THE WORST ADS ON FOOTBALL GAMES MAKE THE MOST IMPACT
Your skip button is irrelevant. Your brain already did its job.
Chris Walker left a comment on my last article that I have not been able to stop thinking about: “I couldn’t tell you a single product I’ve seen advertised on YouTube.”
I wanted to share something important. The fact that we believe our inability to recall proves the ads failed is exactly the cognitive error that has led an entire industry to misallocate hundreds of billions of dollars.
I hold a PhD in Physics and an Executive MBA from IE Business School. I taught machine learning and AI at Sofia University for ten years. I now run a platform that processes data for 250+ B2C brands across 15 countries. This article explains why the ads you skip, mock, and forget are the ones doing the heaviest cognitive lifting, and why the measurement infrastructure that decides your budget cannot see any of it.
Your brain is not your ally here
Robert Zajonc demonstrated this in 1968 in a landmark study published in the Journal of Personality and Social Psychology [1]. Repeated exposure to a stimulus increases preference for it, even when the person has zero conscious memory of the exposure. Zajonc showed participants Chinese characters, nonsense words, and photographs. The participants could not recall seeing them. But when asked to rate them, they consistently preferred the stimuli they had been exposed to before. They could not name the stimulus. They could not describe it. But when forced to choose, they leaned toward the familiar option without knowing why.
This is not a quirk. This is how your purchasing decisions actually work.
The mechanism is called processing fluency [2]. When you encounter something familiar, your brain processes it more easily than something novel. This ease of processing generates a subtle positive feeling that the brain misattributes to the stimulus itself rather than to the processing experience. You do not think “I have seen this before, so my brain is processing it quickly.” You think “I like this brand.” The feeling of fluency is experienced as a feeling of preference. This misattribution is the key to understanding why the mere exposure effect is so powerful and so resistant to conscious override [3].
A meta-analysis by Bornstein (1989) reviewing over 200 mere exposure studies from 1968 to 1987 confirmed the effect’s robustness across stimulus types, exposure durations, and population groups [4]. Further research found the effect works best with 10 to 20 exposures, and that subliminal exposure (below the threshold of conscious awareness) actually produces stronger preference effects than supraliminal exposure [5]. Your brain registered the ad. Your conscious mind did not. The preference formed anyway.
Sofie Sue Rutgeerts, who leads TV and CTV strategy at egta, pushed back on this point in the LinkedIn discussion: “Mere exposure explains part of the story. But reducing effectiveness to familiarity alone risks ignoring the role of meaning, context, and emotion in how brands are actually chosen.” She is right. But the two mechanisms are not in conflict. They compound. Binet and Field’s analysis of the IPA Databank showed that emotionally rich, broad-reach brand campaigns deliver the largest business effects over time [6]. The mere exposure effect is the substrate on which emotional resonance builds. You cannot build emotional associations with a brand the customer has never been exposed to.
The $8 million question nobody asks correctly
Brands spent $8 million for 30 seconds during this year’s Super Bowl. Premium placements went for over $10 million, a new record confirmed by NBCUniversal’s Mark Marshall [7]. Super Bowl LIX in 2025 drew a record 127.7 million average viewers, with peak viewership hitting 137.7 million [8]. Total advertising revenue for a single four-hour broadcast approached $600 million [7].
Marketing Twitter ranked which ads were “best” and “worst.” Cannes Lions will go to the clever ones. Reddit roasted the cringe.
But the ranking is irrelevant to the outcome.
Byron Sharp and the Ehrenberg-Bass Institute call it mental availability: the probability that a brand comes to mind in a buying situation [9]. It does not require you to like the ad. It does not require you to remember the ad. It requires that enough neural pathways exist so that when you are standing in an aisle or typing into a search bar, the brand feels slightly more familiar than the alternative. Mental availability has a direct relationship to market share [10]. Not preference. Probability.
Sharp’s research, documented in “How Brands Grow” (2010), established that brands grow primarily by increasing their buyer base through broad reach and mental availability, not through deep loyalty or emotional differentiation [9]. The implications for advertising are stark: the purpose of most advertising is not to persuade. It is to build and refresh memory structures that make the brand easier to notice and easier to recall at the moment of purchase.
The nuance most people get wrong
I am not arguing that cringe is a strategy. The research is more precise than that.
