Emotion & Attention Statistics 2026: 55+ Sourced Stats on Ads, Eye Tracking and Emotion AI 

Most ads barely get seen, and the ones that do work mostly through emotion. That is the short version of what the research on attention and emotion shows, and it is why the gap between what people say and how they actually react matters so much for brands.

We've gathered the most reliable statistics on advertising attention, emotional response, eye tracking, packaging, the say–do gap, emotion AI and AI in market research. Every figure links to its source and carries its year, so you can cite it with confidence. Where an original report isn't publicly available, we say so and link to the coverage that reported it. We've also flagged the popular "stats" that don't hold up.

Last updated: 8 October 2026. Compiled by the research team at Entropik.

Key statistics at a glance

• 85% of digital ad views get less than 2.5 seconds of active attention, which Amplified identifies as the general threshold for forming a memory. (VCCP Media × Amplified, 2025)

• 2x: emotional campaigns are about twice as efficient, and deliver about twice the profit, of rational ones. (Les Binet & Peter Field, The Long and the Short of It: Balancing Short and Long-Term Marketing Strategies, IPA, 2013)

• 27% → 48%: adding an emotion-based creative score to media spend nearly doubles how well it predicts brand growth. (System1, 2025)

• ~300% more brand recall among viewers who skipped, when the ad showed the brand from the start. (Google/Ipsos, 2015)

• 57% of web page viewing time is spent above the fold. (Nielsen Norman Group, 2018)

• 65% vs. 26%: the share of consumers who say they want to buy sustainable, purpose-driven brands versus the share who actually do. (Harvard Business Review, 2019)

• 71% of market researchers expect synthetic responses to make up more than half of data collection within three years. (Qualtrics 2025 Market Research Trends report, published October 2024)

Attention in advertising statistics

An ad being "viewable" is not the same as an ad being seen. Eye-tracking research shows most impressions get a second or two of real attention, and many get none.

Attention vs. viewability

• 85% of digital ad views receive less than 2.5 seconds of active attention, the general threshold Amplified has found for forming a memory. (VCCP Media × Amplified, 2025)

• Over 80% of ads don't reach the 2.5-second attention-memory threshold. (Amplified)

Attention by platform and format

• Some social ad formats deliver just 0.4 to 1.6 seconds of active attention per view, and most of that attention is captured in the first 2.5 seconds. (Amplified, 2025; 59 billion+ attention data points)

• On YouTube, about 70% of ads are skipped regardless of ad length. YouTube ads averaged 5.6 seconds of active attention, versus 13.7 seconds for broadcaster video on demand. (Amplified for Via, Belgium, 2022)

• Viewer attention to connected TV (CTV) ads reached 51.5% in Q1 2024, closing in on linear TV at 54.5%. (TVision × DoubleVerify, 2024; 5,000-home US panel)

Attention and memory

• 2.5 seconds of active attention is the general threshold Amplified has found for an ad to be committed to memory. (Amplified)

• For brands with strong distinctive assets, such as a well-known logo, colour or character, as little as 1.5 seconds of active attention can be enough to form a memory. Without them, the 2.5-second threshold applies. (VCCP Media × Amplified, 2025)

• Every well-branded second of attention delivered 2.5 times the ROI of an unbranded equivalent. (VCCP, Hacking the Attention Economy)

Screen attention in daily life

• People now stay on any one screen for about 47 seconds before switching, down from about 2.5 minutes in 2003–04. (Gloria Mark, UC Irvine, Attention Span, 2023) This measures how often people switch tasks, not a biological limit on attention.

Emotion and advertising effectiveness statistics

Emotion is the best-evidenced creative driver of long-term advertising results. Both large campaign databases and ad-testing datasets point the same way.

• Emotional advertising campaigns are about twice as efficient as rational ones and deliver about twice the profit. (Binet & Field, The Long and the Short of It, IPA, 2013; 996 campaigns, 700 brands, 83 categories)

• Long-term advertising investment (3+ years) delivers double the profit of a short-term approach (under 1 year). (Binet & Field, The Long and the Short of It, IPA, 2013)

• The most effective budget split is about 60% brand building and 40% sales activation. (Binet & Field, The Long and the Short of It, IPA, 2013)

• Fame campaigns were more than twice as likely as others to deliver very large reductions in price sensitivity (8% vs. 3%). (Binet & Field, The Long and the Short of It, IPA, 2013)

• Media spend alone predicts 27% of brand growth. Adding an emotion-based creative score raises that to 48%. (System1, 2025; 264 brands, 4,000+ ads)

• 92.1% of US ads with top emotional scores (4–5 Stars) also delivered above-average short-term sales impact. (System1; 100,000+ ads tested)

• Nearly half of ads in System1's database score just 1 Star for emotional impact, and only 1% score 5 Stars. (System1)

• The average US ad scores 2.4 Stars on System1's 1.0–5.9 scale. (System1, February 2026)

• Short digital ads (5–13 seconds) average 1.9 Stars, below the 2.2-Star global norm. (System1; 5,000+ short digital ads)

• On Pinterest, ads that left people feeling more positive saw 47% higher return on ad spend and a 20% higher lift in ad recall. (System1 × Pinterest, 2024; 50 brand-lift studies)

• Ads with a high concentration of happiness and surprise drove more total attentive seconds than ads with low concentrations. (Amplified for Pinterest)

A note on these sources: the IPA data comes from campaigns entered for effectiveness awards, and System1 uses its own testing method. Both are widely respected, but neither is a controlled experiment.

