A/B test compares two versions; a holdout compares sending with not sending.
HTML Entity
An HTML entity is a short text code that begins with an ampersand and ends with a semicolon; a browser or email client decodes it into a single character when it renders HTML. Examples: becomes a non-breaking space, & becomes &, € becomes the euro sign. Entities exist for characters that are invisible, reserved by HTML, or hard to type.
Human-in-the-Loop
Human-in-the-loop is a design where a person reviews or approves an AI system's output before it takes effect, instead of the system acting on its own.
I
ID Transformation (Bloomreach Marketing)
ID transformations are Bloomreach Engagement rules that trim leading and trailing whitespace from, and optionally lowercase, every customer ID before it is matched to a profile. They apply to both real-time tracking and imports. Trim is mandatory on every new project. Lowercase is a per-project switch meant for IDs that hold plain-text emails, and it should stay off for cookies and hashed emails. The common failure is leaving lowercase off while the CRM sends `John.Doe@example.com` and the shop sends `john.doe@example.com`: the platform creates two profiles for one person.
Identity Graph
An identity graph is the map that connects the many identifiers (emails, devices, cookies, account IDs, and households) that belong to the same person or family.
Identity Resolution
Identity resolution is the process of matching the many identifiers a single customer leaves across devices, sessions, and tools into one unified profile. It’s how a CDP decides that an anonymous web visitor, an email subscriber, and an app user are the same person.
Imports (Bloomreach Marketing)
Imports in Bloomreach Engagement load customers, events or catalog items from a file, a database or a storage bucket, either once or on a repeating schedule. They are the usual route for data that does not come from tracking: order history, loyalty balances, product feeds, attributes computed in a warehouse.
In-App Message
An in-app message is content shown inside a mobile app while the customer is using it, triggered by their behavior, with no permission prompt needed. It is the mobile equivalent of a weblayer: a banner, a modal or a full screen, shown at a moment the app can see.
In-Session Prediction (Bloomreach Marketing)
In-session prediction is a Bloomreach Engagement prediction model that scores each visitor, from 0 to 1, on how likely they are to buy before the current session ends. The score updates as pages viewed, items browsed and time on site accumulate. It combines live session signals with the customer's history. The result is stored in a customer property named `In-session prediction [time frame]`, recalculated whenever it is read, the same way aggregates and segmentations are. The score is used through segmentations and customer filters: a weblayer audience, a scenario branch or a report split by probability band. Training takes from about 20 minutes to a few hours, and AUC is the quality metric to judge the model by.
Incrementality Measurement
Incrementality measurement estimates what happened because of marketing, often using control groups, rather than crediting everything that happened after an ad.
IndexNow
IndexNow is an open protocol that lets a website notify search engines the moment a URL is added, updated, or deleted, instead of waiting to be crawled.
Inference
Inference is the act of running a trained AI model to produce an output, as opposed to training it; every prompt answered is one inference call.
Infinario
Infinario is the original name of the platform now known as Bloomreach Engagement. It launched for the gaming industry, tracking player behavior and running in-game retention campaigns, before rebranding to Exponea in 2016 to serve broader e-commerce and B2C markets. Bloomreach acquired Exponea in January 2021 and renamed it Bloomreach Engagement. Today the product is called Bloomreach Marketing.
Instance vs Project (Bloomreach Marketing)
In Bloomreach Engagement an instance is the hosted environment on Google Cloud. An account is the customer within it, and it shares contracted volumes and limits. A project is an isolated data space under that account, with its own customers, events, campaigns, project token and access management. Divisions sit beneath a project to scope regions or brands. Instances come in three types: multi-tenant, single-tenant and exclusive. The hierarchy runs instance > account > project > division. No data is shared between projects, which is why teams run separate development, staging and production projects, or one project per country. Roles granted at account or instance level apply to every project below, which is how someone ends up with far more access than intended.
Intelligent Tracking Prevention (ITP)
Intelligent Tracking Prevention (ITP) is Apple’s set of privacy restrictions in Safari and WebKit that limit how long cookies and other identifiers set by scripts can persist.
J
Jailbreak
A jailbreak is a prompt crafted to make an AI model ignore its safety rules or system prompt. It is usually typed by the person using the model, where a prompt injection hides in data the model reads.
Jinja
Jinja is a templating language that inserts data and logic into text using {{ ... }} for output and {% ... %} for logic like loops and conditions. In Bloomreach it is the way a message or weblayer pulls in customer attributes and catalog data, and it runs on the server before the result reaches the browser.
Journey Orchestration
Journey orchestration is coordinating a customer’s interactions across channels and lifecycle stages into one connected experience.
K
Kafka
Apache Kafka is an open-source event streaming platform that moves high volumes of event data between systems in real time, widely used as the backbone of data pipelines.
Key Performance Indicator (KPI)
A key performance indicator (KPI) is a metric chosen to measure progress toward a specific goal, such as conversion rate, retention, or average order value.
Key-Value Pair
A key-value pair is a single piece of data expressed as a named key and its value, such as country set to DE, or plan set to premium.