A 2021 study from the University of Illinois found that emotionally arousing ads hurt immediate memory but improve delayed memory, the kind that matters weeks later when a purchase decision actually happens [11]. The catch: this only works when the emotional arousal matches the ad’s core message. A study in the journal Frontiers in Psychology confirmed that the mere exposure effect in advertising is modulated by attention: when viewers attended to a stimulus, the positive attitude shift was stronger, but the effect persisted even with peripheral processing [12].
There is another boundary. The mere exposure effect works best with stimuli that start neutral or slightly positive [1]. If an ad provokes genuine hostility on first contact, repetition reinforces the negative association. A study on disruptive advertising actually found positive effects on consumer preferences when the disruption was mild, but negative effects when it crossed the threshold into genuine annoyance [13].
The sweet spot: ads that are mildly annoying, slightly cringey, a bit too loud. Those do the heaviest cognitive lifting. Irritating enough to process. Not offensive enough to trigger active rejection.
Folgers ran this exact playbook for 21 years with “Mrs. Olson.” Research found consumers increasingly perceived the campaign as annoying. It made Folgers the number one coffee brand in America [14].
Chris Walker made an insightful observation in the discussion: he noted that during live rugby matches, Samsung and Virgin Atlantic ran split-screen ads that felt “like a jarring intrusion” but the brands stuck. “I didn’t like them,” he wrote, “but I can’t deny that the brands stuck.” That is processing fluency working in real time. The disruption forced cognitive processing. The processing created familiarity. The familiarity created preference. Chris proved the thesis while arguing against it.
The measurement infrastructure is designed to hide this
Standard digital attribution operates on 7-day or 28-day windows. A football game ad that aired on November 12 and influenced a purchase on December 3 does not exist in the data.
This creates a structural feedback loop: channels that create demand look expensive and unproductive. Channels that capture demand look efficient. Budgets shift from creation to capture. The funnel starves from the top.
Brainlabs ran 17 Meta conversion-lift studies for 12 different advertisers and found that paid social drove a 19% incremental increase in search visits [15]. Those visits, and any conversions from them, would have been attributed to paid or organic search, not to the Meta ads that caused them. 71% of the incremental search traffic was organic, meaning the exposure drove users to search for the brand directly rather than click through an ad [15].
Meta’s own Measurement 360 report found that last-click attribution underestimates the value of upper-funnel ads by up to 40% [16].
A MediaScience study reinforced this: when a TV ad preceded a digital ad, participants had 125% higher unaided brand recall and 18% higher purchase intent compared to two digital ads alone [17].
Dennis Stam, a recruitment marketer at Radancy, laid out the diagnostic chain this creates: Reach to eSOV, eSOV to SOM, SOM to Revenue, Revenue to Profit. “Low quality of reach will stagnate eSOV,” he wrote. He is describing the mechanism exactly. When you cut upper-funnel channels because attribution cannot see them, you do not just lose impressions. You break the first link in the chain. eSOV drops. SOM follows. Revenue declines three months later. The dashboard blames the performance campaigns that were actually just riding momentum you stopped building.
The exposure timeline your dashboard cannot see
The brand story does not play out in 30 seconds. It plays out across weeks. A TV ad plants the character. A YouTube pre-roll you skip adds texture. A podcast mention deepens it. A jarring retargeting ad creates friction, a disruption in the narrative. Three weeks later, a branded search resolves the whole arc.
This is the same tension and resolution structure that works inside a single ad, stretched invisibly across the customer journey. The 7-day attribution window only sees the final scene. It credits the branded search click and erases every chapter that built toward it. Brainlabs found paid social drove 19% incremental search visits, fully attributed to Google [15]. The character was built on Meta. The credit went to the last click.
Silvia Pesheva, who connects investors with growth opportunities, captured why this matters operationally: “Marketing is 50% maths and 50% psychology. The Marketing Matrix is true: standard digital attribution operates on 7-day or 28-day windows. A football game ad that aired on November 12 and influenced a purchase on December 3 doesn’t exist in the data.” She also noted that this understanding “does not go bottom-up in marketing structure. It spreads horizontally from one CMO to another, based on results.” This is exactly right. The CMOs who understand the objective function gap between what attribution measures and what actually drives value are the ones reallocating budgets correctly.
Demand creation vs. demand capture: the system rewards capture and starves creation
Binet and Field’s IPA Databank analysis, published as “The Long and the Short of It” (2013), established what many marketers intuited but could not prove: approximately 60% of marketing budget should go to long-term brand building, and 40% to short-term activation [6]. Their subsequent work, “Effectiveness in Context” (2018), refined this to 62:38 and confirmed the finding across hundreds of additional campaign cases [18]. The optimal ratio varies by category, brand maturity, and market position, but the principle is consistent: long-term brand building delivers the majority of profit growth.
