Glossary

An ad blocker is software that stops selected scripts, requests, or page elements from loading. It can block analytics calls as well as ads, which creates gaps in destination reports. Block lists and user settings differ, so the impact varies by browser and audience.

AdAttributionKit is Apple's privacy-preserving framework for attributing app installs and re-engagement to ads. It builds on SKAdNetwork with support for more ad formats and attribution scenarios. Apple returns signed postbacks with limited conversion data rather than a person-level activity trail.

An AI citation is a source link or reference attached to an AI-generated answer. It lets readers verify a claim and can send qualified traffic to the cited page. Citation formats and selection rules vary by product, query, geography, and available sources.

AI referral traffic is website traffic from links in AI assistants, answer engines, and AI-powered search results. Analytics can detect visits when the referrer or campaign tag is present. Some AI visits appear as direct traffic, so measured referrals are usually a partial total.

Android App Links are verified HTTPS links that open matching content in an Android app. The website publishes an association file and the app declares supported URLs. Verified links can bypass the app chooser, while the website remains the fallback when the app cannot open.

Anonymisation changes or removes identifying details so a person cannot reasonably be identified from the remaining data. Deleting names is rarely enough. Rare combinations, precise timestamps, locations, or linkable IDs can still reveal who a record describes.

An anonymous ID is an identifier assigned before a product knows who a visitor is, usually in a browser or app. It links events within that context until a known identity is available. Anonymous describes the account state, not whether the data is legally personal.

App Tracking Transparency is Apple's permission system for tracking users across apps and websites owned by different companies. Apps covered by Apple's tracking definition must show the ATT prompt before tracking or accessing the advertising identifier. ATT does not govern every form of in-app analytics.

Attribution assigns conversion credit to recorded marketing touchpoints. First-touch, last-touch, and data-driven models can credit the same journey differently. Use attribution to compare channels under one reporting rule, not as proof that a credited ad caused the conversion.

An attribution window is the time after an ad click or view during which a platform can credit a conversion to that interaction. Click and view windows often differ. Shorter windows usually report fewer attributed conversions even when total sales stay unchanged.

Average order value is total order revenue divided by the number of orders in the same period. If 100 orders generate $5,000, AOV is $50. State whether revenue includes tax, shipping, discounts, cancellations, and refunds before comparing reports.

BDDK is Turkey's Banking Regulation and Supervision Agency. It oversees banks and parts of the financial sector, including rules that can affect data systems, outsourcing, and service providers. BDDK is a regulator, not a privacy law.

A BigQuery export sends analytics or product data to Google BigQuery for SQL analysis and joins with other datasets. It offers more control over event-level records than a fixed dashboard. Schema, update frequency, query cost, and privacy settings determine what analysts can calculate reliably.

Bounce rate is the percentage of sessions that fail an analytics tool's engagement test. In Google Analytics 4, it is the inverse of engagement rate. Older tools often define a bounce as a single-page session, so compare rates only after checking the underlying rule.

The California Consumer Privacy Act is California's privacy law for covered businesses and residents. It creates rights to access, delete, and correct personal information, and to opt out of its sale or sharing. The California Privacy Rights Act expanded and amended these protections.

Cart abandonment happens when a shopper adds an item to a cart but does not purchase within a defined period. The rate depends on how the store identifies carts, returning shoppers, and completed orders. A cart counted as abandoned today can still convert later.

Channel grouping maps raw sources and mediums into reporting categories such as paid search, organic social, and email. Documented rules make reports easier to compare. Lost campaign tags or rule changes can move the same traffic into a different channel without changing customer behavior.

Checkout extensibility is Shopify's supported framework for customizing checkout with UI extensions, Shopify Functions, web pixels, and related APIs. Merchants can change checkout experiences and measurement without editing template code directly. Available extension points depend on the feature and Shopify plan.

A click ID is a platform-specific value added to a landing-page URL after an ad click. Google's GCLID and Meta's fbclid are common examples. If captured and stored correctly, the value can connect a later conversion to the original click.