Keyword Auto-Responses (Bloomreach Marketing)
Keyword auto-responses are Bloomreach Engagement rules that reply automatically when a customer texts a configured word such as STOP, JOIN or INFO. Each rule can also revoke consent, grant consent, or do nothing beyond the reply. A match requires the keyword at the start of the message as a complete word. Case, accents and extra spaces are ignored, but "Please STOP" does not match because STOP is not the first word. A revoke keyword blocks the number from future messages and records a rejection on matching profiles. A grant keyword creates a profile for an unknown number when "Map new subscribers ID" is enabled. The platform has no reserved keyword list. Every word is yours to define, up to 100 per configuration.
Klaviyo
Klaviyo is a marketing automation and customer data platform focused mainly on email and SMS, widely used by ecommerce brands, especially on Shopify.
Klaviyo Composer
Klaviyo Composer is Klaviyo’s AI agent that builds a campaign or flow from a written brief. It drafts the audience, messages, and steps, and a person approves the result before launch. Klaviyo announced it in March 2026, opened a public beta in June, and added scheduled audits and custom skills in September.
Klaviyo Data Platform
Klaviyo Data Platform is the data layer under Klaviyo’s marketing and service products: profiles, metrics, segments, and the analytics on them. From September 2026 it can be queried through plain-language prompts that Klaviyo translates into SQL. The feature was in preview at launch.
Klaviyo Headless
Klaviyo Headless exposes Klaviyo’s tools through the Model Context Protocol, 260 or more of them at its September 2026 launch. Outside AI agents and apps, such as Claude or ChatGPT, can then read profiles and trigger actions without a person in the Klaviyo interface.
Knowledge Cutoff
A knowledge cutoff is the date after which a language model saw no training data. Events after that date are unknown to it unless they are supplied in the prompt or retrieved.
Known Customer
A known customer is a person a CDP can identify by a stable identifier such as an email address, account ID, or customer ID, as opposed to an anonymous visitor tracked only by a soft signal.
L
Large Language Model (LLM)
A large language model (LLM) is an AI model trained on huge amounts of text to predict the next token, so it can generate, summarize, and reason over language.
Lead Scoring
Lead scoring is ranking prospects by fit, intent, and engagement so sales and marketing can focus effort on the best ones.
Least Privilege
Least privilege is the rule that a user, system, or AI agent gets only the access its task needs. A mistake or an attack can then do only limited damage.
Legitimate Interest
Legitimate interest is a GDPR legal basis for processing personal data without explicit consent. The controller weighs its own interest against the individual’s rights and records the result in a balancing test. Some platforms, Bloomreach Engagement among them, let a consent category be configured on this basis, which flips its default from opt-in to opt-out.
Lifecycle Marketing
Lifecycle marketing is marketing based on a customer’s lifecycle stage and relationship maturity rather than one-size-fits-all campaigns.
Lifecycle Stage
A lifecycle stage is a label for where a customer is in their relationship with a brand, such as new, active, at risk or lapsed. It is usually computed from purchase recency and frequency. It gives every team the same vocabulary for who a customer is right now.
List Hygiene
List hygiene is the routine removal or suppression of addresses that bounce, never engage or were never valid. The goal is a sending reputation built on people who want the mail. A list that is never cleaned slowly fills with dead mailboxes and spam traps, and the mailbox provider notices before you do.
List vs Segment (Klaviyo)
A list in Klaviyo is a mostly static group a contact joins explicitly, such as a newsletter signup, as distinct from a segment, whose membership is dynamic and rule-based.
Lookalike Audience
A lookalike audience is an audience modeled to resemble your existing valuable customers.
Loomi AI
Loomi AI is Bloomreach’s AI layer and the brand its products now sit under, covering personalization, search ranking, predictions, and a set of named agents. Those agents include the Loomi Marketing Agent and Loomi Connect.
Loomi Shopping Agent (Bloomreach)
The Loomi Shopping Agent is Bloomreach's conversational shopping assistant: it guides a shopper to a purchase through natural-language chat on the retailer's site. Bloomreach first announced it in 2023 as Bloomreach Clarity.
Loyalty Marketing
Loyalty marketing is the practice of rewarding repeat customers, through points, tiers, or perks, to increase retention and lifetime value.
Loyalty Program
A loyalty program is a structured scheme that rewards customers for repeat behavior, usually with points, tiers or perks. In exchange, customers identify themselves and buy more often. The rewards are visible; the identification is the part that makes the rest of the data usable.
Loyalty Tier
A loyalty tier is a level in a loyalty program, such as silver or gold, reached by spend or activity. Each tier carries its own benefits and its own rules for moving up and down. Tiers turn a points balance into a status the customer cares about keeping.
M
Mailchimp
Mailchimp is an email marketing platform for small and mid-sized businesses, organized around audiences, tags, and segments, with a built-in automation builder. Owned by Intuit since 2021, it’s often the first tool a brand sends from.
Marketing Automation
Marketing automation is software that triggers and personalizes messages across channels, such as email, SMS, and on-site, based on customer behavior and data rather than manual sending.
Master Data Management (MDM)
Master data management (MDM) is the enterprise discipline and toolset for governing trusted master records for core entities such as customers, products, and suppliers.
