Opening the black box of attention metrics 

Opinion

Greater clarity of attention metrics, be it through self-regulation or industry investigation, will unlock more creative thinking.

2023 has marked a shift from everyone talking about attention metrics to everyone adopting attention metrics. Suddenly, the attention market is no longer the domain of upstart ‘attentionators’. The big guns have arrived.

DoubleVerify’s Authentic Attention product, and IAS’s Quality Attention offering (which Lumen is happy to be a part of) mark a watershed in the industry: shit just got real.

But if it is really the dawn of the Attention Age, then the attention providers have some serious questions to answer. How are we defining attention? Where does your attention data come from? If we are using predictive models based on human datasets to estimate attention, what factors go into those models? If we are claiming superior business results, then who is marking our homework?

Radical transparency needed

The industry can respond to these questions in two ways: opacity or transparency.

The first approach is to say, “who cares how it works – it works!” — I don’t really know how planes fly, and yet I was still happy to get on one to go to Cannes and drink pink wine. Just as you don’t need a degree in aeronautics to step on a plane, perhaps we shouldn’t worry our pretty little heads about how attention tech works, and instead just enjoy the fact that it does.

The evidence from all the various attention companies is capacious and consistent: using this data helps brands avoid wasting ad dollars on unseen ads and helps direct investment to the hardest working media. The results really do speak for themselves.

But while I trust EasyJet to get me to Nice-Cote d’Azur without really knowing how they do it, I also trust that someone at EasyJet — and the people who regulate them — is fully informed. I don’t have to worry about it because someone else definitely is worrying about it.

And the advertising industry is right to worry. There can be few more troubling phrases in the English language than ‘Trust me, I work in adtech.’

And that’s why the second approach — one of radical transparency — is a far healthier option for the industry. Given the newness and, frankly, weirdness of attention predictions, getting straight answers to straight questions is vital.

This can be done by the attention providers, industry bodies, or, in the final analysis, by the buyers of these attention services themselves.

Podcast: Attention measurement is ‘once-in-a-lifetime chance to make a killing’

In terms of self-candour, Lumen’s methodology, data set, predictive model and link to results has just been investigated by PwC (Sam Tomlinson and I will be talking through the results at MADFest on Wednesday).

Lumen shared our data, predictive model and the results from three Blue Chip advertisers with the guys at PwC. They were able to understand our approach, recreate the model, and confirm that attention metrics predicted performance outcomes better than viewability alone. But they also found that more work needed to be done to prove the link between attention metrics and brand lift study results.

Happily, this correlation has now been well established, notably by the dentsu Attention Economy project and meta-analyses produced by Teads. But the fact that PwC were rigorous enough to raise this question speaks volumes about their integrity and the trust you can place in the rest of their report.

Straight answers to straight questions

But self-regulation is only one way of building trust. Not every company will be willing to let PwC root around their algorithms.

Industry regulation — or at least, investigation — is essential to build trust and confidence in the market. The IAB’s Attention Task Force, expertly led by Angie Eng in New York, will provide a comprehensive overview of approaches and methodologies, which will surely build up into a robust set of standards in the future.

But if you can’t wait that long, perhaps the best advice is “caveat emptor”. As a buyer of attention services, you have to ask the tough questions yourself. Here are some areas you might want to shed some light on:

• Where does your original attention data come from?
• Is there any human attention data involved at all?
• How does your attention prediction model work?
• What are the factors that go into the model?
• How does your model work across media?
• How does your model correlate with outcomes like brand lift or sales?
• How much better is your model at predicting sales than sheer viewability alone?

Getting straight answers to these straight questions is the minimum you should be able to expect from your partners.

It is also the only way we will make progress as an industry. Sure, we need clarity about methods and models will build trust and promote trial. But clarity will also unlock creativity.

The more planners and buyers know about how attention works, the more interesting and inventive their thinking. More ideas, more experiments — leading to better understanding and better results.

It’s time for the attention industry to open up the black boxes and let the sunlight in. Let a thousand flowers bloom!