The structural consequence: performance marketing is parasitic on brand marketing. Retargeting ads and branded search do not create demand. They capture demand that was created elsewhere, often by the channels that attribution systems cannot see. When you cut upper-funnel spend because it “does not convert,” you are cutting the thing that makes your lower-funnel spend work.
Every CMO who has cut awareness spend and then watched branded search volume decline three weeks later has lived this. The channels you skip consciously still shape what you choose.
What this means for your stack
The attribution problem is not just a media buying problem. It is a marketing stack architecture problem. If your entire measurement infrastructure is built on last-click attribution, you are systematically overvaluing demand capture and undervaluing demand creation.
This connects directly to what Brinker and Riemersma called “Attribution 2.0” in their 2026 State of Marketing Attribution report: the shift from attribution as credit allocation to attribution as a decisioning discipline [19]. Attribution 1.0 asks: which channel gets the credit? Attribution 2.0 asks: what is the next best action for this customer’s predicted lifetime value trajectory?
The shift requires three architectural changes:
First: measure value, not events. Stop counting conversions in attribution windows. Start measuring predicted customer lifetime value over 90-day and 12-month horizons. The football game ad that does not convert within 7 days may be the highest-value touchpoint in the journey. You will never know if your measurement window is shorter than your customer’s decision cycle.
Second: capture 100% of signals. Server-side tracking captures every visitor interaction regardless of browser restrictions, ad blockers, or cookie deprecation. When your pixel misses 30-40% of traffic, your attribution model is not wrong about which channels work. It is blind to whether the customer was even there. A decisioning platform built on incomplete data makes incomplete decisions.
Third: use an objective function that spans the full journey. The five structural blind spots in most ecommerce stacks all stem from the same root cause: the stack has no shared objective function. Ad tech optimizes for the cheapest next acquisition. Martech optimizes for the next email open. Neither optimizes for the economic outcome of the customer relationship. A decision intelligence platform closes this gap by governing every decision, from product recommendations to email timing to ad spend allocation, with a single number: predicted customer lifetime value.
The question nobody is asking
The question is not whether the football game ad was good. The question is whether you would recognize the brand in a lineup three weeks later. The answer, backed by 58 years of replicated research across over 200 studies [4], is yes.
Your “worst performing” channels might be your most important ones. Your most “efficient” channels might be parasitic, claiming credit for conversions they did not cause.
If you only measure what you can attribute, you will only invest in what you can attribute. And what you can attribute is, by definition, the last thing that happened, not the first thing that mattered.
The system rewards capture and starves creation. A decision intelligence platform is designed to see the full journey, not just the final click. Book a demo to see what the five blind spots look like in your own data.
FAQ
What is the mere exposure effect? The mere exposure effect, first demonstrated by Robert Zajonc in 1968, is the phenomenon where repeated exposure to a stimulus increases preference for it, even when the person has zero conscious memory of the exposure. It works through processing fluency: familiar stimuli are easier for the brain to process, and that ease of processing is misattributed as a feeling of preference.
How does the mere exposure effect apply to advertising? When viewers see an ad, even briefly or peripherally, their brain registers the brand’s visual signature. Later, when making a purchase decision, the brand feels slightly more familiar and trustworthy than alternatives. This effect works even when the viewer cannot recall seeing the ad, and is actually stronger when exposure is subliminal.
What is mental availability? Mental availability, a concept from Byron Sharp and the Ehrenberg-Bass Institute, refers to the probability that a brand comes to mind in a buying situation. It depends on the quality and quantity of memory structures related to the brand. Mental availability has a direct relationship to market share. It is built through consistent, broad-reach advertising, not through persuasion or deep emotional connection alone.
What did the Brainlabs Meta conversion-lift studies find? Brainlabs ran 17 conversion-lift studies for 12 advertisers and found that Meta Ads drove a 19% incremental increase in search visits. 71% of this incremental search traffic was organic, meaning the ad exposure drove users to search for the brand directly. All of these conversions would have been attributed to paid or organic search, not to the Meta ads that caused them.
What is the Binet and Field 60:40 rule? Les Binet and Peter Field analyzed the IPA Databank and found that the optimal marketing budget split for most consumer brands is approximately 60% on long-term brand building and 40% on short-term activation. Their subsequent research confirmed the finding across hundreds of additional cases. Brands that over-invest in activation at the expense of brand building see short-term gains followed by long-term erosion of market share and profitability.
