Google’s data-driven attribution model: Moving away from last click

By Maria Bestard | 07 Jun 2017

Most advertisers measure the success of their online advertising on a "last click" basis, meaning that they give all the credit to the last ad or keyword a user clicked. Last month, Google announced it had released a data-driven attribution model to AdWords as a public beta, which was already available in Google Analytics 360 and DoubleClick. The default attribution model in AdWords is based on last click data, which ultimately gives all credit to the last keyword searched by a user. This has always been a bone of contention when it comes to accuracy as, typically, before making a purchase, users tend to make several searches and also click several ads. Changing our attribution model to a data-driven one grants the opportunity for all keywords to get a portion of credit if they've contributed along the conversion path.

This new data-driven model uses available data to determine how each keyword will impact a user's likelihood to convert, making each model specific to each advertiser.

To assign the right credit weighting to each keyword, Google compares all the click paths of users who convert vs the path of users who don't. The model then identifies patterns related to clicks that lead to conversions and attributes them more credit than to the ones which didn't.

The actual calculation related to conversion credit for each touchpoint relies on comparing all the different permutations of touchpoints, and then normalising across them. This means that the data-driven attribution algorithm considers the order in which each touchpoint occurs and assigns different credit weightings for different path positions.

The benefits of moving to this attribution model are that it:

-        Values early influencer keywords appropriately

-        Avoids underestimating generics or shopping campaigns

-        Gives more accurate bidding optimisation, taking into consideration all campaigns/ads/keywords that participate in the conversion path

-        Drives stronger results and helping to reach business goals

Not all advertisers will have this model available to them, as it requires a certain amount of data to create a precise model for how your conversions should be attributed. To qualify, an account must have at least 15,000 clicks and 600 conversions within 30 days. Once Google has collected sufficient data for the model across 30 consecutive days, the data is then available to see in AdWords.

If traffic drops below 10,000 clicks or 400 conversions within 30 days, Google sends an alert to warn advertisers, and if a low data level persists for 30 days, the attribution model will be automatically switched back to the linear model, which distributes the credit for the conversion equally across all clicks on the path.