A click-through conversion occurs when someone clicks an ad and completes a defined action inside the click attribution window. The platform connects the click and outcome through an identifier or account match. Tracking coverage, window length, and event rules all affect the count.

Client-side tracking records actions in a browser or mobile app, then sends them from the user's device to an analytics or advertising platform. Common methods include JavaScript tags and mobile SDKs. Browser settings, app permissions, consent, and blocked requests can limit the data collected.

CNAME cloaking uses a DNS alias to route a first-party subdomain to a third-party tracking service. The request appears to use the site's domain even though another provider processes it. Some browsers detect and restrict this pattern, and the provider remains a third party.

A cohort is a group of users who share a defined starting event or time period, such as customers acquired in September. Cohort analysis follows that fixed group to compare retention, revenue, or behavior over time without mixing in later users.

A consent management platform collects, records, and shares a visitor's privacy choices. It can display a consent banner, block tags, and send choices to vendors. The business still decides which activities need consent and must verify that every connected tool follows the recorded choice.

A conversion API sends marketing events from a server or backend system to an advertising platform. It can report purchases, leads, and other outcomes without relying only on a browser tag. Duplicate browser and server events need a shared ID so the platform counts them once.

Conversion rate is the percentage of eligible users, sessions, or clicks that complete a defined action. A store might divide purchases by sessions, while an ad platform divides conversions by clicks. Always state both the conversion event and denominator.

Conversions API Gateway is Meta's packaged server that copies Meta Pixel events and sends them to the Conversions API. It runs in a cloud account or through a managed host on a first-party subdomain, so advertisers get server events without writing an integration. It handles web events only.

Cookieless analytics measures website or app activity without storing browser cookies. It can use aggregate counts, server logs, or other event data instead. Cookieless does not mean anonymous or consent-free because IP addresses, device details, and event combinations may still identify a person.

Cookieless attribution links outcomes to marketing activity without relying on browser cookies. Common inputs include click IDs, signed-in first-party data, privacy-preserving postbacks, and aggregate models. Each method covers a different part of the journey and still requires appropriate consent and data controls.

Cookieless pings are the requests Google tags send in Consent Mode advanced mode when a visitor denies analytics or ad storage. They carry the event and consent state but no cookie identifiers. Google Analytics 4 and Google Ads use them to model the conversions and behaviour of visitors who declined.

Cost per acquisition is advertising cost divided by the number of defined acquisitions. If $1,000 in spend produces 20 qualified customers, CPA is $50. Always name the acquisition event and attribution window, because a signup and a paying customer produce different CPAs.

Cost per mille is the advertising cost for 1,000 impressions. If 10,000 impressions cost $80, the CPM is $8. CPM measures delivery cost, not attention, clicks, or sales, so it works best alongside outcome metrics.

Cross-device tracking connects activity from several devices to one person or account. Deterministic methods use a shared login, while probabilistic methods infer a link from weaker signals. Consent choices, platform rules, and shared household devices limit the accuracy of both approaches.

A custom rule is an if-then instruction that changes how a measured touchpoint is classified. A rule might map a known source and medium to Email before attribution runs. Test rules against real traffic and record each change because classification updates can reshape historical reports.

A custom tracking domain is a business-owned hostname used to collect event data, such as metrics.example.com. It gives teams a branded, manageable endpoint for a server-side pipeline. The domain alone does not guarantee longer cookie life, bypass browser controls, or grant permission to share data.

A custom weight assigns a chosen share of conversion credit to positions in a customer journey. A model might favor the first and last interactions over the middle. The weights express a reporting assumption, so they should total 100% and should not be presented as causal evidence.

Customer acquisition cost is total sales and marketing expense divided by the number of new customers acquired in the same period. CAC can include salaries, software, agencies, and advertising. That wider scope separates it from an ad platform's cost per conversion.