Metric (Bloomreach Marketing)
A metric in Bloomreach Engagement is a saved definition of a number, such as total revenue or conversion rate, computed from event data and reused across reports and dashboards. A simple metric aggregates one event type, for example `sum > purchase > total_price`. A formula combines several simple metrics with arithmetic. Each metric can carry its own event filter, customer filter and date filter. The common failure is a date filter saved inside the metric that nobody remembers. The report then shows a narrower period than its own date picker says.
Metric (Klaviyo)
A metric in Klaviyo is a tracked customer event, such as Placed Order, Viewed Product, or Opened Email, that flows and segments can use.
Mobile Push
Mobile push is a notification sent to a customer's phone through an installed app. The app must register the device token with the messaging platform first. The platform delivers to that token, not to the person.
Model Context Protocol
The Model Context Protocol (MCP) is an open standard that lets AI assistants and agents call a product’s tools and read its data through a defined server. Marketing platforms adopted it in 2026: Klaviyo Headless, Bloomreach Loomi Connect, and Brevo’s Claude connector all expose their platforms this way.
Model Drift
Model drift is the slow loss of accuracy in a deployed model as the real-world data it sees moves away from the data it was trained or tuned on.
Multi-Agent System
A multi-agent system splits a task across several AI agents, each with its own role and tools, coordinated by an orchestrator agent or a fixed workflow.
Multimodal AI
A multimodal model is an AI model that works with more than one kind of data, such as text, images, audio, and video, as input or output, in the same system.
N
Naming Convention
A naming convention is the agreed pattern for what events and attributes are called in a customer data platform, and it is decided before tracking is built. It covers the case, the separator, the order of words, and whether a name describes the action or the object it happened to.
Next Best Action (NBA)
Next best action (NBA) is the single action a system recommends as most likely to move a customer forward: which offer, message, or channel, at which moment.
NIS2
NIS2 is the EU's cybersecurity directive, due in national law since October 2024, setting security and incident reporting duties for in-scope companies.
Non-Breaking Space
A non-breaking space is a space character (Unicode code point U+00A0) that keeps the tokens on either side on the same line, so the renderer never breaks between them. It looks like an ordinary space but behaves like glue.
O
Omnichannel Marketing
Omnichannel marketing is coordinated communication across channels such as email, SMS, app, web, and ads, built on a consistent view of the customer.
On Date Attribute Trigger (Bloomreach Marketing)
The on date attribute trigger is a Bloomreach Engagement scenario trigger that fires a set number of days before or after a date stored on the customer profile or in an event. Typical dates are a `birthday` or a subscription end date.
One-Click Unsubscribe (Bloomreach Marketing)
One-click unsubscribe is the list-unsubscribe header that Bloomreach Engagement adds to marketing email so mailbox providers can show an Unsubscribe button next to the sender name. Gmail and Yahoo require the header.
Open Rate
Open rate is the share of delivered emails that recipients opened, calculated as unique opens divided by delivered messages.
Opened (Campaign Status)
Opened is the Bloomreach Engagement campaign status recorded when the tracking pixel in an email was loaded, which usually, but not always, means a person looked at it. Some mail clients block the pixel, so a customer can show clicked without opened; others load it automatically, so an open can happen with nobody reading.
P
Performance Dashboards (Bloomreach Marketing)
Performance dashboards are Bloomreach Engagement's pre-built analytical views of project revenue, email channel performance and campaign evaluation. Channel performance covers email revenue, engagement and deliverability; campaign evaluation covers scenarios and email campaigns.
Personalization
Personalization is adapting a customer’s experience based on what is known about them, whether through rules a marketer wrote, segments the team defined, or a model’s decisions.
Personalization Token
A personalization token is a placeholder in a message template, such as a first-name field, that is replaced with each recipient's own data at send time. In Bloomreach Engagement it is written in Jinja; other platforms call it a merge tag.
Personally Identifiable Information (PII)
Personally identifiable information (PII) is data that identifies a specific person, such as an email address, phone number, or home address.
PII Redaction
PII redaction is the removal or masking of personal data such as names, emails, and phone numbers from text before it goes to an AI model or into its logs.
Points Ledger
A points ledger is the transaction record of every point a loyalty member earned, redeemed, or lost to expiry, from which the current balance is calculated. It lives in the loyalty system; the CDP needs a copy of both the balance and the transactions to act on them.
Post-Purchase Flow
A post-purchase flow is a series of triggered messages after an order: confirmation and delivery, then usage guidance, a review request, and a reorder prompt. It runs on the order event and the products in it, so the content can match what the customer bought.
Post-Purchase Window
The post-purchase window is the period from order confirmation to first use of the product, when a customer is most attentive. It is the cheapest time for a marketing program to earn the second purchase.
Preblocked (Campaign Status)
Preblocked is the Bloomreach Engagement campaign status recorded when the email provider refused to attempt delivery because the recipient or their domain is on the provider's own suppression list. It is a block at the provider level, before the message ever reaches the receiving server, usually the result of earlier hard bounces or complaints.
