114 terms from Analytics and Tracking, each defined in a sentence or two, with a link to the calculator that works the number out where one exists.
Reporting on ad delivery and cost, usually from the platforms themselves.
The specific case where a tool estimates from a subset once data volume passes a threshold.
A generated identifier for someone not logged in.
Pulling numbers programmatically instead of exporting by hand.
Reports showing credit across sources and paths.
The rules deciding who belongs to a segment.
An event that adds someone to an audience automatically.
Analysis of actions rather than demographics.
Streaming raw analytics data into a warehouse. The escape hatch from sampled, aggregated reporting.
Removing automated traffic from reports.
Any measurement that depends on the visitor browser cooperating.
Assigning credit for a call to a source.
Attributing phone calls back to the ad or keyword that produced them.
Open the calculator →A numeric identifier for a campaign, more stable than its name.
A unique identifier attached to a click so a conversion can be matched to it later.
The analytics identifier for a browser. Lost when cookies clear, which is often.
Recording events in the browser. Easy to set up, and increasingly lossy.
Grouping people by when they joined and tracking them over time. The only honest way to see retention.
Open the calculator →Making sure one sale is counted once across channels. Without it, channel reports sum to more than the business made.
A snippet that fires when a conversion happens.
Any tag whose purpose is to record a completed action.
Matching identifiers between two ad systems. Central to programmatic, and quietly dying.
Identifying returning visitors with a stored file. Shrinking in usefulness every year.
An identifier stitching one person across several devices.
Keeping one session alive across separate domains, such as a site and a hosted checkout.
The same across subdomains. Easier, and still commonly broken.
An extra attribute you define and send, such as customer type or margin band.
A numeric value you define and send yourself.
Analysis of who your customers are and what they are worth over time.
Open the calculator →Whether what was captured matches reality.
Whether everything that happened was captured.
Whether the same question gets the same answer across tools.
Sending raw data out for analysis elsewhere.
How recent the data is. Decisions made on stale data feel confident and are not.
The rules for who can define, change and access data.
Bringing outside data into an analytics tool, such as cost or product margin.
A structured object on the page holding values for tags to read. Stops tags scraping the DOM and breaking on redesigns.
Whether the numbers can be trusted enough to act on.
Comparing two sources and explaining the gap. Platform revenue versus the finance system, usually.
Checking incoming data against expected rules before it reaches a report.
Turning numbers into something a human can read quickly.
A match based on a shared known identifier. Accurate, and limited in reach.
An identifier tied to a phone or tablet rather than a browser.
The wider practice covering sites, apps and campaigns together.
Short for dynamic number insertion.
Showing a different phone number per visitor so calls can be attributed.
Analysis built on discrete tracked actions instead of page views.
Removing the same event counted twice.
An identifier used to match the same event arriving twice from different sources.
The structure of each event and the properties it carries.
A tag firing on a specific interaction rather than a page load.
The naming system for events. Get it wrong once and every future report inherits the mess.
A short summary for people who will not open a second tab. Fewer numbers, more context.
A free-form analysis view rather than a fixed dashboard.
The Meta click identifier.
A cookie set by the site being visited. Still broadly accepted, with shortened lifespans on some browsers.
The Google Marketing Platform conversion tag, used with Campaign Manager and DV360.
Measuring drop-off between defined steps. Shows where to work rather than how bad things are.
Open the calculator →The most common tag manager. Powerful, and a common cause of duplicate conversions when several people have access.
Updating only what changed instead of rebuilding the whole dataset.
The LinkedIn tracking tag.
Visits from your own team. Filter it or your conversion rate is quietly wrong.
Clicks or impressions that should not be billed. Platforms refund some of it, never all.
A view restricted to the handful of metrics that actually matter.
Channel and campaign performance in one place.
A written agreement on what gets measured and why, before anything is built.
Analysis across media channels rather than inside one platform.
The Microsoft Advertising click identifier.
Sending sales completed outside the site back into ad platforms. The fix for lead campaigns optimising toward noise.
Open the calculator →The same for any non-web action, not only purchases.
The same value as your ecommerce system knows it.
Studying the routes people take through a site or funnel.
A live view of the numbers a team acts on.
A cookie that survives between visits, up to its expiry.
A match based on inference. Wider reach, and some of it is wrong.
Measurement of behaviour inside a product rather than on marketing pages.
The same with cost subtracted. Rarer, and the version that changes decisions.
Open the calculator →Telling analytics to ignore certain referrers, usually payment providers, so returning visitors do not start a new session.
A tag that adds visitors to an audience list.
The part of the stack that turns raw data into readable output.
A pixel whose job is to build an audience of visitors.
Analysis of where money comes from and how reliably.
Reporting on a subset instead of everything. Fast, and it hides small segments.
Comparing subsets rather than trusting the average. Where the real story usually is.
Matching browser and server versions of the same event, usually by event ID.
Sending events through your own server before they reach vendors. Faster pages and better data control.
The general practice of recording events server-side rather than in the browser. Survives blockers.
A cookie deleted when the browser closes.
An identifier for a single visit.
Junk referrals that inflate sessions and depress conversion rate.
Managing tracking scripts without redeploying the site.
The tool that does it.
A cookie set by another domain. Blocked by default in most browsers now.
A general term for any small snippet reporting an event back.
The technical version, listing every event and property to be captured.
The document developers implement from. Vague specs produce data nobody trusts.
A unique identifier for an order. Essential for deduplicating conversions.
The TikTok click identifier.
A report showing individual user journeys. Useful for debugging, useless for strategy.
Your own identifier for a logged-in person. The most durable identifier you control.
Connecting records that belong to the same person.
The tagging convention for identifying traffic sources in analytics.
The campaign name. Inconsistent naming here wrecks more reporting than any tool ever fixes.
Which specific creative or link was used.
The type of traffic, such as cpc, email or social.
Keeping campaign parameters through a multi-page or multi-session journey. Without it, most conversions look direct.
Where the visit came from, such as google or newsletter.
The keyword, mostly used in manual search tagging.
Measurement of what people do on a website. The oldest and still the messiest part of the stack.