Analytics Engineering — Key Terms
Definitions written for marketing and growth teams — not just engineers. Every term here is relevant to how your tracking infrastructure works and why it matters for your ad performance.
GA4 (Google Analytics 4)
Google's current analytics platform, replacing Universal Analytics. GA4 uses an event-based data model instead of session-based, which means every user interaction is tracked as an individual event with parameters. Proper GA4 implementation requires configuring events, parameters, and conversions that align with your actual business KPIs — not just accepting default settings.
GTM (Google Tag Manager)
A tag management system that lets you deploy and manage tracking tags on your website without modifying code directly. GTM uses a container system with tags, triggers, and variables. A well-structured GTM container follows naming conventions and documentation standards so any team member can understand and maintain it — not just the person who built it.
Event-Based Tracking
A tracking model where every user interaction — page views, clicks, form submissions, purchases — is recorded as a discrete event with associated parameters. GA4 runs on this model. The quality of your data depends entirely on how well your events are defined, named, and validated against your actual business logic.
Conversion Event
A specific user action that you've marked as a key business outcome in GA4 or your ad platform. Examples include purchases, form submissions, phone calls, and booking confirmations. Misconfigured conversion events — firing too early, too late, or multiple times — are one of the most common causes of inflated ROAS and misallocated ad spend.
Data Layer
A JavaScript object that sits between your website and your tag management system. It passes structured data — transaction IDs, product information, user attributes — to GTM without relying on scraping the page DOM. A properly implemented data layer is the foundation of reliable tracking and makes your setup resilient to website design changes.
Event Dictionary
A document that defines every tracked event on your site — its name, trigger conditions, associated parameters, and expected values. An event dictionary is a deliverable, not an optional extra. Without it, your tracking setup becomes unmaintainable the moment the person who built it is no longer available.
UTM Parameters
URL parameters (utm_source, utm_medium, utm_campaign, utm_content, utm_term) that pass campaign attribution data into your analytics platform. Inconsistent UTM naming — mixing uppercase and lowercase, using different conventions across platforms — is one of the leading causes of fragmented attribution data and unreliable channel reporting.
Server-Side Tracking
A tracking architecture where data is sent from your web server to analytics platforms and ad networks — rather than from the user's browser. This bypasses ad blockers, browser privacy restrictions, and iOS tracking limitations. Server-side tracking is the most reliable way to recover signal lost after iOS 14 and browser privacy updates.
sGTM (Server-Side GTM)
Google Tag Manager deployed on a server rather than in the browser. sGTM acts as a proxy between your website and destination platforms like GA4 and Meta. It enables server-side event processing, data enrichment before sending, and improved data control — including the ability to strip or modify data before it reaches third-party platforms.
First-Party Data
Data collected directly from your own users through your own properties — your website, app, or CRM. In a privacy-first environment where third-party cookies are being deprecated, first-party data collected through server-side infrastructure is the most durable and reliable signal available to ad platforms for optimization and attribution.
Custom Domain for sGTM
Deploying your server-side GTM container on a subdomain of your own domain (e.g., metrics.yourbrand.com) rather than on a generic Google Cloud URL. This makes the tracking requests appear as first-party, reducing the likelihood of being blocked by browser privacy protections and ad blockers.
Signal Loss
The percentage of real user actions — purchases, leads, conversions — that go unrecorded by your tracking setup due to ad blockers, iOS privacy restrictions, or browser limitations. Signal loss typically ranges from 20–60% depending on your audience and traffic sources. Unaddressed signal loss causes ad platforms to optimize against incomplete data.
Meta CAPI (Conversions API)
Meta's server-side API for sending conversion events directly from your server to Meta — bypassing browser-based pixel limitations. CAPI is the primary solution for recovering iOS signal loss in Meta Ads. Proper CAPI implementation requires event deduplication logic to prevent the same conversion from being counted twice (once from the pixel, once from the server).
Event Deduplication
The process of ensuring that the same conversion event is not counted multiple times when it arrives through multiple channels — for example, once from the browser pixel and once from the server-side API. Deduplication in Meta requires matching event_id values between pixel and CAPI events. Missing deduplication is one of the most common causes of inflated conversion counts.
Event Match Quality (EMQ)
A Meta metric (scored 0–10) that measures how well your conversion events can be matched to Facebook users. Higher EMQ means better attribution and more efficient ad delivery. EMQ improves when you send enriched customer data — email, phone, name — alongside conversion events through CAPI.
