Paying (for) Attention: Using Attention Data to Optimise Media Spend


Paying (for) attention

A First Aid kit for your brand - May 2018 event

At our May 2018 event, Mike Follett (Lumen) was joined on stage by Patrick Smith (British Gas) to discuss how British Gas  is using attention data to optimise media buying and enhance creative performance. Lumen have provided additional insights to accompany their talk.


How British Gas is using attention data to optimise media spend and enhance creative development

We all have an intuition that there is a difference between having an ‘opportunity to see’ something and actually looking at it. Just because people CAN see something does not mean that they WILL see it. This is especially important when it comes to advertising because ads can’t deliver a message unless consumers engage with them visually. 

But what is the difference between ‘OTS’ and actual attention? How many viewable ads do in fact get viewed?

For the last two years, Lumen Research have been running the world’s largest eye tracking panel to try and answer exactly this question. In total 766 households have been recruited and equipped with laptop-mounted eye tracking cameras that passively monitor what people notice when they go online - and what they ignore!

The learnings from the panel have important implications for marketers:

  • Only 22% of viewable digital advertising actually gets looked at – meaning that 78% of ads get entirely ignored. This compares to newspaper ads, where 76% of viewable ads will get viewed. Advertisers have to earn the attention of their audiences, and cannot assume that everyone is watching.
  • Dwell times with digital ads are much shorter than most marketers imagine. The average digital ad will get looked at for 1.2 seconds. The average newspaper ad will get 2.2 seconds of attention. This means that attention patterns to digital advertising has much more in common with out-of-home than DM.

The key driver of attention are:

  • Viewable time: the longer an ad is available to be seen, the more chance it has to be seen. Understanding the minimum threshold for viewable time is crucial in optimising media spend.
  • Format and location of ads: larger ads in prominent positions get more attention than smaller ads in secluded positions. Understanding how much to pay for ad size and position can help deliver greater efficiencies.
  • Site: context makes a different. The same ad, viewable for the same amount of time, will get more attention on some sites rather than others. Choosing the right sites for your ads to appear on can have enormous effects on the efficiency of your media spend.

Working with British Gas, we have been able to apply these learnings to deliver immediate business impact.

  • Historical review: we applied our learnings to historical data from British Gas digital campaigns. We were able to establish that ads that are likely to get high levels of visual attention were significantly more likely to drive sales in the short term.
  • Predictive model: using data from the historical review, we have been able to identify the sites and formats that are most likely to drive sales for British Gas. We have been working with their media agency to use this insight to inform buying decisions, which has dramatically reduced the amount of money spent on ads that are unlikely to ever get seen.
  • Enhancing creative: in addition to the insight into media, we have conducted a series of cheap and quick creative tests to identify the creative drivers that lead to increased attention. As a result of these changes, attention to British Gas ads has dramatically increased and click through rate has improved by 25%

For years, advertisers have talked about ‘buying eyeballs’. This used to be a colourful analogy, but is now an affordable and scalable reality.



Mike Follett, Managing Director, Lumen


Exclusively for Members

As always, our seminars are exclusive to AURA members. 

Following the event, additional material will be available on the AURA website. Follow this link to see additional event content.