Predictions (Bloomreach Marketing)
Predictions in Bloomreach Engagement are built-in machine learning models that score each customer on an outcome, such as likelihood to buy or to churn. The score is stored as a customer attribute, so it can be used like any attribute: in a segment, a journey condition or a personalized message.
Predictive Analytics (Klaviyo)
Klaviyo’s predictive analytics are model-generated forecasts on a customer profile, such as predicted lifetime value, churn risk, and expected date of next order.
Predictive Scoring
Predictive scoring is a model’s estimate of the likelihood of an outcome such as churn, conversion, purchase, or engagement.
Preference Center
A preference center is the interface where a customer manages topics, channels, frequency, and permissions.
Probabilistic Matching
Probabilistic matching links records using signals that suggest the same person, such as device, IP, location, or behavior patterns, but do not prove it.
Progressive Profiling
Progressive profiling asks a customer one question at a time, inside messages or pages they are already on, so the profile fills in gradually, with no long form at signup. Each question is asked because a specific campaign needs the answer, and that campaign sends within a week of it.
Prompt
A prompt is the text given to a language model to shape its answer; the system prompt sets standing rules and the user prompt carries the task.
Prompt Injection
Prompt injection is an attack where text the model reads, such as a web page, an email, or a customer field, carries instructions that hijack what it does.
Propensity Model
A propensity model estimates how likely a specific customer is to take a specific action, such as purchasing, churning, clicking, or converting.
Pseudonymization
Pseudonymization replaces direct identifiers with artificial IDs, so records can be processed without directly exposing the person’s identity.
Publisher First-Party Data
Publisher first-party data is audience data owned by publishers or media companies, collected directly from their own readers and viewers.
Q
Quality Assurance (QA)
Quality assurance (QA) is the practice of systematically checking that something works as intended before it goes live. In a data stack, that means checking tracking, consent, and campaigns.
Query Parameter
A query parameter is a key-value pair appended to a URL after a question mark that passes extra information to the page or to analytics, as in ?utm_source=newsletter.
Queue
A queue is a holding area for data or tasks waiting to be processed, used to absorb bursts, retry failures, and decouple systems that run at different speeds.
Quiet Hours
Quiet hours are a window in the customer's local time, typically overnight, during which no marketing message is sent on any channel. Transactional messages are the exception. Several countries regulate marketing SMS hours by law, so the window should match the strictest market a program sends to.
R
RCS Campaigns (Bloomreach Marketing)
RCS campaigns are Bloomreach Engagement messages sent over Rich Communication Services (RCS), the carrier standard that upgrades SMS. RCS adds images, video, reply buttons, carousels, and a verified brand name and logo in the thread.
Real-Time Activation
Real-time activation is triggering a message, offer, or decision as soon as possible after a customer signal, instead of on a schedule.
Real-Time Profile
A real-time profile is a customer profile updated as soon as new behavior is processed, rather than in a nightly batch.
Reasoning Model
A reasoning model is a language model that works through a problem step by step before answering, trading more time and tokens for better results on hard tasks.
Recommendation Engine
A recommendation engine is a system that suggests products, content, or offers to a customer based on behavior, context, or similarity to other customers.
Recommendations+ (Bloomreach Marketing)
Recommendations+ is the journey-based version of the Bloomreach Engagement `Personalized recommendations for you` template. It predicts the next product from the order of a customer's recent interactions.
Red Teaming
Red teaming is a deliberate attempt to make an AI system fail, misbehave, or leak, done before launch so your own people find the failures, not customers.
Repeat Purchase Rate
Repeat purchase rate is the share of customers who bought more than once in a period, a direct measure of whether lifecycle marketing is working. Acquisition fills the top; this number says how much of it stays.
Replenishment Flow
A replenishment flow is a triggered message timed to when a consumable product is likely to run out, based on the last order date and the typical usage period. It applies to anything bought on a cycle: coffee, pet food, supplements, printer ink, contact lenses.
Report (Bloomreach Marketing)
A report in Bloomreach Engagement is a pivot-style table analysis where rows and columns split customers or events by attributes. Each cell shows a metric such as a customer count, an event count or revenue.
Retail Media Network (RMN)
A retail media network (RMN) is an advertising network run by a retailer, using its own shopper data and inventory to let brands target and measure ads.
Retention Rate
Retention rate is the share of customers from a starting group who are still active, usually still buying, after a set period. For subscriptions it is clear; for retail it depends on what counts as active, which is the first thing to agree on.
Retentions (Bloomreach Marketing)
Retentions is the Bloomreach Engagement analysis that shows, cohort by cohort, what share of customers who took an initial event came back later to take a returning event. Examples are a `session_start` followed by another `session_start`, or a first `purchase` followed by a second.
Retrieval-Augmented Generation (RAG)
Retrieval-augmented generation (RAG) fetches relevant records and places them in the model's prompt, so the answer draws on your data, not only on its training.
Reverse ETL
Reverse ETL is the process of sending modeled data from a data warehouse back into the business and activation tools that act on it.
Reward Trigger
A reward trigger is a loyalty event, such as a tier change, points about to expire or a reward becoming available, that starts an automated journey the moment it happens. It is what connects the loyalty system to the messaging the customer sees.
RFM Segmentation
RFM segmentation scores customers on recency of last purchase, frequency of purchases and monetary value, then groups them into segments such as champions, loyal or at risk. It needs nothing but order history, which is why it is usually the first segmentation a CDP runs and often the one that lasts.
Role-Based Access Management (Bloomreach)
Role-based access management is how Bloomreach Engagement grants permissions. Admins assign predefined or custom roles at instance, account or project scope, and user groups restrict who can edit specific items. Role names end in an extent level, from Viewer up to Admin. Each level inherits everything below it, so an Editor cannot publish or export. A role granted at a higher scope reaches every account and project beneath it. Every user needs at least one standalone role to log in. Personal Data Viewer, the flag that reveals personal data (PII) fields, is never part of a predefined role; someone has to add it on purpose. Custom roles can stack up to 32 inherited roles and cannot remove a permission from any of them.
Running Aggregate (Bloomreach Marketing)
A running aggregate is a Bloomreach Engagement definition that computes a value for each event from the events around it. An example is the country of the last `session_start` before each `purchase`. A regular aggregate produces one value per customer, for example lifetime spend. A running aggregate produces one value per event, so an analysis can be split by an attribute the event itself never carried. An including/excluding switch decides whether the event being computed for (the anchor) counts toward its own value. That matters when the anchor and the source event are the same type. Running aggregates work in analyses but not in Jinja personalization; only aggregates work there.
S
Salesforce Marketing Cloud
Salesforce Marketing Cloud is Salesforce's enterprise marketing automation suite: Journey Builder for journeys, data extensions for data, AMPscript for logic. The edition most teams mean, now sold as Marketing Cloud Engagement, is the one with that stack. It grew out of ExactTarget and is common in large organizations that already run Salesforce CRM. A second edition, Marketing Cloud Next, shipped in 2025 on the core Salesforce platform and Data Cloud.
Scenario
A scenario in Bloomreach is the visual flow that defines an automated journey: what starts it, the conditions and waits along the way, and the actions it takes.
Schema Drift
Schema drift is the slow, unannounced change of a data structure over time: renamed fields, new values, changed types, properties that stop arriving. Nobody decides it; it happens one release at a time, and the segments built on the old shape keep running on fewer and fewer customers.
Second-Party Data
Second-party data is another organization’s first-party data, shared through a direct partnership rather than bought from a third-party broker.
Seed Address
A seed address is a test mailbox that receives a campaign before or alongside the real audience. Teams usually keep a small segment of them across Gmail, Outlook, and a corporate domain. That way they can check rendering, links, personalization, and inbox placement on real email clients and phones.
Segment
A segment is a defined group of customers who share criteria, such as recent buyers, lapsed subscribers, or high-value customers, used for targeting and analysis.
Segmentation Fork (Bloomreach Marketing)
A segmentation fork is a Bloomreach Engagement scenario node that splits customers into separate branches, one for each segment of a saved segmentation. Each group then takes its own path. A scenario can hold up to 5 segmentation forks. A segmentation used this way can have up to 20 segments. Segmentations with parameters are not supported. The common failure is editing the segmentation afterwards. A deleted or renamed segment keeps its branch but passes 0 customers, and a new segment gets no branch. Both stay that way until you open the node, which shows a warning, and click `Update Segmentation`.
Send-Time Optimization
Send-time optimization picks the hour each customer is most likely to open or click, based on their past engagement, instead of using one send time for everyone. The platform spreads a campaign over a window and delivers to each person at their predicted best time.
Sender Reputation
Sender reputation is the score mailbox providers assign to a sending IP and domain, built from complaints, bounces, spam traps and engagement. It decides whether mail reaches the inbox. Each provider keeps its own; Google exposes part of it in Postmaster Tools, most others do not.
Sent (Campaign Status)
Sent is the Bloomreach Engagement campaign status recorded when a message was submitted to the network for delivery. Channels where the provider reports submission as a separate step use it. In email reporting it is usually zero, because the email provider reports delivered or a bounce instead.
Server-Side Anonymous Identity Management (Bloomreach)
Server-side anonymous identity management is the set of Bloomreach Engagement methods that keep an anonymous visitor's ID stable across visits. They do it by setting the identity cookie from a server instead of from JavaScript. A server-set cookie outlasts Safari's Intelligent Tracking Prevention, which caps JavaScript-set cookies at 7 days. The documentation lists three routes. The first is a custom tracking domain delegated with NS zone records (a CNAME record still gets the 7-day cap). The second is a server-side Google Tag Manager container. The third is a cookie set by your own backend and passed to the SDK as a soft ID such as `anonymous_id`. Without one of them, a returning Safari visitor gets a new anonymous ID after a week and their visits split into disconnected profiles. The third route does not switch off the SDK cookie, so changing your own cookie value can still merge into the previous profile.
Server-Side Rendering
Server-side rendering means the final markup is generated on the server before reaching the browser, rather than being assembled by JavaScript in the visitor’s browser. The browser receives finished markup.
Server-Side Tracking
Server-side tracking sends event and identity data to a server you control, which then forwards it to your analytics and marketing tools, instead of the browser sending it straight to each vendor. Moving collection off the browser keeps it out of reach of ad blockers and some browser privacy limits.
