Attribution and AdWords

Attribution models are designed to help marketers understand how their marketing efforts — in this case, search ads — played a role in getting a user to convert. How do you know which keywords and ads contributed to a conversion?

This question is increasingly important as users research and buy across multiple devices and channels.

It’s important to note that attribution models in AdWords only account for ad clicks and engagements on Search and Shopping ads. No other touch points on the way to conversion are factored into the models. Google has an enterprise product, Attribution 360, that brings conversion data from other channels in for cross-channel modeling and is rolling out a free version of this product. The models are the same across all properties (Fig. 11).

Data-driven attribution is Google’s machine learning-powered model and requires campaigns have a minimum number of clicks and conversions. Generally, an account must have at least 15,000 search clicks, and a conversion action has to accrue at least 600 conversions within 30 days for the data-driven modeling to begin running. Advertisers only see the data-driven option available in AdWords when it has enough data.

Fig. 11 – Attribution models available in Google

What’s the problem with last-click attribution?

Last-click has been the default attribution model for years, and Google is likely to finally kill it off within a year. Why? It simply allocates 100 percent of the credit for a conversion to the last activity, in this case, the last AdWords ad a user clicked or engaged with before converting.

Why is this a problem? Typically, users don’t click one ad and convert. Instead there may be two, three, four or more ad interactions — across multiple devices — before a conversion occurs.

Consider the following scenario: A user searches for “tax preparation software.” She clicks on an ad from LoopholeTaxPrep.com (not a real site.) She then goes back to search to comparison-shop. She then searches for “online tax software reviews” and clicks on an ad that leads to LoopholeTaxPrep.com’s reviews page. A couple of days later, she’s ready to buy, searches the brand name Loophole Tax Prep, clicks on the brand’s ad and converts.

There have been three ad clicks to the site throughout the process. In a last-click attribution model, the brand ad would get all of the credit. The non-brand ads would get none, and their contributions to the conversion would be invisible to the advertiser. In AdWords, the non-brand ads will have spent money but show no conversion outcomes.

The advertiser could end up overvaluing the brand ad and undervaluing the higher-funnel non-brand ads, then lower bids or cut budget on those non-brand campaigns. This is why there has been a significant push to move away from using last-click attribution.

Read more of The Search Engine Land Guide to PPC: