How can you tell if viewers are really paying attention to a video ad? What an International Journal of Advertising study found
Advertisers are being urged to buy attention instead of impressions, and most attention data come from tracking eyes. A lab comparison of eight body signals, checked against brainwaves, found that a viewer’s heart rate separated well-watched ads from weaker ones far better than their gaze did.
Hartnett et al. find heart rate sorted 9 of 10 video ads into high or low attention in a lab test, while eyes-on-screen tracking did no better than guessing.
Marketing in 3 · 2024 in research · No. 49 · Paper published 3 December 2024
Counting eyes on a screen is easy. Knowing what gets through is harder. Photo: Krists Luhaers on Unsplash.
The short answer
Watch their heart rate, not just their eyes. In a lab, 261 viewers sat through an hour of TV with ad breaks while sensors recorded eight body signals, which were then checked against brainwave readings and the viewers' own ratings. A slight slowing of the heart, compared with each person's resting rate, was the only scalable signal that reliably separated ads people attended to from ads they half-watched. Gaze-based scores missed the difference.
Evidence level
★★★One study
★★★Several studies
★★★Meta-analysis
Key takeaways
A heart beating more slowly than at rest put 9 of the 10 test ads in the right group, high or low attention, matching the brainwave benchmark.
Eyes-on-screen time and fixation counts labelled all 5 low-attention ads as high-attention ones, because viewers kept looking at the screen through the weaker ads too.
Averaged across the eight body signals, accuracy was 54%, barely above the 50% that pure guessing would score.
Only the high-attention ads changed which brand people picked: they captured 27% of the possible gain over viewers who saw no ads, while low-attention ads captured none.
What the researchers did
Nicole Hartnett, Steven Bellman, Virginia Beal and Rachel Kennedy of the Ehrenberg-Bass Institute for Marketing
Science at the University of South Australia in Adelaide, Claire Charron of Google's Market Insights team in London
and Duane Varan of MediaScience in Austin, Texas, looked at a question that now shapes how media gets bought. Agencies
and industry groups want ads priced on the attention they win rather than on impressions, and the attention figures
sold today are mostly based on where people's eyes point. The team asked whether the eyes are the right thing to watch.
They tried eight body signals that could in principle be gathered from large numbers of people: heart rate, four
eye-tracking measures, two readings of skin conductance (tiny changes in sweat on the fingertips) and smiling picked
up by a camera. Each one was set against EEG, a cap of electrodes that reads brainwaves. EEG is treated as the gold
standard for attention, but it needs a lab, so it cannot follow whole audiences.
To know in advance which ads deserved attention, the team began with 324 thirty-second commercials that had aired in
the United States, chose 100 varied ones and ran an online pre-test in which consumers rated how closely they had
followed each one. The five rated highest and the five rated lowest became the test ads. In the lab, every viewer
first watched four minutes of a calm Zen garden video, which gave a resting reading on every sensor. Then came an
hour of television with ad breaks.
Test ads
Ten 30-second commercials from ten brands, in categories such as snacks, skin care, cars, cleaning products and internet services
Pre-test
636 US consumers from an online panel each rated 10 of the 100 shortlisted ads for how much attention they gave them
Viewers
261 people in a lab, 181 with complete data on every measure; each analysis used all the data available
Session
A four-minute calm video for the resting reading, then a drama, sitcom or reality show of the viewer's choice with 30 ads in five breaks, shuffled for each person
Sensors
Finger electrodes for heart rate and skin conductance, infrared eye trackers, a camera for facial expressions and a nine-electrode EEG cap
Afterwards
Brand recall, recognition of each ad's scenes, a 0 to 100 attention rating per ad and a choice between rival brands; 145 people who saw no ads answered the same survey
Funding
Paid for by Google; the authors note that both research groups also receive money from Google's competitors
Which signal told the two kinds of ad apart?
Each measure was given a rule taken from earlier research. An ad counted as attended to if viewers' readings moved
away from their resting state in the predicted direction: a slower heart, more looking at the
screen, more smiling, and so on. A good measure should flag the five strong ads and leave the five weak ones alone.
Heart rate did that for nine ads in ten, the same score as the EEG cap. It also lined up with how viewers rated each ad
themselves (a correlation of 0.85, ignoring the sign) and with the brainwave reading (0.79).
Nothing else beat chance. Blink length, and how far a viewer's gaze strayed from where everyone else was looking,
reached six in ten. Short spikes in skin conductance did worst, at two in ten, and went against the prediction
altogether: they tended to rise as attention fell.
