Insights · Measurement

Marketing attribution and analytics

There is no single source of truth anymore. Modern measurement is a layered stack — attribution for in-flight signal, marketing mix modeling for budget, incrementality for causal proof — built on consented first-party data.

The measurement reset

There is no single source of truth in marketing measurement anymore. As tracking signal degrades, the industry has moved to a layered stack: attribution for in-flight, tactical signal; marketing mix modeling for strategic budget allocation; and incrementality testing for causal proof. One honest correction up front: Google did not remove third-party cookies from Chrome — it reversed that plan — so the shift is driven by privacy and signal loss generally, not a single cookie deadline.

Attribution models: last-click vs data-driven

Google Analytics 4 uses data-driven attribution (DDA) as its default and recommended model, distributing conversion credit across touchpoints with machine learning instead of fixed rules; GA4 removed four older rule-based models in November 2023. One trap to know: when a property lacks sufficient conversion volume, GA4 silently falls back to a last-click-style model — so two accounts can be measuring on different logic without anyone noticing.

The marketing mix modeling revival

Marketing mix modeling (MMM) — a statistical, privacy-durable method that needs no user-level tracking — is back in force. Roughly 47% of US marketers plan to increase MMM investment over the coming year (eMarketer/TransUnion). The barrier to entry has fallen too: Google released its open-source MMM library Meridian in January 2025, alongside Meta's Robyn and PyMC-Marketing. MMM answers the strategic question attribution can't: how to allocate budget across channels.

Incrementality: proving causation

Attribution shows correlation; incrementality testing proves causation. By comparing an exposed (test) group with an unexposed (control) group — commonly through geo experiments or audience holdouts — you measure the true lift a channel delivers, not just the conversions it happened to touch. It is the antidote to over-crediting channels that would have converted anyway.

First-party data and server-side infrastructure

The durable foundation is consented first-party data. Server-side tagging gives marketers a first-party collection layer with better PII control, longer cookie lifespans and cleaner governance; techniques like Enhanced Conversions use hashed first-party data to recover otherwise-lost conversions; and data clean rooms have matured into mainstream, privacy-safe collaboration infrastructure, especially in retail media. The direction is unmistakable: own your data, measure responsibly.

Privacy and consent: GDPR and Consent Mode v2

Since March 2024, Google requires advertisers serving ads in the EEA to pass user-consent signals via Consent Mode v2. Important nuance for legal-safe copy: Consent Mode is Google's policy requirement, not itself a GDPR mandate — it is one mechanism for respecting consent, not a substitute for a lawful consent basis and a proper consent management platform. Treat privacy as a design constraint that, handled well, becomes a measurement advantage.

For your brand

Where Rabbit fits

As signal loss pushes value toward owned, first-party foundations, organic discovery becomes the most durable channel of all. Our own multilingual network is found without ad spend, and a multilingual website with international SEO can give your brand the same owned, lasting reach. We would welcome a quiet conversation whenever it suits.

See the live case study →

A small studio, network-grade work

Proof, not promises.

Most agencies describe their craft. We prefer to show it. For a traditional craft business we built a multilingual knowledge network that is found organically around the world — no ad budget behind it, just structure, language and patience.

See it live: rohrgeruestbau.de and special-scaffolding.com (a 17-language scaffolding knowledge net). If you would like that kind of quiet, lasting visibility for your brand, Rabbit Marketing can help.

Start a conversation →

FAQ

Questions, answered

What is the difference between last-click and data-driven attribution in GA4?

Last-click gives all credit to the final touchpoint. Data-driven attribution (GA4’s default) uses machine learning to distribute credit across touchpoints based on their measured contribution. Note that GA4 silently falls back to last-click when conversion volume is too low.

Why are marketers returning to marketing mix modeling?

Because MMM is privacy-durable — it needs no user-level tracking — so it survives signal loss. Around 47% of US marketers plan to increase MMM investment, helped by free open-source libraries like Google’s Meridian, Meta’s Robyn and PyMC-Marketing.

How does incrementality testing differ from attribution?

Attribution assigns credit to touchpoints that were involved (correlation). Incrementality testing compares an exposed group with a control group — via geo experiments or holdouts — to prove the causal lift a channel actually delivered.

Is Google Consent Mode v2 legally required for GDPR?

Not exactly. Consent Mode v2 is Google’s own policy requirement for serving ads in the EEA since March 2024. It is a mechanism for passing consent signals, not a substitute for a lawful GDPR consent basis and a proper consent management platform.

Did Google deprecate third-party cookies in Chrome?

No. Google reversed its deprecation plan; third-party cookies remain in Chrome. Measurement is still shifting toward first-party data and privacy-durable methods because signal loss continues for other reasons.

Sources

Where this comes from

Research date: June 2026. Figures are industry estimates where indicated; they are illustrative, not advice, and not a promise of results. Company and brand names are used for editorial reference only and imply no affiliation with Rabbit-Marketing OÜ.

Related insights

Keep reading

All insights →

Measuring what matters?

Start a conversation →