What is data-driven attribution?
Data-driven attribution uses machine learning to estimate how much each touchpoint contributed to a conversion. Here is how Google's model works, how much data it needs, a worked example, and why it is not the same as incrementality.
Data-driven attribution (DDA) is a model that uses machine learning on your own conversion data to estimate how much each touchpoint contributed to a conversion. Instead of fixed rules, it compares the paths of people who converted with the paths of people who did not, and gives more credit to the touches that changed the odds of converting.
How does data-driven attribution work?
Google's GA4 attribution guide says the model uses "a counterfactual approach": it contrasts what happened with what could have occurred without a given touch. The Google Ads data-driven attribution page says it is "the default attribution model for most conversion actions" and analyzes interactions on Search (including Shopping), YouTube, Display and Demand Gen ads.
| Rules-based models | Data-driven | |
|---|---|---|
| How credit is set | Fixed weights you choose | Learned from your paths |
| Transparency | Fully explainable | Hard to audit |
| Data needed | Any volume | Works better with more |
| Scope in an ad platform | That platform's ads | That platform's ads |
Data-driven attribution example
An illustrative UK retailer records £120,000 of Google Ads conversion value in a month. Last click credits £90,000 to brand search and £30,000 to non-brand search and YouTube. If paths with a YouTube ad convert more often than similar paths without one, data-driven might move £20,000 of brand search credit to YouTube and non-brand search. The real split depends entirely on your data.
Limits of data-driven attribution
- One platform's view: Google's model credits Google ads, Meta's credits Meta ads. Neither sees the other's touches.
- Volume: Google says all conversion actions are eligible, but recommends at least 200 conversions and 2,000 ad interactions within 30 days.
- Not proof of cause: it finds patterns in paths. Only an incrementality test shows what a channel caused.
Related terms
Data-driven is one kind of attribution model. Compare it with rules-based multi-touch attribution such as position-based.
How Grower helps
Grower gives you a cross-channel view next to each platform's own model. Its attribution engine de-duplicates conversions from every connected channel and app into one count, then applies the model you pick: last click, first click, linear, time decay, position-based or custom weights.
Frequently asked questions
Is data-driven attribution the default in Google Ads?
Yes, for most conversion actions. Google's help page calls it the default attribution model for most conversion actions. When Google retired first click, linear, time decay and position-based models in 2023, it moved conversion actions still using them to data-driven. Last click remains available if you prefer a rules-based model inside Google Ads.
How many conversions do you need for data-driven attribution?
Google says every conversion action is eligible regardless of volume, but it recommends at least 200 conversions and 2,000 ad interactions in supported networks within a 30-day period. The model still runs with less data, and Google notes that it performs better with more. Low-volume accounts should treat the results with extra caution.
Is data-driven attribution the same as incrementality?
No. Data-driven attribution redistributes credit for conversions that already happened, based on patterns in conversion paths. Incrementality measures how many conversions a campaign caused, by comparing a group that saw the ads with a control group that did not. A channel can earn plenty of data-driven credit and still show little incremental lift.
Put each platform's data-driven numbers next to one de-duplicated, cross-channel view: book a Grower demo, or read marketing attribution models explained.