A customer data platform collects customer data from several sources, resolves identities, and creates profiles or audiences for analysis and activation. Its matching rules decide which records belong together. Data quality, consent, and destination controls remain separate responsibilities.

A customer journey is the sequence of interactions someone has with a business, from discovery through purchase, use, and renewal. Analytics captures only the touchpoints its systems can observe. Journey analysis reveals friction in that recorded path, not every influence on the customer.

Customer lifetime value estimates the revenue or profit a customer will generate over a defined period. Teams calculate LTV from historical purchases or a forecast. The time horizon, margin, refunds, and service costs must be clear before comparing it with acquisition cost.

Customer Match is a Google Ads feature for building audiences from a business's first-party customer data. Google matches uploaded contact details, often after hashing, to signed-in accounts across eligible inventory. Source data, targeting, exclusions, and consent must follow Google's policies.

Daily active users counts distinct users who complete a defined activity within one day. The rule might be opening an app or completing a meaningful product action. User identity and time zone affect the result, and daily totals cannot be added to calculate monthly users.

A data discrepancy is a difference between reports that appear to measure the same activity. Time zones, event definitions, consent coverage, attribution rules, currencies, and processing delays commonly create gaps. Align those settings and inspect raw events before deciding that either system is wrong.

Data governance defines who owns data, who can access it, and how teams collect, name, share, retain, and delete it. For event tracking, governance includes the tracking plan, consent rules, destination permissions, quality checks, and approval process for changes.

Data Hub is Hardal's workspace for building datasets and reports from marketing and product events. Teams combine metrics, dimensions, filters, and attribution views to answer specific questions. Reliable outputs depend on consistent event names, complete source data, and shared metric definitions.

A data layer is a structured object or event queue that exposes website activity to a tag manager. It can hold a product ID, order value, page type, or consent state. A stable data layer separates tracking logic from page markup and keeps event data consistent.

A data pipeline moves data from sources through processing steps to one or more destinations. It can validate, transform, enrich, deduplicate, and filter records along the way. Reliable pipelines log failures and retries so missing delivery does not look like missing customer activity.

A data processing agreement defines how one organization processes personal data for another. It usually covers instructions, security, subprocessors, rights requests, audits, and deletion or return. A DPA sets contractual duties, but does not replace a required legal basis or privacy notice.

Data quality measures whether data is accurate, complete, timely, consistent, and fit for a specific use. Event-tracking checks often cover missing fields, invalid values, duplicates, delays, and naming errors. A large dataset can still be low quality if it records the wrong behavior.

Data residency describes the country or region where data is stored or processed. A company might require EU customer records to remain in the EU. A complete policy also covers backups, remote access, subprocessors, and transfers to other locations.

Data retention is how long an organization keeps a dataset before deleting or anonymizing it. Different records can have different operational, reporting, or legal periods. A usable policy names the data, purpose, owner, retention period, and deletion method, including backups and downstream copies.

Data sampling analyzes a subset of records to estimate results for a larger dataset. It speeds up expensive queries but introduces sampling error, especially in small segments. Check whether a report is sampled before treating minor channel differences as real.

Data sovereignty means data falls under the laws and authority of a jurisdiction. Storage location can affect those rules, but so can the people in the data, the organization processing it, remote access, and cross-border transfers.

Data thresholding hides or groups small report segments that could reveal information about individuals. Google Analytics 4 can withhold rows even when it collected the underlying events. A missing row may therefore reflect a privacy threshold rather than missing tracking.

A data warehouse stores structured data from multiple systems for joint analysis. Marketing teams can query orders, ad spend, CRM records, and product events in one place. The warehouse provides storage and computation, while collection, identity, quality, and access rules require separate design.

Data-driven attribution analyzes observed conversion paths to estimate how much credit each touchpoint should receive. The result depends on the platform's model, data volume, and tracking coverage. It captures more journey patterns than a fixed rule, but remains an attribution estimate rather than a causal test.

Deep linking opens a specific screen or piece of content inside a mobile app. A product URL, for example, can open the matching product page instead of the app home screen. Good implementations also provide a useful web or store fallback when the app is unavailable.

