cross channel attribution model at work
Back to Blog

The pitfalls of cross-channel attribution models (and how to dodge them)

  • LiveRamp
  • 3 min read

Welcome to the perennial challenge of measuring marketing impact.

As marketers, we have always strived to figure out which activities lead to revenue and which channels bear the strongest influence. The problem is, with an ever-increasing pool of touch points in a typical purchase journey, how do you make sure each one gets the credit it deserves?

The simplest approaches are still the most common, but these can still result in skewed perspectives of what’s working…

‘Last click’ attribution

With ‘last click’ attribution, whatever the user did just before buying gets the entire credit for the sale. However, this can easily become problematic. For example, let’s say you spend millions on a TV campaign. A prospective buyer watches your TV advert, but then types your brand name into the search bar to find your site and make their purchase. All the credit for that transaction goes to Search. Extrapolate this across your entire customer base and revenue streams, and you end up allocating too much of your budget to Search without increasing your TV spend.

‘First click’ attribution

As the name implies, ‘first click’ attribution simply credits the sale to whatever interaction first put a user on your radar. Say the buyer’s first interaction with your brand takes place when they search for something and your ad appears, which they click but don’t convert. Instead, after seeing several retargeting adverts in the following days, they eventually return to the site and buy the original product, seven days after first searching. Despite the time lag and all the interactions in between, the original ad is credited with the sale.

Clearly, first click attribution is just as misleading as its last click counterpart.

Connecting the attribution dots

As marketing teams recognise the distorting effects and pitfalls of these simple models, they have started to build better, more weighted models that start to connect the dots across their online assets. The e-Commerce system may be integrated with the CRM system, the ad-serving platform might integrate with the on-site product recommendation tool. With a persistent user portrait, you can start to understand customer preferences across these linked channels – as long as the user signs in and authenticates in each instance.

The result is a slightly better picture. But only slightly…

Enter the next pitfall… Although this model is starting to look sophisticated, it can still lull us into a false sense of what is working and what is not. The most glaring example of all is the case of offline behaviour. Despite the online revolution, over 90% of all retail purchases are still made offline, meaning it’s just as crucial to build a clear picture of offline attribution as online.

So what’s the answer?

To see attribution success marketers need a complete, omnichannel picture – encompassing both online and offline interaction. Data onboarding and identity resolution make it possible to connect data across channels and platforms, allowing you to build an holistic attribution model and really understand what’s working.

Take a look at this eBook to understand more about how identity resolution helps you connect the dots across your entire marketing ecosystem.