Branding, memory and recall statistics

When attention is this short, when the brand appears matters as much as whether it appears.

• Among viewers who skipped an ad, the version that showed the brand from the start generated nearly 300% more brand recall. Overall brand recall rose 200%, with no drop in view-through rate. (Google/Ipsos, Libresse case study, 2015; 1,200 respondents in Denmark)

• YouTube ads that show the brand in the first five seconds achieve higher ad recall and brand awareness lift, although viewers are also more likely to skip them. (Think with Google, "The First 5 Seconds," 2015; thousands of ads across 16 countries)

• Google's creative guidance, based on analysis of 5,000 YouTube ads: introduce your product or brand in the first five seconds. (Google/Ipsos, ABCD framework)

• Campaigns that created very large salience effects were almost twice as likely to deliver top-level short-term sales growth and long-term share growth. (Binet & Field, The Long and the Short of It, IPA, 2013)

Eye-tracking statistics: visual attention on websites

People scroll more than they used to, but the top of the page still gets most of the attention.

• Users spend about 57% of their page-viewing time above the fold, and 74% within the first two screens. (Nielsen Norman Group, 2018; lab eye-tracking, desktop)

• The second screen of a page gets about 17% of viewing time. The remaining 26% is spread across everything below it. (Nielsen Norman Group, 2018)

• In 2010, 80% of viewing time was spent above the fold. By 2018 that had fallen to 57%, and 81% fell within the first three screens. (Nielsen Norman Group, 2018)

Packaging and shelf attention statistics

Many buying decisions still happen at the shelf, which makes the few seconds a pack gets noticed critical.

• 76% of purchase decisions in grocery stores are made in store. (POPAI Shopper Engagement Study, 2012, as reported by Drug Store News)

• In mass-merchant stores, 82% of purchase decisions are made in store. (POPAI Mass Merchant Study, 2014, as reported by KDM)

• 66% of observed product "grabs" resulted in a purchase, and nearly 1 in 6 brand purchases happened when a display for that brand was present. (POPAI, 2012, as reported by Chief Marketer)

• 13% of shoppers' eye fixations in store landed on in-store displays. (POPAI, 2012, as reported by Chief Marketer)

• Attention to a product falls as it moves further from eye level, and shelf position mattered more than packaging factors. (Eye-tracking study of in-store displays)

The POPAI figures count "generally planned" and substitute purchases as in-store decisions, so they describe how much the shelf shapes the final choice rather than pure impulse buying.

The say–do gap: why self-reported research falls short

What people say they will do and what they actually do often diverge. That gap is why researchers increasingly measure behaviour and emotion alongside survey answers.

• 65% of consumers say they want to buy purpose-driven brands that advocate sustainability, yet only about 26% actually do. (Harvard Business Review, "The Elusive Green Consumer," 2019)

• 63% of consumers globally say they prefer to buy from companies whose purpose reflects their values. (Accenture Strategy, 2018; ~30,000 consumers, self-reported)

• GfK valued the gap between sustainable-buying intentions and behaviour, covering 36% of consumers, at $806 billion of untapped opportunity in consumer goods. (NielsenIQ/GfK, 2022)

• Many shoppers report comparing brands and prices at the shelf, but watching the same shoppers shows they often don't look at alternatives at all. (Gerald Zaltman, Harvard Business School, 2003)

UX frustration and abandonment statistics

Friction costs sales. Most shoppers who add to cart never finish, and poor checkout design is a leading reason.

• The global average online cart abandonment rate is 70.22%. (Baymard Institute, 2026)

• 59% of leading US and European e-commerce sites have "mediocre" or worse checkout UX performance, and only 2% rate as "good." (Baymard Institute, 2026)

• Extra costs being too high, such as shipping, taxes and fees, is the leading cause of cart abandonment apart from just browsing, at 40%. (Baymard Institute, 2026 quantitative study)

• Across 14 years of usability testing, Baymard has consistently found that checkout design and flow are often the sole cause of users abandoning a purchase. (Baymard Institute, 2025)

Emotion AI statistics: market size, accuracy and regulation

Emotion AI is growing fast, but forecasts vary widely and the science has real limits. Both sides belong in any honest picture of the field.

Market size

Estimates for the emotion AI market in 2025 range from about $3 billion to over $40 billion, depending on how each firm defines the market. Forecast growth rates range from about 15% to 42% a year.