Deferred deep linking sends a new user to the intended app content after installation and first open. The original link context must survive the app-store step. Platform privacy controls can limit that context, so the fallback should still open a useful screen.

A destination is a system that receives data from a pipeline, such as an analytics platform, warehouse, CRM, or ad network. Routing rules should define which events and fields it receives, why it needs them, and how the pipeline handles delivery failures.

The Digital Markets Act is an EU law for designated large digital platforms known as gatekeepers. It sets obligations around data use, interoperability, competition, and access to core platform services. The DMA is separate from GDPR, although both can shape advertising and measurement workflows.

E-commerce events record actions across a shopping journey, including product views, add-to-cart actions, checkout starts, purchases, and refunds. Consistent product, currency, order, and value fields make the funnel comparable and prevent a checkout start from being mistaken for completed revenue.

Engagement rate is the percentage of sessions that meet an analytics tool's engagement criteria. In Google Analytics 4, an engaged session lasts more than 10 seconds, records a key event, or includes at least two page or screen views. Other tools can define engagement differently.

The ePrivacy Directive is the EU framework for electronic communications privacy and access to information stored on a user's device. It covers many uses of cookies and similar technologies. EU countries implement it through national laws, which teams must read alongside GDPR.

An event is a recorded action or state change with a name, timestamp, and optional properties. Page views, app installs, signups, and purchases are common examples. Events can originate in a browser, mobile app, connected device, or backend system.

Event deduplication recognizes multiple records of the same action and counts them once. It prevents a browser pixel and server API from creating two purchases for one order. Most implementations match a shared event ID plus compatible event details.

Event enrichment adds trusted context to an event before delivery. A server might attach a product category from its catalog or a campaign name from an internal lookup. The added fields should come from documented sources and follow the same consent rules as the original event.

Event Match Quality is Meta's rating of how useful an event's customer information is for matching it to a Meta account. Permitted email, phone, browser, and click identifiers can affect the score. EMQ measures matchability, not whether a purchase happened or whether the data is accurate.

Extract, load, transform is a pipeline pattern that copies source data into a destination before reshaping it there. ELT preserves raw input and uses the destination's compute engine for modeling. Access controls, validation, cost management, and retention still apply to the loaded data.

Extract, transform, load is a pipeline pattern that reads source data, cleans or reshapes it, then writes it to a destination. Transforming first can enforce a common schema before storage. Reliable ETL also handles failed records, reruns, and source-field changes.

_fbp and _fbc are Meta browser identifiers used for event matching. _fbp identifies a browser, while _fbc stores a Facebook click ID when one is available. They serve different purposes and should be collected and sent only under the site's approved consent setup.

A first-party cookie belongs to the website a visitor is using. Sites use it to maintain sessions, remember settings, or store an identifier between requests. Browser limits and consent rules still apply, and a first-party cookie does not make every downstream data use first party.

First-party data is information a business collects through direct interactions with customers or visitors. Purchases, account details, support requests, and on-site events are common examples. Direct collection gives the business more control, but the data still needs a clear purpose and sharing rules.

First-touch attribution gives all conversion credit to the earliest eligible interaction in the recorded journey. It highlights channels that introduce customers to the business. The result depends on the lookback window and tracking history, since an older unseen touchpoint cannot receive credit.

A funnel is an ordered sequence of steps toward an outcome, such as product view, add to cart, checkout, and purchase. Funnel analysis shows how many observed journeys reach each step. Order requirements, time limits, and session boundaries change the result.

GA4 Data Import uploads external data into a Google Analytics 4 property, such as cost data from non-Google ad platforms, item details, user attributes or offline events. GA4 joins each file to collected data on a key field, like campaign ID, item ID or user ID.

The General Data Protection Regulation is the European Union's main law for processing personal data. It sets rules for lawful basis, transparency, security, data minimization, and individual rights. GDPR scope depends on the organization's activity and audience, not only the location of its servers.

