Every other attribution model in Google Ads makes a rule. Data-driven attribution (DDA) skips the rules entirely and lets a machine learning model figure out which touchpoints actually drove conversions.
That sounds like a clear upgrade. But understanding how it works — and where it falls short — matters before you flip the switch on your campaigns.
What is data-driven attribution?
Data-driven attribution is a model that uses your own account’s historical conversion data to assign credit to touchpoints in the customer journey. Instead of applying a fixed rule (like “split credit equally” or “give it all to the last click”), DDA analyzes the actual paths that led to conversions versus paths that didn’t, then weights each touchpoint based on its observed impact.
It’s part of the broader family of attribution models — but it’s the only one in Google Ads that isn’t rules-based.
How the data-driven attribution model works
Google’s DDA uses a counterfactual approach. For every path a user took, the model asks: what would have happened if this touchpoint hadn’t been there? If removing a specific ad interaction would have significantly reduced conversion probability, that touchpoint gets more credit. If removing it barely changes the probability, it gets less.
The model considers signals like:
- Which keywords and ads appeared in converting vs. non-converting paths
- Where in the sequence each touchpoint occurred
- How much time passed between interactions
- Device, match type, and ad format
The result: a set of credit weights specific to your account, recalculated continuously as new data comes in.
Account requirements
DDA requires enough conversion data to train reliably. Google’s thresholds are:
- At least 300 conversions in the past 30 days for the conversion action
- At least 3,000 ad interactions in the path to conversion
If your account falls below these thresholds, Google will either fall back to another model or gray out the DDA option. Low-volume advertisers — most local service businesses — often can’t meet the minimums.
Data-driven attribution vs. rules-based models
Every other Google Ads model works from a fixed rule, regardless of your actual data:
- Last click: 100% credit to the final ad before conversion
- First click: 100% credit to the first ad interaction
- Linear: equal credit split across all touchpoints
- Time decay: more credit to touchpoints closer to the conversion
- Position-based: 40% each to first and last, 20% split across the middle
These rules are easy to understand and consistent. They don’t require any minimum data volume. The tradeoff: they apply the same logic to every account, which means they don’t reflect how your specific customers actually behave.
Multi-touch attribution as a category includes both rules-based approaches (linear, time decay, position-based) and data-driven. DDA is the most sophisticated version of multi-touch.
The limitations of data-driven attribution
It’s a black box
You can see the credit weights in Google Ads reports, but you can’t see the logic that produced them. If DDA assigns 60% of credit to a broad match keyword that you wouldn’t have expected to matter, you have to take Google’s word for it. For advertisers who want to audit or explain their attribution, that opacity is a real problem.
It only covers Google’s own touchpoints
Google data-driven attribution can see Google Ads clicks, Google Display impressions, and YouTube views. It cannot see what happened on Meta, email, organic search, or any other channel. If a customer clicked a Facebook ad three days before converting on a Google search, DDA treats the Google click as if it happened in a vacuum.
This is the central limitation of any platform-native attribution model: it only accounts for what the platform can see.
It doesn’t connect to revenue
DDA tells you which touchpoints drove conversions — form fills, calls, purchases. It doesn’t tell you which touchpoints drove revenue. A campaign that drove 50 low-value conversions might score better than one that drove 10 high-value clients. Unless you pass actual revenue values back to Google Ads, the model optimizes for volume, not quality.
What data-driven attribution means for WordPress service businesses
If you’re running Google Ads for a service business — roofing, senior living, medical practice, home services — you’re likely below the 300-conversion threshold for most months. That means DDA isn’t available to you in any practical sense.
More importantly, DDA doesn’t solve the core problem that service businesses face: connecting a Google click to a closed deal. A roofing company doesn’t want to know which ad drove the most form fills. They want to know which ad drove the most signed contracts.
That connection — from ad click to sale — doesn’t happen inside Google Ads. It happens when you close the lead in your CRM and pass that outcome back to your attribution system. WordPress attribution tools built around offline conversion tracking make this possible without moving your clients off their existing stack.
In other words: data-driven attribution is a useful model for understanding Google’s view of your funnel. But it’s not a substitute for tracking what actually happens after the lead comes in.
Should you use data-driven attribution?
If you have the data volume to support it, yes — DDA is generally better than last-click because it accounts for the full path to conversion rather than giving all credit to the final step. Google also uses DDA as its default model in Smart Bidding, so opting in puts your bid optimization on the same foundation as your attribution.
If you don’t have the volume (most local and service advertisers), last-click or position-based are pragmatic defaults. The bigger leverage comes from improving what you’re measuring, not which model you use to weight it.
The models available in Google Ads — DDA included — only touch the top of the funnel. The sale happens at the bottom. Knowing which channel actually drove revenue requires tracking that extends past the conversion event and into your CRM or deal record.