Google Enhanced Conversions
Google's server-side conversion solution that supplements browser-based conversion tracking by sending hashed first-party customer data (email, phone) to Google. Enhanced Conversions improve attribution accuracy for Google Ads campaigns — particularly for users who have opted out of cookies or use browsers that restrict third-party tracking.
ROAS (Return on Ad Spend)
Revenue generated divided by ad spend. ROAS is the primary performance metric for most paid media campaigns — but it is only as accurate as the tracking data feeding it. Inflated ROAS caused by duplicate conversion events or misattributed purchases leads to budget decisions that scale underperforming campaigns and cut winners.
View-Through Attribution
An attribution model that gives credit to an ad impression — even if the user never clicked — when a conversion happens within a defined window after seeing the ad. View-through attribution significantly inflates reported conversions for display and video campaigns and should be evaluated carefully against your actual conversion data.
BigQuery
Google's cloud data warehouse, part of Google Cloud Platform. In a marketing analytics context, BigQuery is used to store raw event data from GA4, combine it with ad platform data, and run custom attribution models and cohort analyses that are not possible inside GA4's interface. BigQuery gives you full SQL access to your data without sampling.
Looker Studio
Google's free business intelligence and data visualization tool, formerly known as Data Studio. Looker Studio connects to GA4, BigQuery, Google Ads, and other sources to build dashboards. The quality of a Looker Studio dashboard is entirely dependent on the quality of the underlying data — garbage in, garbage out.
Multi-Touch Attribution
An attribution model that distributes conversion credit across multiple touchpoints in the customer journey — rather than giving all credit to the first or last click. Multi-touch attribution requires clean, deduplicated event data across all channels and is typically implemented in BigQuery where cross-channel data can be joined and modeled.
Data Sampling
A GA4 limitation that applies when queries involve large datasets — GA4 analyzes a subset of data rather than the full dataset and extrapolates results. Sampling introduces inaccuracy in reports. BigQuery export of raw GA4 data eliminates sampling by allowing queries to run against the complete, unsampled dataset.
Cross-Domain Tracking
A GTM and GA4 configuration that allows user sessions to be tracked continuously across multiple domains — for example, from a marketing site to a checkout domain. Without cross-domain tracking, GA4 treats the checkout as a new session with direct traffic as the source, breaking attribution for any conversion that spans two domains.
Revenue Attribution
The process of connecting ad spend and marketing activity to actual revenue generated. Accurate revenue attribution requires clean purchase event data with correct revenue parameters, matched against your actual order management system or CRM. A common benchmark is keeping GA4 revenue within ±2% of Shopify or your source of truth.
Consent Mode (Google)
A Google framework that adjusts how GTM tags behave based on user consent choices. When a user declines cookies, Consent Mode signals this to Google tags — which then operate in a cookieless mode and model behavior using aggregated data. Consent Mode v2 is required for Google Ads conversion tracking in the EU and is best practice globally.
iOS 14+ Impact
Apple's App Tracking Transparency (ATT) framework, introduced in iOS 14.5, requires apps to request permission before tracking users across apps and websites. This significantly reduced the signal available to Meta and other ad platforms from iOS users — resulting in underreported conversions, inflated CPMs, and degraded ad algorithm optimization. Server-side tracking with CAPI is the primary technical mitigation.
Third-Party Cookie Deprecation
The phased removal of third-party cookies from major browsers — already completed in Firefox and Safari, and in progress in Chrome. Third-party cookies were the foundation of most browser-based ad tracking and attribution. Their deprecation makes server-side tracking and first-party data collection essential for maintaining attribution accuracy.
Tracking Audit
A systematic review of your existing GA4, GTM, and ad platform tracking configurations to identify measurement gaps, duplicate events, misconfigured conversions, and signal loss. A proper audit produces a prioritized fix roadmap (P0 critical, P1 high, P2 medium) with an estimate of revenue impact from each gap.
QA Validation
The process of testing every tracking event against real traffic — not just GTM preview mode — to confirm it fires correctly, passes the right parameters, and deduplicates properly. QA validation is the step most agencies skip. Without it, you cannot confirm your implementation is actually working in production.
P0 / P1 / P2 Prioritization
A framework for prioritizing tracking fixes by business impact. P0 = critical issues causing immediate revenue measurement failure (missing purchase events, duplicate conversions). P1 = high-priority gaps affecting attribution quality (broken UTMs, missing lead type segmentation). P2 = medium-priority improvements that enhance reporting but don't break core measurement.
Your tracking setup has gaps. The audit finds them.
Free for the founding cohort. Written report + recorded walkthrough. No obligation.
Get a Free Tracking Audit →