Generative engine optimization improves how clearly AI search and answer systems can discover, understand, and cite website content. GEO favors crawlable pages, direct answers, specific evidence, strong internal links, and credible sources. It overlaps heavily with technical SEO and good editorial structure.

Global Privacy Control is a browser or device signal that tells websites a user wants to opt out of certain data sales or sharing. Sites can detect the signal and apply it to relevant processing. The required response depends on the privacy law covering the interaction.

The Google Advertising ID is Android's resettable identifier for eligible advertising measurement and personalization. Access depends on device settings, Google Play policies, and user choices. Apps should expect the value to be unavailable or reset and should not use it as a permanent account identifier.

Google Consent Mode passes a visitor's consent choices to Google tags and SDKs. Those tools adjust cookie storage and measurement behavior based on the signal. Consent Mode works with a separate banner or consent platform, and its basic and advanced setups collect different data after a denial.

Google Enhanced Conversions uses hashed first-party customer data to improve conversion matching in Google Ads. An advertiser might send a normalized, hashed email with a recorded purchase. The feature supplements the conversion event and must follow Google's eligibility, consent, and data-use requirements.

The Google Play Install Referrer API gives an Android app referral details associated with its Play Store install. Campaign parameters and click timestamps can help connect the install to a marketing link. The data describes the store referral, not every earlier ad view or cross-device interaction.

The Gramm-Leach-Bliley Act is a US law governing how financial institutions handle customers' nonpublic personal information. Its rules cover privacy notices, information sharing, and security safeguards. GLBA applies to defined institutions and activities, not to every company collecting marketing data.

gtg_health=1 is a query parameter on a lightweight Google Tag Manager request that checks whether the Google tag loaded correctly. The request can go directly to googletagmanager.com even when other tag traffic uses a first-party or server-side endpoint. It is diagnostic traffic, not a conversion event.

Hashed PII is personal information transformed with a one-way hashing function before transmission. Ad platforms commonly use hashed emails or phone numbers for event matching. Hashing protects the raw value in transit, but it does not make the data anonymous or remove privacy obligations.

The IAB Transparency and Consent Framework is a standard for communicating privacy choices across publishers, consent platforms, and advertising vendors. It stores purpose and vendor choices in a TCF string. The framework carries the signal, but does not determine whether a specific data use is lawful.

The Identifier for Advertisers is Apple's resettable device identifier for eligible advertising measurement and personalization. Apps need App Tracking Transparency authorization before accessing it for tracking. A user can deny access or reset the identifier, so apps cannot treat IDFA as a permanent user ID.

The Identifier for Vendors is an Apple-generated device identifier shared across apps from the same vendor. Developers can use it to relate activity within their own app portfolio. IDFV is separate from IDFA and cannot be used to bypass Apple's cross-company tracking rules.

An identity graph stores links among identifiers such as account IDs, devices, cookies, and permitted email hashes. It helps connect records across systems under defined matching rules. Each link needs a source and confidence level because one bad match can contaminate many profiles and reports.

Identity resolution links records that may belong to the same person or account across sessions, devices, and systems. A signed-in account ID provides stronger evidence than a probabilistic device match. Clear rules should define allowed identifiers, match confidence, and how conflicting links are resolved.

Incrementality is the additional result caused by a marketing activity compared with what would have happened without it. Holdout tests and controlled experiments estimate this lift. Attribution assigns credit within observed journeys, while incrementality asks whether marketing changed the total outcome.

Install attribution assigns an app install to an earlier eligible ad interaction. Evidence can come from click IDs, app-store referral data, or privacy-preserving platform postbacks. Ad networks and measurement partners may credit the same install differently because their windows, matching methods, and rules differ.

Intelligent Tracking Prevention is Safari's system for limiting cross-site tracking. It restricts some cookies, storage methods, and request patterns used to recognize people across websites. Because Safari updates ITP over time, teams should test real implementations instead of relying on a fixed cookie lifetime.

