Cumulative Attention: A new way to conceptualise effective frequency

Opinion

Instead of counting the number of discrete ads served to a consumer, we should instead measure the cumulative attention they give to a campaign, Lumen’s Mike Follett writes.

We have long known that repetition is required for memory formation. In the Dissoi Logoi, an Ancient Greek training manual for sophists and rhetoricians (i.e., communications professionals like you and me), we find three pieces of advice on how to retain information. 

The first lesson is music to the ears of an attention researcher like me: If you pay attention, your judgment will better perceive what’s going through your mind.

But the second piece of advice is no less interesting: Repeat again what you hear; for by often hearing and saying the same things, what you have learned comes complete into your memory 

In past times, repetitive learning approaches were beaten into medieval schoolchildren, quantified by Hermann Ebbinghaus, humanised by Jean Piaget and set free by Maria Montessori. But whether the learning sessions are spaced or massed, self-directed or teacher-directed, audio or visual, they are always assumed to be multiple. 

This is not a message that media planners want to hear. Spending money to serve an ad to the same person time and again sounds like a waste of precious resources, which could be spent reaching new people. 

Repetition – otherwise known as frequency – is a curse to be endured, minimised or even exorcised.

The title of Herb Krugman’s 1972 paper that gave us his famous ‘rule of three’ was Why three exposures may be enough. It conceptualises frequency as something to be reduced to a bare minimum. By November 2001, Erwin Ephron and Melissa Heath had gone even further, entitling their Admap article ‘Once may not be enough, but it is the best we can do’. 

This isn’t just wishful thinking. Numerous studies have shown that in advertising, the first cut is the deepest and that subsequent exposures yield diminishing returns.

The rise of ad targeting technologies has given all of us the experience of ‘ad bombardment’: seeing the same CTV ad five or 10 times in the course of an evening, or endlessly in your feed. Excess frequency can lead to ad wear-out or banner blindness for a brand, and reduce trust in advertising in general. 

So we reach an impasse

For educationalists, repetition is indispensable; for media planners it is indefensible. 

And it is here that the horsetrading begins: should the optimal OTS be one or three or seven? Is this the same in all media? Is the Rule of Three in TV a good enough rule of thumb for Digital?

Perhaps there is another way of tackling the problem. Instead of counting the number of discrete ads served to a consumer, we should instead measure the cumulative attention they give to a campaign – whether it’s across one, three or 10 ads.

This is the provocative contention of some new attention research from Amazon Ads, Cint and Lumen. We have conducted a meta-analysis of over 270 brand lift studies conducted amongst 191,000 respondents in the US, Brazil and EU5 for a wide range of advertiser categories. 

For each respondent in the study, Cint collected data on how many ads they saw and what types they were: Amazon Prime, Twitch, and third-party digital display.

Lumen then applied average attention scores to this exposure data to estimate not only how much attention each impression received, but also the total amount of attention that a particular individual is likely to have given across multiple impressions.

This ‘Cumulative Attention’ can then be correlated to shifts in brand awareness, consideration, preference and claimed purchase intent versus a matched control sample. 

This approach builds on the one we adopted with Havas and Brand Metrics in 2024, popularised by Dr Grace Kite and Tom Roach as ‘Lots of Littles’. Only this time, we’re analysing the impact of lots of ‘Littles’, ‘Bigs’, and a combination of both. 

Lots of bigs and lots of littles: Lumen finds cumulative attention drives brand outcomes

Imagine three different respondents: 

Respondent A has been exposed to three ads on Amazon Prime, each of which we estimate will generate 16 seconds of attention, for a Cumulative Attention of 48 seconds. 

Respondent B is also exposed to three ads, but they are digital display ads, each generating 1.5 seconds of attention, for a cumulative attention of 4.5 seconds. 

Respondent C has a mix: one Prime ad and 10 display ads, for a Cumulative Attention total of 31 seconds and a frequency of 11. 

If repetition were all that mattered, Respondent C should be much more likely to remember the brand. If repetition suffered from massively diminishing returns, Respondent B should be as likely to remember the brand as the other respondents. But, of course, Respondent A is more likely to remember and prefer the brand. 

Consistently across categories and countries, we found that Cumulative Attention was far more predictive of memory and attitudinal outcomes than ‘naive’ frequency counts. 

And, in line with our previous research, we found that increases in Cumulative Attention are consistently correlated with improved brand lift outcomes.

Different outcomes require different volumes of Cumulative Attention

Merely reminding people that a brand exists might require 21 seconds of attention; getting people to change their minds about a product might need 40 or more seconds; getting people to put it on their shopping list might require 68 seconds or more.

None of these objectives can be achieved via attention to a single impression – even a single CTV impression. Advertisers need a certain level of cross-media frequency to cross the minimum thresholds required to create or maintain memories. How many ads are served matters less than how much Cumulative Attention is delivered. 

This is a groundbreaking insight and is likely to have major implications for how we plan and buy media for years to come. But it’s not an iron rule. While the correlations between Cumulative Attention and memorisation are strong, the data vary, and the results include outliers.

Which brings us to the third lesson of the Dissoi Logoi, the cutting-edge communications insight from 2,500 years ago: What you hear, place on what you know…[f]or courage, place it on Mars or Achilles; for metal working, on Vulcan; for cowardice, on Epeus.

What the anonymous sophist is telling us is that some things are much easier to remember than others – things that are associated with well-known stories, or myths or celebrities are shortcuts to memory.

A psychologist like Orlando Wood might call these fluent devices; a classicist like Paul Feldwick might call it fame. 

Good creative – that grabs attention and is easy to remember – is likely to require less Cumulative Attention than the dull and ordinary fare that is served up to us most of the time. Its power lies in its attentional efficiency – doing more work with less attention. 

The concept of cross-media Cumulative Attention can help us optimise to a robust and reliable norm. But there’s always room for great creativity to outperform the algorithm. 

Mike Follett is the founder and CEO of Lumen Research.