Skip to content
SealMetrics
Part 04 · Where we goChapter 11

Glossary

Opinionated definitions of the terms used across Open. Where we use a word differently from the rest of the industry, we say so.

7 min readLast revised · May 2026

Most glossaries are neutral on purpose. This one isn't. We use a handful of words in ways that differ from how the rest of the analytics industry uses them — and a few words we wish the industry would stop using altogether. This chapter is where we say which is which.

Each term links to the chapter that develops it. If you arrived here from another chapter looking for a one-line definition, the link beneath every term will take you back to the full treatment.

Method and measurement

How we describe what we do and what we refuse to do. The vocabulary that defines complete data, last-click on 100%, and the things we say no to.

Complete data#complete-data

Measured. Not sampled, not modelled, not filled in by a regression when consent is rejected. Every visitor, every event, captured before the cascade fires.

Develops in chapter 03.

Last-click attribution on 100%#last-click-attribution

Credit assigned to the final measurable channel before conversion, applied across the entire population — not the consenting fraction. The model is simple by design; what is not simple is feeding it complete data.

Develops in chapter 09.

Multi-touch attribution#multi-touch-attribution

Distributing credit across the channels involved in a conversion path. Requires per-user, cross-session journey reconstruction. In Europe that journey is reconstructable for roughly 13% of visitors. We do not do this.

Develops in chapter 09.

Sampling#sampling

Analysing a subset of events and extrapolating the result as if it were a full measurement. GA4 applies sampling above 10M events per query. We don't.

Develops in chapter 03.

Modelling (modelled conversions)#modelling

Statistical fills for events the analytics tool cannot measure. Output is presented as a number; source is a regression. Acceptable for directional analysis. Dangerous when used to allocate media budget.

Develops in chapter 03.

Server-side measurement#server-side

Event ingestion that happens on a server the customer controls, not in the browser. Survives ad blockers, ITP, and consent rejection because the request never has to leave the first-party domain.

Develops in chapter 05.

Legal compliance

Regulatory language used across Open. We name the article numbers and the rulings so a DPO can pick up the chapter cold.

GDPR by architecture#gdpr-by-architecture

Compliance produced by what the data model never collects, rather than by what a consent UX manages to obtain. The product cannot be configured to handle personal data because no field for it exists.

Develops in chapter 06.

ePrivacy Article 5(3)#eprivacy-article-5-3

European rule that governs access to terminal equipment. Cookies, localStorage, fingerprinting all trigger it. We do none of them.

Develops in chapter 06.

Schrems II#schrems-ii

Court of Justice EU ruling, July 2020, that invalidated the Privacy Shield framework for US data transfers. The reason our infrastructure runs in Dublin with zero US sub-processors for the data plane.

Develops in chapter 06.

DPA (Data Processing Agreement)#dpa

Contract signed before data flows. We are the processor, you are the controller. Covers scope, lawful basis, sub-processors, TOMs, rights handling, breach notification. Included by default in Scale plans.

Develops in chapter 06.

Legitimate interest (Art. 6(1)(f))#legitimate-interest

Lawful basis under GDPR for processing without consent when the interest does not override the data subject's rights. Documented in a Legitimate Interest Assessment (LIA), bundled with the DPA.

Develops in chapter 06.

Technical architecture

How the pixel, the storage, and the data plane are built. Where industry words mean something specific in our case, we say so.

Cookieless#cookieless

Measurement that does not set or read cookies. The word is overused — many tools that call themselves cookieless rely on fingerprinting or localStorage instead, which trigger ePrivacy Article 5(3) anyway. We are cookieless in name and in fact.

Develops in chapter 05.

Fingerprinting#fingerprinting

Combining terminal signals — UA, canvas, fonts, audio context, IP — into a near-unique browser identifier. Functionally equivalent to a cookie under ePrivacy Article 5(3). We strip the vectors before storage.

Develops in chapter 09.

First-party measurement#first-party

Events flow through the customer's own domain, not a third-party endpoint. Survives ad blockers and cross-site cookie restrictions because there is no third-party request to block.

Develops in chapter 05.

Intelligent Tracking Prevention (ITP)#itp

Safari's mechanism for limiting cross-site tracking. Expires or blocks cookies that rely on cross-site context. One of the three layers in the cascade.

Develops in chapter 03.

Pixel#pixel

The small client-side script that captures and dispatches events to ingestion. Ours is under 4 KB gzipped, loads asynchronously, and sets no terminal storage.

Develops in chapter 05.

Commercial vocabulary

Words used about the customer, the spend, and the canonical figures repeated across Open.

The 13%#the-13-percent

The canonical estimate of GA4 measurement coverage in the EU after consent rejection (~55%), ad blockers (~40% of the remainder), and browser restrictions. The end of the cascade.

Develops in chapter 03.

Performance media#performance-media

Paid acquisition channels measured by direct response — Google Ads, Meta, TikTok, affiliates. Distinct from brand media. The category where measurement gaps hit budget hardest.

The operator#the-operator

Our buyer profile. Defined by intolerance to broken data rather than by sector, geography, or revenue band.

Develops in chapter 02.

If a word we use feels imprecise, write to us — clarity in this document is part of the product. The standard reference glossary, with broader industry definitions, lives at /glossary.

Open ends here for now. Back to the index, or start over at chapter 01.

Was this chapter useful? Tell us →