Last non-direct click credits the most recent recorded marketing interaction before a conversion and ignores a later direct visit. It prevents direct traffic from replacing a known channel near checkout. Inconsistent UTM tags or channel rules can still move credit between sources.

Last-touch attribution gives all conversion credit to the final eligible interaction before the outcome. It is easy to explain, but often favors channels close to purchase, such as branded search and retargeting. The attribution window and treatment of direct traffic shape the result.

LinkedIn Conversions API sends conversion events from a server to LinkedIn Campaign Manager. It can report outcomes such as qualified leads and completed purchases after the original website visit. Accurate timestamps, match keys, and deduplication settings connect those events to eligible ad interactions.

An LLM crawler is an automated bot that fetches web content for an AI provider's search, retrieval, or model-development systems. Providers use different user agents for different purposes. Site owners can review server logs and provider documentation before setting access rules.

llms.txt is a proposed root-level text file that points language-model tools to useful site content. It is a community convention, not a web standard, ranking factor, or replacement for crawlable HTML and sitemaps. Publishing the file does not guarantee discovery, retrieval, or citation.

Marketing attribution connects conversions to earlier marketing interactions and divides credit among them. Results change with tracking coverage, attribution windows, and the chosen model. It supports channel reporting and budget analysis, while experiments are better suited to measuring causal lift.

Marketing mix modeling uses aggregated data such as weekly sales, media spend, price, promotions, and seasonality to estimate channel contribution. It does not require user-level tracking. Reliable results depend on enough historical variation, sound model assumptions, and validation against experiments.

Measurement Protocol is the Google Analytics 4 HTTP API for sending events from a server or another non-browser system. It suits backend events such as offline purchases and refunds. Each request needs an API secret, and a client or app instance ID to join an existing user.

Meta Conversions API is Meta's server interface for receiving events such as purchases, leads, and registrations. It often runs alongside the Meta Pixel. When both send the same action, matching event names and IDs help Meta deduplicate the records.

Meta Signals Gateway is Meta's data routing tool for collecting first-party events from websites, apps, CRMs and offline sources. It can forward those events to Meta and to other destinations. It extends the Conversions API Gateway model beyond web events and runs on infrastructure the advertiser controls.

Mobile analytics measures app acquisition, usage, performance, and retention through events such as installs, screen views, purchases, crashes, and returns. SDKs and server pipelines collect the data. Reports depend on event definitions, identity rules, permissions, and delayed offline uploads.

A mobile measurement partner collects app events and reconciles campaign results across ad networks. It connects an SDK or server events with network attribution APIs to report installs, re-engagement, and conversions. Results depend on platform permissions, attribution windows, network rules, and available identifiers.

Monthly active users counts distinct users who complete a defined activity within a calendar month or rolling 30-day window. Each user counts once per period. Reports should state the activity rule, window type, and identity method before comparing products.

Multi-touch attribution gives conversion credit to several recorded interactions in a customer journey. A linear model splits credit evenly, while position-based and data-driven models use different weights. Missing devices, untracked views, and consent gaps change the path the model can see.

Offline conversions are outcomes recorded outside a website or app, then uploaded to a measurement platform. Examples include signed contracts, phone sales, and in-store purchases. Matching them to earlier marketing activity requires a permitted identifier, an accurate conversion time, and consistent event rules.

Order ID deduplication uses a stable order identifier to prevent multiple records of one purchase from inflating revenue. It handles retries and duplicate browser or server delivery while preserving real later transactions and refunds. Some destinations also require a separate event ID.

A pageview is an event recorded when someone loads or views a web page. Single-page apps must often send a new pageview after a route change even without a browser reload. Repeat views count again, so pageviews measure content views rather than unique people.

KVKK commonly refers to Turkey's Personal Data Protection Law No. 6698 and sometimes to its data protection authority. The law governs personal-data processing and grants rights to data subjects. Tracking programs in Turkey must account for its notice, legal-basis, security, and transfer rules.