Ads correctly classed as high or low attention against the viewer's resting state (%)
Ads correctly classed as high or low attention against the viewer's resting state (%)
Group
Value (%)
Heart rate
90
Brainwaves (EEG benchmark)
90
Gaze straying from the crowd
60
Blink length
60
Smiling
50
Fixations per second
50
Eyes on screen
50
Skin conductance level
50
Skin conductance spikes
20
Lab experiment with 261 viewers (202 to 256 usable readings per measure) and 10 test ads. Guessing would score 50%.
Why the eye-tracking measures were fooled
The trouble came from the weaker ads. Set against the calm video, they still drew more looking and more fixations
than people gave at rest, so the eye-based measures mistook them for strong ones. Across the eight signals, strong ads were spotted
75% of the time but weak ones only 33%.
Ads are made to catch the eye, the authors point out. Movement and other eye-catching features pull the gaze on their own,
whether or not anyone is thinking about the message. Eyes show where attention points; they say little about whether
the ad is being processed. A slower heart reflects that processing.
Smiling failed for another reason. All ten ads were light and upbeat, so viewers smiled at the weak ones too. Skin
conductance may need stronger material: these ten commercials probably stirred people far less than the bursts of noise
used to validate the measure, and the authors suggest testing it next with fear appeals.
A second yardstick: each viewer's average ad
Because a calm video sets such a low bar, the team tried a yardstick with no theory behind it but obvious practical
appeal: how each viewer responded on average across all 30 ads in the session. Judged against that, smiling caught
up with heart rate, and two of the eye-tracking measures, fixations per second and how far the gaze wandered from
the crowd, reached seven in ten.
The authors see a use for this in large panels. If consenting viewers each had an average built from thousands of
ads, even small shifts above or below it would carry weight. On this yardstick, blink length and eyes-on-screen
already pointed the right way for eight ads in ten, although several of those shifts were slight and could have come about by luck.
Ads correctly classed when each one is compared with the viewer's average ad (%)
Ads correctly classed when each one is compared with the viewer's average ad (%)
Group
Value (%)
Heart rate
90
Smiling
90
Fixations per second
70
Gaze straying from the crowd
70
Eyes on screen
60
Blink length
50
Skin conductance level
40
Skin conductance spikes
40
Same lab session; the yardstick is each person's mean response over all 30 ads shown. The EEG cap scored 80% here.
Does the extra attention pay?
Attention is only worth buying if it changes something. The survey answers were set against those of a control group
that never saw the ads. Both kinds of ad raised brand recall and recognition, so the weaker ones were
not wasted, but the strong ones did far more. Recognition of the ads' scenes rose furthest, close to the ceiling
after a high-attention ad.
The gains are expressed as a share of the room each brand had left to grow: the rise over the control group,
divided by the most it could have risen. That puts big and small brands on equal terms.
Gain over an unexposed control group, as a share of the most each measure could rise (%)
Gain over an unexposed control group, as a share of the most each measure could rise (%)
Group
Value (%)
Brand recall, low-attention ads
10
Brand recall, high-attention ads
25
Recall with category hint, low
17
Recall with category hint, high
38
Ad scenes recognised, low
59
Ad scenes recognised, high
93
Brand chosen, low
0
Brand chosen, high
27
Viewers in the lab against a control group of 145 who did a decoy task. Every gain was above zero except brand choice after low-attention ads.
Marketing in 3 steps
Ready. Set. Market.
1
ReadyDecide and prepare
Ask what sits behind the attention score. Before you pay for media priced on attention, ask the seller in writing which signal produces the number: eyes on the screen, time in view, or a body signal such as heart rate. If it is only eyes or time in view, read the score as a chance to be seen, not proof the ad got through.
Build a yardstick from your own ads. Pick five of your ads that customers clearly remember and five they barely recall. Keep these ten as a test set for any attention tool you are offered.
2
SetBuild it
Make every tool pass the test. Ask each vendor to score your ten ads before you sign. Check that the five memorable ads get higher scores than the five forgettable ones. If the tool mixes them up more than twice, ask for a different measure.
Add heart rate to your biggest launch. For your largest video campaign of the year, book a viewing panel that records heart rate with a wristband or finger sensor. Ask for a resting reading from each viewer first, taken during four minutes of calm footage such as a nature film.
3
MarketLaunch and measure
Ask viewers straight after. After each ad test, add one question: how much attention did you pay to this ad, from 0 to 100? Put the answers next to the sensor data and look again at any ad where the two disagree.
Check recall and brand choice. A few days after launch, ask people who saw the campaign and people who did not to name brands in your category and to choose between yours and a rival. Move budget towards the versions that shift both answers.