Service Worker
A service worker is a script the browser runs in the background, apart from the page, to intercept requests, cache responses, and receive push notifications. Web push and some synchronous experiment setups depend on one being registered from the site's own root path.
Sessionization
Sessionization groups a stream of individual events into visits, usually by closing a session after a period of inactivity, so that visits can be counted and analyzed. The inactivity window, what counts as the first event, and whether a campaign click starts a new session are all choices, and each tool makes them differently.
Shadow AI
Shadow AI is the use of AI tools by employees without the company's approval or knowledge, typically personal chatbot accounts fed with work data.
Shopify
Shopify is a hosted ecommerce platform whose store, checkout, and customer events are the usual data source behind Klaviyo and Bloomreach Marketing (Engagement) setups. Its integrations sync customers, orders, products, and cart activity into the marketing platform, and its cart permalinks let a message rebuild a cart from variant IDs.
Sign-Up Form (Klaviyo)
A sign-up form in Klaviyo is a popup, flyout, embedded, full-page, or banner form used to collect email and SMS subscribers and their consent directly on a website.
Single Customer View
A single customer view (SCV) is one profile per customer that combines identity, behavior, transactions, consent, and attributes from every source, replacing the scattered records the same person leaves across systems. Customer 360 and unified customer profile are other names for the same thing.
Smart Sending (Klaviyo)
Smart Sending is a Klaviyo setting that skips contacts who were recently messaged on the same channel, to prevent over-messaging.
Soft Bounced (Campaign Status)
Soft bounced is the Bloomreach Engagement campaign status recorded when the receiving server rejected a message temporarily, for example because the inbox was full. The platform still attempts later messages to that address. Repeated soft bounces for one address eventually count as a permanent failure and trigger suppression.
SPF
SPF is a DNS TXT record on your domain listing which servers are allowed to send email on behalf of your domain. Receiving mail servers check it on every message; without it, they cannot verify whether a sender is authorized, which hurts deliverability and invites spoofing. Publishing SPF is a DNS edit, not a software project.
Standard Contractual Clauses (SCCs)
Standard contractual clauses (SCCs) are the European Commission's approved contract terms for sending personal data outside the EU with no adequacy decision.
Structured Output
Structured output is a model response forced into a fixed format, such as JSON with named fields, so other software can read it without parsing free text.
Sub-Account
A sub-account is a separate contact base, sender set, and sending reputation for one brand or client under a central admin account. In Brevo it is an enterprise feature: each sub-account has its own IP address and API key, with central dashboards and per-user permissions above them.
Success (Campaign Status)
Success is the Bloomreach Engagement campaign status recorded when a customer was added to an advertising audience through an ads node. Email messages never receive it, so in an email report it stays at zero.
Suppressed (Campaign Status)
Suppressed is the Bloomreach Engagement campaign status recorded when the platform itself decided not to send. The reasons are missing consent, a frequency policy, repeated bounces, a suppression list or the list hygiene filter. The event's `type` attribute says which rule applied. Nothing reached the provider.
Suppression Audience
A suppression audience is a group deliberately excluded from a campaign, such as recent buyers, opted-out customers, or existing subscribers.
Surveys (Bloomreach Marketing)
Surveys are Bloomreach Engagement hosted forms, built under `Campaigns > Surveys` and sent through a personalized link. Because the link is personal, every answer lands on the customer's profile as a `survey` event instead of an anonymous response. Five question types are available: single choice, multiple choice, rating stars, text field and matrix. The link goes into a message with `{{ surveys['Name of the survey'] }}`. Each answered question is tracked as its own event with `question`, `answer` and `survey_id` attributes. Setting a survey as `recurrent` lets a customer submit again, but the new answer replaces the old one. Events are only recorded if `Allow tracking of events not defined` is on in the Data Manager, or if the `survey` event has been added there by hand.
System Events (Bloomreach Marketing)
System events are the events Bloomreach Engagement generates itself in every project, each with a fixed attribute set. Examples are `session_start`, `session_end`, `first_session`, `page_visit`, `campaign`, `merge`, `ab test`, `experiment`, `survey` and `anonymization`. They need no custom integration and look the same across projects, which is why performance dashboards are built on them. The `campaign` event is the one marketers read most: its `status` attribute records each send's outcome, from `sent` and `delivered` to `clicked` or `suppressed`. Two caveats: `page_visit` is off by default in the JavaScript SDK, and a sleeping computer whose browser wakes up can produce ghost sessions with no real activity.
T
Target Event
A target event is the specific, time-bound outcome a propensity model predicts, written as a filter the platform can evaluate. Examples: a second purchase within 60 days of the first, or a renewal in the 45 days before expiry. "Purchase, generally" is not a target event; it produces a score that correlates with already being a customer.
Temperature
Temperature is a setting that controls how random a language model's word choices are: low values give consistent answers, high values give more varied ones.
