Beyond Last-Click Attribution
Choose an attribution model that best fits your needs
There are numerous types of attribution models. You’ll want to choose a model that best fits your needs.
A brand new hotel with a big budget might value introducing people to their brand, while a trip planning software company cares about every ad interaction on the journey. It depends on what you want to get out of Google Ads. You’ll want to choose a model that best fits your needs.
Whenever you start the process of picking a new attribution model, remember to test how that model matches your overall approach to Google Ads. You want to see if it drives more value or more conversions and then decide if you’ve made the right choice.
If you have enough conversions to qualify, consider using data-driven attribution (DDA)
Choosing the right attribution model is a big decision, and it’s sometimes best to let the numbers do the talking. That’s where data-driven attribution can step in. A data-driven approach takes the guesswork out of choosing a model.
If you have enough information to use data-driven attribution, it can paint the clearest possible picture of success in your account. Data-driven attribution was already available in Analytics 360, Attribution 360, and Google Marketing Platform.
Attribution works for all conversions from your website, including those imported from Google Analytics. Attribution does not apply to other conversions such as calls and app downloads at this time.
Data-driven attribution methodology
Sophisticated algorithms evaluate all the different paths in your account (both converting and non-converting) to determine which touchpoints are the most influential. Factors such as the number of ad interactions, the order of exposure and the creative assets used in each conversion path are all incorporated into results. Using a counterfactual approach, the algorithms contrast what actually happened with what could have happened to determine which ad clicks are most critical for a conversion.
There are multiple benefits of switching to data-driven attribution:
- Values all steps on the conversion path
- Works with automated bidding (as do the other attribution model options)
- Works even on very short conversion paths
- Quick and easy to implement
The choice of models is ultimately yours, and there are some cases where you may find a more optimal model for your account. But if you have enough data to qualify, plan on using DDA.
If you can’t use DDA, consider a rules-based attribution model
Data-driven attribution is powered by your account’s history, and if you don’t have enough traffic you might not be eligible to use it. That doesn’t mean that you need to continue using last-click attribution, though.
Like DDA, linear, time decay and position-based models all break up one conversion across each touchpoint. Splitting up a single conversion across all steps on the conversion path can give a clearer sense of a keyword’s value. And all models let you take advantage of automated bidding as well.
Picking a model should connect with your goals for your Google Ads account. Certain strategies tend to be growth-oriented, while others are more focused on efficiency.
|Last-click (DEFAULT)||Gives all credit for the conversion to the last-clicked keyword||Most conservative|
|First click||Gives all credit for the conversion to the first-clicked keyword||Most growth-oriented|
|Linear||Distributes the credit for the conversion equally across all clicks on the path||Moderate|
|Time decay||Gives more credit to clicks that happened closer in time to the conversion||Conservative|
|Position-based||Gives 40% of credit to both the first- and last-clicked keyword, with the remaining 20% spread out across the other clicks on the path||Growth-oriented|
|Data-driven||Gives credit to clicked keywords based on how imperative they were in the conversion process||Based on account’s performance|
The approach that you choose is going to determine the relationship between you, your Google Ads account, and your customer’s path to purchase. Attribution is about putting the performance of different keywords into the proper perspective. Performance that seemed typical under pre-existing last-click models could be very different when you evaluate those keywords with a new model.
Keywords earlier in the click path (often generic terms) usually behave differently than keywords later in the click path (like brand terms).
|Early influence keywords||Late influence keywords|
Types of keywords
("things to do in Tuscany")
("Mom and Pop's Tuscan Tours")
A more aggressive model, like first click, will shift performance stats to reward keywords earlier in the click path, while a more conservative model, like last click, will reward keywords that occur later in the click path. As you update the model you use, it might also make sense to re-evaluate the performance goals you have for your account. Many advertisers see better performance with lower costs-per-acquisition for keywords earlier in the click path once they move to an aggressive model, such as first click. If this is the case for you, consider increasing bids to get more volume at your previous CPA.
Here’s something to consider, especially in the days and weeks after you move away from a last-click model: your costs will likely remain the same but your conversions may show a small, temporary drop. That's because the conversion lag for non-last-click models tends to be higher than for last-click models. This could make it seem, for a brief period of time, that performance is getting worse. Things should quickly stabilize as your account adjusts to new ways of counting conversions, but know that slight performance drops are expected after changing models.
When reviewing performance data, remember to choose an appropriate history window. If you update your history window, you may see additional conversions that originated from clicks that fall outside of your current date range. While your cost data will remain the consistent (as it’s aligned with the date range you select), you may see more conversions appearing as you extend your history window.