Personally identifiable information is data that identifies a person on its own or when combined with other data. Names and email addresses are obvious examples, while persistent device or account identifiers can also qualify. Privacy laws define personal information differently, so PII is not a universal legal test.

A postback is a server message confirming that an event or attribution result occurred. In mobile advertising, an operating system, ad network, or measurement partner can send one after an install or in-app outcome. The sender's rules determine its fields, timing, and privacy limits.

Privacy-first analytics collects the minimum data needed for a defined measurement purpose. It applies consent choices, limits access and retention, and controls which fields reach each destination. The label describes the full data practice, not a specific analytics tool or collection method.

Probabilistic matching estimates whether records belong to the same device or person without a direct shared identifier. It combines weaker signals and assigns a likelihood rather than a confirmed link. False matches are possible, and platform policies or privacy laws may restrict the inputs.

Pseudonymisation replaces direct identifiers with codes while keeping a separate way to reconnect the data to a person. It reduces exposure when the linking key is protected. The records remain personal data if re-identification is reasonably possible.

Re-engagement is a user's return to an app or service after a defined period of inactivity. An ad, push notification, or email can prompt the return. Campaign reports should state the inactivity threshold and distinguish an app open from a later purchase or subscription.

Real-time analytics shows events within seconds or minutes of collection. Teams use it to monitor launches, campaigns, and tracking changes. Counts remain provisional because late events, deduplication, attribution, and processing rules can update the final report.

Reddit Conversions API sends advertiser events to Reddit from a server. It supports purchases, signups, leads, and other configured conversions without relying only on Reddit Pixel. A shared deduplication key prevents Pixel and API copies of one action from inflating results.

Retention rate is the percentage of a starting cohort that returns or remains active after a set period. Day-30 retention, for example, measures users who complete the chosen return event around day 30. The cohort, event, and time window must be stated together.

Return on ad spend is attributed revenue divided by advertising cost. A campaign that produces $500 in attributed revenue from $100 in spend reports a 5x ROAS. The result changes with the revenue definition, attribution model, conversion window, and refund handling.

Reverse ETL syncs prepared warehouse data into operational tools such as CRMs, support systems, and ad platforms. A team might send a consented customer segment to a campaign tool. Each sync needs clear fields, update frequency, ownership, and removal or opt-out behavior.

Schema validation checks data against expected names, types, formats, and required fields. A purchase event might require an order ID, currency code, and numeric value. Pipelines can reject invalid records or quarantine them for repair before they corrupt reports.

A self-attributing network is an ad platform that uses its own impression and click data to claim conversions. Mobile measurement partners send eligible outcomes to the network and receive attribution responses. Teams should compare each network's windows and rules before reconciling reports.

Server-side Google Tag Manager is a Google Tag Manager container that runs on a server. It receives event requests, turns them into a common format, and forwards selected fields through tags. Teams often place the endpoint on a first-party domain and manage destinations centrally.

Server-side measurement processes analytics events on a server before they reach reporting or advertising platforms. The server can validate fields, remove unwanted data, apply consent choices, and control destinations. It improves data control, but cannot recover an action that was never captured.

Server-side tracking routes events through a business-controlled server before sending approved data to analytics and advertising platforms. Teams can validate, enrich, filter, and route each event in one place. Device actions still need client-side collection, and consent rules still apply.

Server-to-server tracking sends data directly between backend systems through an API or webhook. An order service, for example, can report a paid purchase without waiting for the checkout page to stay open. S2S is one delivery method within a server-side tracking setup.

A session groups website or app events into one visit under an analytics tool's rules. Many tools end a session after a period of inactivity. Timeout, time zone, campaign, and identity settings can change the total, and one person can create several sessions.

Shopify Web Pixels are app or custom pixels that subscribe to Shopify customer events and send selected data to analytics or marketing tools. Standard events include product_viewed and checkout_completed. Shopify runs app pixels in a sandbox and applies the store's customer privacy settings.