Found this useful?
Marketing in 3 is free and ad-free, with new issues every Monday and Thursday. A coffee keeps it that way.
Small heart-rate shifts. Average readings moved by only one or two beats a minute between the calm video and the ads, so each viewer needs a resting reading and the panel needs to be large.
Gaze scores bunch near the top. Viewers' eyes were on the screen about nine tenths of the time, so in an attentive setting eye data will rate most ads highly.
Returns level off. Past research quoted by the authors shows that twice the attention does not bring twice the effect, so more attention helps but each extra unit buys less.
More than one number. No single signal covered everything. The right mix depends on whether you need ads to be consciously noticed or are content with light, passing exposure.
Where it does not work
Forced viewing in a lab. People sat facing a large screen with nothing else to do: no phones, no chatter, no second screen.
Ten upbeat ads. All were 30-second US commercials with a positive tone, and the authors concede that their low-attention group was middling rather than ignored.
Lab-grade sensors. Heart rate came from finger electrodes. Reading it through webcams or wearables, which the authors propose for large-scale use, was not tested.
Two levels only. Ads fell into just two groups, strong or weak on attention, with no finer grades and no measure of deeper thinking about the message. Pupil size and facial muscle electrodes were left out.
Survey outcomes. Recall, recognition and brand preference were asked straight after viewing; nobody's shopping was tracked.
Questions marketers ask
Which measure tracks attention to video ads most accurately?
Of eight signals that could be collected at scale, heart rate came out best in a lab comparison. A small slowing of the heart, measured against each viewer's resting rate, told the well-attended ads from the rest as well as a far costlier brainwave reading did (Hartnett et al., 2025).
Does eye tracking measure attention to ads?
It measures where the eyes point. Ads are full of movement that draws the gaze automatically, so someone can look at the screen without taking much in. In a controlled lab, eye-based scores rated weak ads as highly as strong ones. The authors think simple eyes-on-screen counts may do better in distracted, real-world viewing, where looking away is common.
Why does heart rate slow down when people pay attention?
When something catches our interest, the body briefly calms so the senses can take it in, and the heart beats a touch more slowly. This is known as an orienting response. Because it reflects processing rather than gaze, it works for sound as well as pictures, which makes it usable for radio too.
Is it worth simply asking viewers how much attention they paid?
Yes, as one signal among several. Viewers' own 0 to 100 ratings tracked the brainwave reading closely, with a correlation between them of 0.88, ignoring the sign. The authors argue that asking captures the conscious processing advertisers care about, though it misses attention people are unaware of.
What is an attention currency in media buying?
A way of pricing ad space on the attention an ad receives instead of on impressions. One industry proposal prices ads per 1,000 seconds of attention, judged by eyes on the screen. On this evidence, eyes alone can overstate attention, and heart rate is a candidate alternative.
When creators speak out on causes such as gender-based violence or racism, does teaming up with a brand or a charity help? A run of experiments and a review of Instagram comments found followers judged the creator more genuine when the partner suited the creator and the cause matched what the creator already stood for. That sense of sincerity warmed people to the creator and led to small real actions, from looking up the issue to signing a petition. A creator's record on the cause is worth reading before any contract is signed.
Two experiments with about 400 people, one run in the field and one built on scenarios, compared chatbots that came across as friendly with ones that came across as capable. A friendly manner gave the bot a more human feel, capability made its advice more convincing, and both left people readier to buy and to recommend it. People who already knew a fair bit about chatbots were even quicker to see the friendly bot as human. A chatbot's script deserves the attention usually reserved for the software running it.
Source
Hartnett, N., Bellman, S., Beal, V., Kennedy, R., Charron, C., & Varan, D. (2025). How to accurately measure attention to video advertising. International Journal of Advertising, 44(1), 184-207. https://doi.org/10.1080/02650487.2024.2435164
This brief is written for practitioners from the published article. Open access (CC BY 4.0): free to read at the publisher. Numbers are as reported by the authors; read the article for full methods.
Facebook reported a return of 7.47 on retargeting. A field experiment across 210 markets found the retargeting budget could be cut to zero without losing a single order.
Shooters are the favourite of 36% of console players and 17% of phone players; strategy games 35% on laptops and 4% on consoles. How to read the gaming channel, from 116,047 players.
The same airline ad scored lower (4.31 vs 4.82 out of 7) when placed among flood posts. How to pretest social ads in a realistic feed, for free.
Bruno Schivinski is Associate Professor of Marketing at Gdańsk University of Technology. He studies how consumers engage with brands in digital and social media.