Template Parameters (Bloomreach Marketing)
Template parameters are Bloomreach Engagement placeholders written as `[[ parameter ]]` in weblayer, experiment, tag manager and webhook templates. To a non-technical editor, each one appears as a plain input field such as a text box, color picker or dropdown. A developer declares the type and default, for example `[[ method : enum(GET, POST) | GET ]]`. A JSON form adds tooltips, categories and ordering, so a complex template reads like a settings page. Parameters are not Jinja. Parameters set visual and configuration values when the template is used; Jinja inserts customer data at render time. New HTML blocks use their own, separate parameter editor.
Third-Party Data
Third-party data is data collected and aggregated by outside providers, then sold for targeting, with no direct relationship to the customer.
Throughput Policy (Bloomreach Marketing)
A throughput policy is a Bloomreach Engagement project setting that caps the number of concurrent connections shared by every webhook assigned to it. The cap protects a third-party API that has a request limit. You create the policy under project settings, the same way as a frequency policy, and select it inside each webhook node. The concurrency limit only approximates requests per second, because real throughput depends on how fast the receiving API responds. Switching a running webhook from one policy to another applies only to new requests. Changing the limit inside a policy applies immediately. Webhooks in on-event scenarios lose some processing priority when they share a policy, which keeps the load fair across campaigns.
Token
A token is the unit of text a language model reads and writes, about three-quarters of an English word. Vendors bill and limit in tokens, not characters.
Tone of Voice (AI Setting)
Tone of Voice is the setting in an AI content tool, Bloomreach's Loomi among them, where brand guidelines are stored once: register, banned words, claims never made, how the customer is addressed. Every generated draft starts from it, so nobody has to repeat the guidelines in each prompt.
Tool Use
Tool use is a language model's ability to call a defined function, such as a database query or an API, that the surrounding software then runs for it.
Tracking Plan
A tracking plan lists every event and attribute a site or app sends to the customer data platform: its name, fields, when it fires, and who owns it. It’s written before tracking is built and kept current after, so segments, campaigns, and reports all read from names that mean one thing.
Transactional Data
Transactional data is records of purchases, orders, subscriptions, payments, and refunds.
Transactional Message
A transactional message is a communication the customer expects because of an action they took, such as an order confirmation, a shipping notice, or a password reset. It is sent without marketing consent and outside frequency caps and quiet hours, because the customer complains when it doesn't arrive.
Trends (Bloomreach Marketing)
Trends is the Bloomreach Engagement analysis that plots how often chosen events occur over time, as either a count of events or a count of unique customers per day. You pick one or more events, apply date, customer and event filters, and the chart draws itself. Typical uses are page traffic, visits and purchases over a period, or two related events side by side. Trends charts event counts only. To see an event attribute over time, such as revenue per day, use a report with `timestamp` in rows.
Trifft
Trifft, marketed as TRIFFT Loyalty Cloud, is a full-featured loyalty platform for building points, tiers, and reward programs, with a native two-way integration into Bloomreach Marketing.
Trigger
A trigger is the node that starts an automated scenario and decides when customers enter the flow.
Trigger Filter (Klaviyo)
A Klaviyo trigger filter is a condition evaluated once when a contact enters a flow, deciding who gets in. It is not re-checked at later steps; that is the flow filter's job.
Two-Way Messaging (Bloomreach Marketing)
Two-way messaging is the Bloomreach Engagement feature that records customer replies to SMS, MMS, RCS and WhatsApp as `campaign` events with the status `incoming_message`. A reply can then trigger a scenario, answer a survey or open a support thread. It needs three things. First, a dedicated short or long code, because customers cannot reply to an alphanumeric sender. Second, a lowercase `phone` attribute in international format starting with "00" or "+". Third, the Track incoming messages toggle in Project settings. A reply from a number that matches no profile is dropped: two-way messaging never creates profiles. At most 10 profiles per number are processed.
U
Unsubscribe Rate
Unsubscribe rate is the share of delivered messages that led the recipient to opt out, usually expressed per thousand delivered. It is the earliest reliable signal that a program is over-sending or mistargeting. It is also usually blamed on the wrong message: the last one, when the four before it did the damage.
Unsubscribed (Campaign Status)
Unsubscribed is the Bloomreach Engagement campaign status recorded when the recipient opted out through the mail client's list-unsubscribe feature, the one-click button in Gmail, Apple Mail and similar clients. An opt-out through a link in the message body is recorded differently: as a click on that link plus a consent change.
Uplift Modeling
Uplift modeling predicts how much a specific action, such as a discount, changes an individual customer's behavior, separating the persuadable from the people who would buy anyway. A propensity model asks who will buy; an uplift model asks who will buy because of this.
Use Case Center (Bloomreach Marketing)
The Use Case Center is a Bloomreach Engagement library of ready-made campaigns that you download into a project and then adjust. Each use case bundles the scenarios, templates and an evaluation dashboard. You open it from the icon at the bottom of the left menu. It has three tabs: Home, All use cases (filterable by channel, goal, tag, KPI and industry) and Project use cases, which lists what is already downloaded. A tile shows `Ready to use` only when the project meets the requirements. If a required module is missing, the tile cannot be opened. After download, the Use case setup section lists the items that need customizing, usually the email or banner design. Leave the remaining items as they are.
