Data and Tracking
Which channel really brought the revenue?
Google Ads today knows two models: last click and data-driven, with data-driven attribution used by default for most conversion actions g-attr. The four older models first click, linear, time decay and position-based have been retired by Google, with affected conversions switched to data-driven g-attr. The more important question is a different one, though: attribution says who gets credit for a conversion, not whether it would have happened without the channel.
What we do: We set one model with you and stick to it, separate attribution from impact, plan tests with control groups into the year and let ourselves be measured against an outside number; you know what a channel adds, not just who gets credit for a conversion.
By Robin Kardinal, Chief Operation Officer (COO)
Last updated on
What it is about and when it becomes relevant
What attribution can do
It distributes a conversion across the touchpoints before it. Google Ads uses data-driven attribution for this by default, distributing based on your account's data; last click is available as the alternative g-attr. The older models are gone g-attr.
What attribution cannot do
It does not answer what would have happened without the channel. A channel can look good in every model and still be doing nothing but collecting conversions that would have arrived anyway. Only a comparison against a group that did not see it can tell.
When it becomes relevant
As soon as several channels feed the same goals and budgets are shifted between them. Then attribution decides over money, and the question of how reliable it is becomes a business question.
Common mistakes and how you spot them
Budget shifted by last click
How you spot itChannels early in the journey lose budget, then total revenue drops, without the last-clicked channel having gotten any worse.
Every system with its own model
How you spot itAd account, analytics and business system report three different numbers, and in every meeting the convenient one is used.
Model switch mid-evaluation
How you spot itThe time series breaks on exactly the day the model was switched. Before and after can no longer be compared.
Not a single measurement without attribution
How you spot itThere is no test in which a channel was ever held out. So there is no statement about what it adds.
How we go about it
Pick one model and stay with it
We agree with you which model carries the steering, and document every switch with its date so time series stay readable.
Separate attribution from impact
Attribution steers the day-to-day. For the question of what a channel adds, we build tests with control groups.
Make tests plannable
Regional tests or deliberately holding out audiences need preparation and a runtime that matches the buying cycle. We plan them into the year instead of improvising.
Verify against an outside number
We ourselves insist on being judged against a number that does not come from our own report.
Further reading
This topic belongs to Tag Management.
Frequently asked questions
Which attribution models remain in Google Ads?
Last click and data-driven. For most conversion actions, data-driven attribution is the default g-attr.
Which models has Google retired?
First click, linear, time decay and position-based. Affected conversions were switched to data-driven g-attr.
What is data-driven attribution?
A model that distributes the value of a conversion across touchpoints based on your account's actual data g-attr, instead of following a fixed rule.
How do I measure whether a channel really brings additional revenue?
Through control groups instead of attribution: regional tests or deliberately holding out audiences. The setup costs preparation, but it is the only route to a reliable statement.
Is this exactly where you are stuck?
Write to us about where it is stuck. Robin Kardinal and the team will look at what is concretely behind it in your case, and tell you where the biggest lever sits.
Sources
Every statement about ad products comes from the official documentation of the respective platform, with the date we retrieved it. Products change; if something looks outdated to you, write to us and we will check it.
- g-attrGoogle Ads-Hilfe, Attributionsmodelleretrieved on 21.9.2026
Content and images on this page may have been created with AI support.