Accuracy and limits

• A major scientific review found that how people express anger, disgust, fear, happiness, sadness and surprise varies substantially across cultures, situations and individuals, and that the same facial movement can signal more than one emotion. (Barrett et al., Psychological Science in the Public Interest, 2019)

This is why facial coding works best as one signal among several, combined with context, eye tracking, voice and what participants say, rather than as a direct readout of what someone feels.

Regulation

• Since 2 February 2025, the EU AI Act bans AI systems that infer emotions in workplaces and schools, except for medical or safety reasons. Violations carry fines of up to €35 million or 7% of global turnover. Emotion recognition in other settings, such as consented consumer research, is classed as high-risk. (EU AI Act, Regulation (EU) 2024/1689, Article 5(1)(f))

AI in market research statistics

AI has moved from experiment to everyday tool in market research, and researchers expect AI agents and synthetic data to handle much more of the work within a few years.

• 89% of market researchers use AI tools regularly or experimentally, and 83% say their organisations plan to significantly increase AI investment. (Qualtrics 2025 Market Research Trends report, published October 2024; 3,198 researchers, 14 countries)

• 71% agree that within three years, synthetic responses will make up more than half of data collection. (Qualtrics 2025 Market Research Trends report, published October 2024)

• 74% of researchers who use AI regularly have seen demand for qualitative research increase, compared with 22% of those not using AI. (Qualtrics 2025 Market Research Trends report, published October 2024)

• 78% of researchers predict AI agents will run more than half of all research projects by 2028. 15% already use agentic AI. (Qualtrics 2026 Market Research Trends report, published November 2025; 3,000+ researchers)

• Research teams using only basic AI are four times more likely to lose influence in their organisation. (Qualtrics 2026 Market Research Trends report, published November 2025)

• 67% of research suppliers now build generative AI directly into client deliverables. (Greenbook GRIT Report, 2025; article)

• 50% of insights professionals are reskilling to work with AI, and 33% of research buyers now value AI literacy over traditional expertise. (Greenbook GRIT Report, 2025; article)

Market research industry statistics

The global insights industry passed $150 billion in 2024, and software is now its largest segment.

• The global insights industry was worth about $153 billion in 2024 and was expected to pass $160 billion by the end of 2025. (ESOMAR Global Market Research 2025, published in ESOMAR's Research World)

• Of that total, research software accounts for $62 billion, traditional market research for $56 billion and reporting for $35 billion. (ESOMAR, 2025)

• Within the $56 billion market research sector, about 62% is full-service research, 31% subscription and software services, and 7% consultancy. (ESOMAR, 2025)

Attention and emotion myths, debunked

Some of the most-shared "statistics" about attention and the brain have no reliable source. Here's what to use instead.

How we compiled these statistics

We traced each figure to its original source wherever possible and linked directly to it. We prioritised peer-reviewed research, industry bodies such as the IPA and ESOMAR, and large-sample studies from established measurement firms. Older landmark studies are labelled with their year. Where the original report isn't publicly available, as with the POPAI shopper studies, we say so and link to the coverage that reported it.

Much attention research comes from companies that sell attention measurement, including Amplified, TVision and System1. These are the industry's standard datasets, but they are not peer-reviewed, so we name the source for every figure. We update this page regularly and remove figures that are superseded.

How to cite this page

You're welcome to use any statistic on this page. Please cite the source linked beside it, and link to this page as your reference: Entropik, "Emotion & Attention Statistics 2026," entropik.io.

Want to measure attention and emotion on your own ads?

Decode by Entropik shows where people look and how they feel, using webcam eye tracking, facial and voice emotion AI, and predictive attention heatmaps.

Frequently asked questions

1. What is the average attention span for ads?

It depends on the format. Some social ad formats get just 0.4 to 1.6 seconds per view. Overall, 85% of digital ad views get less than 2.5 seconds of active attention.

2. Is the 8-second attention span real?

No. The "8 seconds, shorter than a goldfish" claim has no underlying research behind it. Research by Gloria Mark at UC Irvine shows people switch screens about every 47 seconds, but that measures task-switching, not a limit on how long people can pay attention.

3. Do emotional ads really work better?

Yes, according to the largest datasets available. The IPA's analysis of 996 campaigns found emotional campaigns were about twice as efficient and twice as profitable as rational ones.

4. How accurate is facial emotion recognition?

Facial expressions vary across cultures, situations and individuals, so no system can read a person's feelings from their face alone with certainty. Facial coding is most reliable when combined with other signals, such as eye tracking, voice and self-reported responses, and when results are compared across groups rather than individuals.

5. Is emotion AI legal?

It depends on the setting. In the EU, emotion recognition in workplaces and schools has been banned since February 2025, except for medical or safety reasons. In other settings, such as consented consumer research, it is allowed but classed as high-risk.