Model Marketer: how to choose the right attribution model for you

24th May 2018 - 6 minutes read  Data & Analytics
James Wolman - Data Scientist

There’s no point beating about the bush – nowadays, we live in touch-point central.

Gone is the era of last click attribution. The customer journey has, over the years, become more and more complicated, and Salesforce now reckons it takes six to eight touches to establish a fertile sales lead. Furthermore, 60% of the sales cycle is believed to be over before a buyer talks to your salesperson.

Plus, as the customer journey extends in one direction, it’s also spreading itself across multiple platforms. Most digital marketers are deploying a range of content, running several campaigns and targeting potential buyers from numerous angles. So measuring the ROI on your various channels becomes a bigger and trickier task. How much of a contribution are your SEO efforts making? What kind of impact did your Twitter ad click have? What about your AdWords retargeting campaign? And if more conversions are coming from PPC, should you bin your social media efforts altogether?

It can feel like a scary (and apparently never-ending) set of questions, but luckily for all of us, there are answers out there. The answers aren’t too hard to find, either, but if you want access you’ll have to embrace the multi-touch attribution marketing model – it’s this approach that allows you to manage, optimise and scale your campaign performance.

Allocating accurate value to each of your marketing touchpoints is crucial for monitoring and adjusting your ROI on each channel, giving you the chance to make your spending as efficient as possible.

If you’re reluctant to trust blindly in models which might incorrectly attribute credit and consequently encourage you to wind down channels which are in actual fact high performing, your concerns are fair. There are plenty of models out there, and the pitfalls could potentially be numerous.

The most frequently used models are as follows:

  • First/Last Click, which assigns 100% of the credit to the touchpoint in action just before conversion. It’s conveniently simple, but not necessarily accurate.
  • Linear, which splits attribution evenly between all the touchpoints that play a role in conversion.
  • Time Decay, which credits touchpoints according to how close in time they are to conversion.
  • Position Based, which gives 40% of the credit to the first and last touchpoints involved in conversion, and then distributes attribution between the touchpoints that were active in between.

There’s also a new, more sophisticated data-driven attribution model available, which is Google’s recommendation. It’s more complicated to use, but rather than working on a rules-based system, data-driven models use machine learning to ramp up the accuracy with which they allocate credit.

Before the complexity of the data-driven model becomes standard-issue for all, though, digital marketers have a choice to make. Luckily, there’s a reliable set of tools and tricks to help you avoid the rookie errors and select the perfect model for your business.

Here’s our take on the key factors to consider…

  1. Map your customer’s journey

Create a detailed and comprehensive sketch of touchpoints and channels with potential buyers. Where do they interact with you during the buying cycle? Are there touchpoints that you’d to develop or add? Do you understand customers’ responses to each point of contact?

  1. See the big picture

The model you choose should provide a satisfactory answer to the executive question: if you reign in spending or stop it all together, will revenue suffer and will individual components of the operation experience a drop in performance, too?

  1. Look at your leads

Work out the rates at which leads become paying customers and/or active users.

  1. What’s your goal?

It’s critical that while choosing a model, you’re led by your fundamental objective. Be it branding, limited-time promotion or anything else, different kinds of campaigns can suit different models, so don’t forget your overarching aim.

  1. Review and revise

Unfortunately, the model selection process is never fully complete. It’s important that you commit to regular testing and iterating. You might need to optimise your model, or further down the line switch it up completely.

It’s a good idea to start with a simple model, and experts encourage marketers to give their first selection a decent innings. But if you take your time choosing, monitor progress carefully, and take decisive action if eventually, you think the changes should be rung; you shouldn’t go too far wrong.