Google Ads moves to algorithmic model for determining conversion

Google is moving away from the most popular ad conversion model. Google Ads will use data-driven attribution by default in the future instead of last-click attribution. This means that the last click is no longer leading to determine the effect of an advertisement.

Google will implement the changes to the advertising platform Ads. There, a last-click attribution model was used as standard. That model looks at where a customer comes from at the last minute before making a purchase. That spot, the ‘last click’, then gets the full credit for a purchase. That model has always worked well for internet advertisements, because an advertiser can find out exactly from which website or search engine request a customer comes. The model has become obsolete in recent years. That last click isn’t always the best indication of where a buyer is coming from. For example, they may have heard of a product in several places and only make a purchase after several visits.

Google Ads is therefore switching to so-called ‘data-driven attribution’. The company uses unspecified machine learning to determine which different data points contributed to a conversion to a purchase. This can also be other websites, for example, or previous visits to a website. Google claims that this is more beneficial for advertisers. With data-driven attribution, advertisers would see more conversions at a lower cost.

Data-driven attribution isn’t entirely new to Google Ads. Advertisers could already choose this as an option, in addition to a last-click and a first-click model, a linear model and models that look at the length of time between an impression and a purchase or the position of an ad. The data-driven model was not available to everyone, because large amounts of data are needed to use it optimally. Small businesses, in particular, did not earn enough from advertising to make that model attractive. Google says it is now removing the limit. The company can do this because the system has become smarter in recent years and therefore requires less data.

Although the data-driven model will become the standard, the other models will still be available as an opt-in method. Data-driven attribution will become the default model for all ads in the search engine, Google Shopping, and YouTube in October.

Cookieless

Google says the new way of attribution is more privacy-friendly. “In the changing privacy landscape, advertisers need new metrics to achieve their goals,” the company says. The model does not use third-party cookies, but looks at different data points. As an example, Google cites the size of ads and the time between the moment a user sees an ad and eventually makes a purchase. In this way, for example, fingerprinting of a user can also be prevented.

The switch to the new advertising model is part of Google’s plans to make the Chrome browser tracking-free in the long term. However, data-driven attribution is not covered by the plans for the controversial Privacy Sandbox.