UTM Parameter
A UTM parameter is a tag added to a URL’s query string that tells analytics tools where a click came from. The five standard ones are utm_source, utm_medium, utm_campaign, utm_term, and utm_content.
V
Value-Based Segmentation
Value-based segmentation groups customers by their economic value to the business, such as predicted lifetime value, rather than by demographics or behavior alone.
Variant ID
A variant ID is the identifier for a specific purchasable version of a product, such as a particular size and color, as distinct from the parent product ID.
Vector Database
A vector database stores embeddings and finds which stored items are closest to a given one; it is the lookup behind semantic search and RAG.
Visitor ID
A visitor ID is an identifier assigned to an anonymous visitor, usually stored in a cookie, that ties their activity together before any hard identifier, such as an email address, exists.
Vouchers (Bloomreach Marketing)
Vouchers in Bloomreach Engagement are imported pools of unique discount codes, grouped under a pool name. Campaigns assign the codes to customers through Jinja, and the platform tracks when a code is redeemed. The Jinja call `assign_available()` picks a random code from the pool, marks it assigned and renders it. Within one email the same customer gets the same code unless `cached = false` is set. The platform does not apply the discount itself; your store's voucher system still does that. A mapping in project settings tells the platform which purchase attribute carries the redeemed code. When a pool runs dry, emails fail with `enqueue_failed` and weblayers are not displayed.
W
Warehouse-Native CDP
A warehouse-native CDP is a CDP approach where the data warehouse, such as BigQuery, Snowflake, or Databricks, is the single source of truth, with profiles and activation built directly on it.
Web Analytics
Web analytics is the measurement of website traffic and on-site behavior, such as pages viewed, sessions, and conversions, to understand how visitors use a site.
Web Push
Web push is a notification sent to a browser the customer has given permission to, delivered while the site is closed, through a service worker registered on the domain. It needs no app and no email address, which makes it the one channel available for anonymous visitors.
Webhook
A webhook is an automated HTTP request one system sends to another when an event happens, used to push data or trigger an action in near real time.
Weblayer
A weblayer is a Bloomreach overlay injected into a website: a popup, banner, inline block, slider, or any other on-page surface. It’s configured as three fields, HTML, CSS, and JavaScript, that Bloomreach renders and delivers to the visitor’s page, controlled by segments and scenarios.
Weblayer Variant Generator
The Weblayer Variant Generator is a Bloomreach Loomi tool that produces alternative versions of a weblayer for testing, varying content such as headline, copy, and image. A team can test more variants than a designer could produce by hand.
WhatsApp Campaigns (Bloomreach Marketing)
WhatsApp campaigns are Bloomreach Engagement messages sent through a WhatsApp Business account, using the Infobip or Sinch mobile messaging integration. They come in two forms: Meta-approved templates and free-form replies. A template message can be sent at any time, but Meta must approve it first. A free-form message needs no approval, but it can only be sent within 24 hours of the customer's last message. Outside that window it is not delivered. Each message carries a Meta category (marketing, utility or service) that sets its price. Meta also caps how many new conversations a business can start per day. Since April 1, 2025, Meta has paused marketing template messages to US (+1) numbers; utility templates are unaffected.
Widget Webhook (Bloomreach Marketing)
A widget webhook is a Bloomreach Engagement webhook preset whose settings screen is a custom page hosted on your own server. The page is embedded in the scenario builder as an iframe, so marketers configure a third-party action without leaving Bloomreach. A webhook preset on its own is a reusable webhook template created under `Data & Assets > Integrations`, and it appears among `Other` actions in scenarios. The widget variant adds a `postMessage` exchange: Bloomreach asks the widget for its state on save or test. The saved `widget_state` is capped at 4 KB. Only string values are supported for event properties. Updating a preset does not update the webhook nodes already placed in scenarios.
Win-Back Campaign
A win-back campaign targets customers in the lapsed lifecycle stage with a message chosen by the likely reason they lapsed, sent once, followed by silence if nothing changes. A quarterly discount to everyone quiet is a broadcast, not a win-back.
X
X-Header
An X-header is a custom, non-standard email header, conventionally prefixed with X-, used to carry extra metadata such as campaign or system information alongside a message.
XML Feed
An XML feed is a structured file, in XML format, that a system publishes so other systems can import its data. It is commonly used for product catalogs, pricing, and content syndication.
Y
Year-over-Year (YoY)
Year-over-year (YoY) is a comparison of a metric to the same period a year earlier, used to measure growth while controlling for seasonality.
Yield Management
Yield management is the discipline of setting prices and allocating limited inventory to earn the most revenue from it. It comes from airlines and hotels and now applies to ad impressions, media placements, and retail-media slots, where the day-to-day work is usually called yield optimization.
Z
Zero Data Retention
Zero data retention is a contract term under which an AI vendor doesn't store prompts and outputs sent through its API beyond the time needed to serve them.
Zero-Party Data
Zero-party data is information a customer chooses to share with you deliberately: quiz answers, stated preferences, survey responses, saved filters. They knew they were giving it and shared it on purpose.
These definitions come out of the work.
If a term describes a problem you are dealing with, we design and implement the systems behind it.