Credit where credit is due: Tips for using attribution models

The rise of new direct-response marketing channels, such as performance display, has put pressure on advertisers to find more effective ways to measure the impact of these channels on sales. To better track advertising spend, marketers have turned to attribution models to find out which part of their marketing efforts, if any, should be credited for a specific sale.

There are several types of attribution models. Models such as last click, first click and first touchpoint are single-touchpoint models, where one touchpoint gets full credit for the sale. On the other hand, models such as a linear, U-shaped and time decay are multi-touchpoint models, in which credit for the sale is spread over several marketing channels.

Types of attribution models

Why are we still using last click?

According to insights derived from Criteo’s day-to-day business with more than 5,000 clients globally, and from our research on the role of marketing attribution in today’s digital marketing landscape, close to 80 per cent of advertisers still use last click as their primary attribution model, even when they consider it insufficient.

This shows that many organisational constraints and technical challenges that stand in the way of widespread adoption of newer multi-touchpoint models. The top three reasons why advertisers are not yet taking the leap are:

• Switching attribution models requires new technology platforms and skills
• Switching attribution models are highly disruptive
• Switching attribution models implies new relationships with external vendors

Marketing attribution: no silver bullet to managing complexity

However, advertisers are becoming more sophisticated in their approach to measuring their return on investment, with the emergence of new marketing attribution models and methods.

Advertisers may not have fully switched to more sophisticated attribution models yet, but they are moving away from relying solely on last click. They are injecting intelligence into their marketing strategy by resorting to A/B testing or by using new attribution models as a second view to last click.

As we move towards a multi-device world, we must be able to track users across their various gadgets. This is one of the most important challenges faced by the marketing industry today as there will not be a one-size fits all solution capable of tackling the complexity of multi-channel attribution.

Three key lessons for advertisers

Whether advertisers use new-generation methods or stick to the tried-and-true last click model, here are three key lessons that will help manage the increasing complexity of multi-channel attribution:

Lesson 1: Challenge whatever your model tells you

No matter how cutting edge it is, a model remains a simplified version of reality used for practical purposes. But simple shouldn’t mean simplistic. In particular, make sure that you get a clear view on what touchpoints your model is missing. If you don’t, you might miss sales.

Lesson 2: Stop assuming, start testing

Respondents said it loud and clear, and we fully agree: the best way to demonstrate causality is by testing. If you doubt the value of a marketing channel, or if you would like to know what doubling your spend on it would really do to your sales, the best way to find out is to set up a test.

Lesson 3: Focus on touchpoints that truly have influence on the purchasing decision

Many touchpoints serve only to help a user navigate from one place to another; they do not actually influence any buying decision. Removing such “navigational” touchpoints from your model is less game changing than switching to a new attribution model, but it has a big impact on your results—and on your ability to generate more sales.

As a marketer, your job is to know how your customers get to your cash register. So strategically deciding what your advertising spend should be. And the good news is no machine is going to take this away from you anytime soon.

Yuko Saito is managing director, Southeast Asia, at Criteo