February 21, 2020
Five Ways to Do Incrementality Testing with Digital Ads
By changing your AdWords attribution model strategy from last-click conversion attribution to position-based attribution, you gain more insight into the customer journey, which allows you to see what top-of-funnel and mobile clicks are driving later conversions. This data will help you channel your marketing funds into the campaigns that drive new customers, ultimately helping increase total conversions across your account. These additional insights, and their ability to help you ramp account growth, make position-based attribution a must-have for nearly all advertisers. Here are some details on how position-based attribution works, and what insights you should expect.
If you’ve only been evaluating your paid search campaigns using last click attribution, you are likely undervaluing non-brand queries and campaigns. Last click attribution gives the full conversion credit to the last click before a user converts, which means branded queries will almost always see a large percentage of conversions. Position-based attribution, on the other hand, splits conversion value among all clicks in a given path to conversion. To be more precise, position-based attribution gives 40% of the conversion value to the first click, 40% to the last click, and divides the remaining 20% to the clicks in between.
For example, if you’re looking to buy a pair of binoculars for birding, you might start with a broad top-of-funnel search, “binoculars.” Then you might start looking up more research-focused keywords, such as “binoculars for birding reviews.” After doing your research, you find that Brenda’s Birding Supplies would be the perfect place to buy from, so you type in “brenda’s birding supplies” and make your purchase.
Using last click attribution in the scenario above, 100% of the conversion credit would be given to the brand keyword, even though there were other searches that led to the conversion. Using position-based modeling, 0.4 conversions would be given to binoculars, 0.2 would be given to binoculars for birding reviews, and the remaining 0.4 would be given to the branded query (adding up to 1 total conversion). Position-based modeling gives you different and potentially more valuable insights into your campaign, as you’d be able to see and take action based on top-of-funnel keyword data.
While Google Analytics and third-party platforms such as Kenshoo have previously allowed for position-based attribution and modeling, Google AdWords has historically only used last click attribution. Just within the last year, however, Google rolled out position-based attribution in AdWords. We have run position-based attribution on Kenshoo for several years for most accounts, and we love the insights it gives us on high-funnel account opportunities. More recently, we ran some tests on AdWords to measure the influence of this new feature on performance and account strategy.
The first and most important insight we noticed from using position-based attribution in AdWords was that non-brand conversion values increased significantly – which was no surprise. On average, non-brand conversion totals increased 11%, with individual campaign increases varying between 5% and 20% as we transitioned away from last click attribution. Because of position-based conversion data, we had more non-brand data to work with. That additional data allows us to make more bid adjustments and ad copy changes than if we just rely on AdWords conversion data.
More importantly, the additional non-brand conversion data can often point to new areas and directions to grow the account based on the additional non-brand insights. If just last click attribution was in place, we would have missed out on optimizations and new areas of growth among our non-brand keywords.
Another area where we saw opportunities for account growth was mobile search. The change to position-based attribution drove small increase in our mobile conversion numbers, helping us to more accurately evaluate the impact of mobile search. Because mobile is most frequently a first-touch marketing channel, position-based attribution will help you identify customers who begin the purchase process on a mobile device but only convert after finishing on a desktop device.
When you transition to position-based attribution in AdWords, you will likely see significantly more backfill of conversion data than you would see with last click attribution. This is because many conversion paths started days or weeks ago and are only converting recently. In this case, the value of the first and middle interactions are attributed retroactively, in addition to the last interactions. With last click attribution, only the last interactions see backfill.
For one account, we saw a conversion lift of 3% for December one week after the month had ended. During week two of January, we saw another 0.4% total increase in conversions. In week three of January, we saw an additional 0.3% increase in December’s total conversion volume.
In another example, we looked back at data for a period two months prior and saw a conversion lift of 7% from what was originally reported at the end of the month.
Your cookie window will also impact reporting, so you should make sure to look back at longer date ranges after you apply position-based attribution. A longer cookie window will generally result in more backfill as customers continue to convert multiple months after interacting with your ads. You can expect the data to continue backfilling across the length of your cookie window, which will impact optimizations such as mobile bid adjustments and dayparting analyses. To ensure you capture the full length of the conversion cycle, make sure to review date ranges that encompass your entire cookie window.
One surprising finding was a 6% conversion decrease in remarketing lists for search ads (RLSA) campaigns when using position-based attribution, even though those campaigns are usually strong performers. This made sense upon closer evaluation. Remarketing ad clicks are usually the final interactions before a conversion, since we’re targeting people who have previously expressed interest in our product or service. However, with position-based attribution, additional value is given to top-of-funnel, non-brand interactions that searchers had prior to clicking on a remarketing ad. With position-based attribution, expect to see more conversion credit shifted from RLSA to non-brand Search campaigns.
Ultimately, it’s up to you and your paid search goals to determine how you want to attribute conversion credit in your campaigns, but we strongly recommend moving away from last click attribution. If you are looking for new areas of growth through non-brand, then position-based attribution could be the right fit for you.