Signal loss is the gap between customer actions and the usable data a measurement system receives. Consent choices, blocked requests, app permissions, missing identifiers, and broken tracking can all create it. Teams estimate the gap by comparing platform data with first-party business records.

Signal optimization improves the quality, timing, and usefulness of the events sent to advertising, analytics, and measurement platforms. It covers event quality, matching, deduplication, value fields, consent rules, and destination delivery so campaign and product decisions use cleaner inputs.

SKAdNetwork is Apple's privacy-preserving framework for measuring app ad installs and selected post-install activity. Apple sends delayed attribution postbacks with limited detail instead of exposing a user-level journey. Conversion values and privacy thresholds determine how much campaign information advertisers receive.

A software development kit is a packaged set of code, documentation, and tools for adding a service to an app. A measurement SDK can collect events and send them to an analytics provider. Teams should review its data access, permissions, performance, security, and update process.

A source is a system or touchpoint that produces data for a pipeline. Websites, mobile apps, checkout services, CRMs, and databases are common sources. Recording the source helps teams trace errors and apply the right schema, consent, and ownership rules.

Streaming processes events as they arrive, while batch processing handles a collected set on a schedule. Streaming supports fast alerts and activation. Batch simplifies large reconciliations and backfills. Many data systems combine both to account for late events and corrections.

A tag management system manages analytics and marketing tags from one interface. Triggers decide when each tag runs and variables determine which data it receives. Web containers run in the browser, while server containers process incoming requests on a server.

A third-party cookie belongs to a domain other than the website a visitor opened. Ad networks have used these cookies to recognize browsers across sites for targeting and measurement. Modern browsers restrict them, so reliable measurement cannot assume they are available.

TikTok Events API receives website and app events from a server or approved partner. Advertisers use it to report registrations, purchases, and other outcomes with less dependence on browser delivery. Event IDs prevent double counting when TikTok Pixel reports the same action.

A tracking pixel reports a view or action through a small network request. Traditional pixels load a tiny image in a page or email. Ad platforms also use the term for browser scripts that send events such as purchases, registrations, and page views.

A tracking plan documents the events a product collects, when they fire, and which properties they include. It also defines owners, naming rules, and destinations. Teams use the plan to keep web, app, and backend data consistent and catch missing or duplicate events.

Universal Links are verified HTTPS links that open matching content in an iOS app. Apple requires the website and app to publish an association. If the app is missing or cannot handle the route, the same URL opens as a normal web page.

A user ID is a stable internal identifier assigned to a known account or person, often after login. It can join permitted activity across sessions and devices. A separate account key is usually safer than using raw personal details such as an email address.

UTM parameters are URL query values that label a visit's source, medium, campaign, term, and content. Analytics tools use them to group traffic and compare campaigns. Consistent lowercase names prevent values such as Newsletter and newsletter from splitting into separate report rows.

utm_source=chatgpt.com is a campaign parameter ChatGPT adds to many links it shows in answers. Analytics tools read it as the traffic source, so the visit is reported as a ChatGPT referral instead of direct traffic. Links copied out of ChatGPT or opened in its apps may arrive without it.

Value-based bidding uses conversion values such as revenue, margin, or predicted customer value to guide automated ad bids. Consistent, timely values help the system favor higher-value outcomes. Duplicate purchases, inflated values, and missing refunds can train it toward the wrong goal.

A view-through conversion occurs when a platform credits a conversion after an ad impression with no recorded click. The conversion must happen inside the platform's view window. This shows an association between exposure and outcome, not proof that the impression caused the sale.

Web-to-app is a journey that moves a mobile visitor from a website into a native app. Deep links, banners, and install prompts can preserve the product or task across the handoff. Measurement should separate link clicks, installs, app opens, and in-app conversions.

A webhook is an HTTP request sent automatically when a specific event occurs. A checkout platform might notify an analytics pipeline when an order is paid. Receivers should verify the sender, handle retries, and use stable IDs to prevent duplicate processing.