Why does last-click attribution undervalue upper-funnel channels? Last-click attribution credits the final touchpoint before conversion. Upper-funnel channels like TV, YouTube, and podcast ads typically influence purchases weeks after exposure. Since the conversion happens outside the attribution window, the upper-funnel channel receives zero credit. Meta’s Measurement 360 report found this undervaluation can be as high as 40%.
How does attribution connect to customer lifetime value? Attribution 1.0 asks which channel gets credit for a conversion. Attribution 2.0, as described by Brinker and Riemersma, asks what is the next best action for maximizing a customer’s predicted lifetime value. A decision intelligence platform replaces channel-level credit allocation with customer-level value optimization across the entire journey.
5. REFERENCES
[1] Zajonc, R. B. (1968). “Attitudinal Effects of Mere Exposure.” Journal of Personality and Social Psychology, 9(2, Pt. 2), 1-27. https://doi.org/10.1037/h0025848
[2] Winkielman, P., Schwarz, N., Fazendeiro, T., & Reber, R. (2003). “The Hedonic Marking of Processing Fluency: Implications for Evaluative Judgment.” The Psychology of Evaluation, 189-217.
[3] Atticus Li (2026). “The Mere Exposure Effect: Why Retargeting Works Even When Users Don’t Click.” https://atticusli.com/blog/posts/mere-exposure-effect-retargeting/
[4] Bornstein, R. F. (1989). “Exposure and Affect: Overview and Meta-Analysis of Research, 1968-1987.” Psychological Bulletin, 106(2), 265-289. https://doi.org/10.1037/0033-2909.106.2.265
[5] Zajonc, R. B. (2001). “Mere Exposure: A Gateway to the Subliminal.” Current Directions in Psychological Science, 10(6), 224-228.
[6] Binet, L. & Field, P. (2013). The Long and the Short of It: Balancing Short and Long-Term Marketing Strategies. IPA.
[7] European Business Magazine (2026). “Super Bowl Ads Cost $10M for 30 Seconds.” NBCUniversal confirmed $8M average, premium slots exceeding $10M. https://europeanbusinessmagazine.com/business/super-bowl-ads-cost-10m-for-30-seconds-who-pays-and-why/
[8] Adwave (2026). “How Much Does a Super Bowl Commercial Cost in 2026? Super Bowl LIX (2025) averaged 127.7 million viewers.” https://adwave.com/resources/super-bowl-commercial-cost
[9] Sharp, B. (2010). How Brands Grow: What Marketers Don’t Know. Oxford University Press.
[10] Ehrenberg-Bass Institute (2020). Vaughan, K., Corsi, A., Beal, V. & Sharp, B. “Measuring the Effect of Advertising on Brand Mental Market Share.” International Journal of Market Research. https://marketingscience.info/
[11] University of Illinois (2021). Study on emotional arousal and delayed memory in advertising contexts. Published in Journal of Advertising Research.
[12] Frontiers in Psychology (2018). “The Contribution of Attention to the Mere Exposure Effect for Parts of Advertising Images.” https://pmc.ncbi.nlm.nih.gov/articles/PMC6134073/
[13] Bell, R. & Buchner, A. (2018). “Positive Effects of Disruptive Advertising on Consumer Preferences.” Journal of Interactive Marketing, 41, 1-13.
[14] Muthukrishnan, A. V. & Kardes, F. R. (2001). “Persistent Preferences for Product Attributes: The Effects of Exposure.” Journal of Consumer Research, 28(4), 486-492.
[15] Brainlabs (2026). “Proving Paid Social’s Halo: Meta Conversion Lift. 17 studies, 12 advertisers. 19% incremental search visits.” https://www.brainlabsdigital.com/paid-social-measurement-meta-incrementality-search-lift/
[16] Meta (2023). Measurement 360 Report. “Last-click attribution underestimates upper-funnel value by up to 40%.”
[17] MediaScience (2022). Study on cross-media effects. “TV ad preceding digital ad produced 125% higher unaided brand recall, 18% higher purchase intent.”
[18] Binet, L. & Field, P. (2018). Effectiveness in Context. IPA. Optimal budget ratio: 62:38 brand to activation.
[19] Brinker, S. & Riemersma, F. (2026). 2026 State of Marketing Attribution Report. “Attribution 1.0 is dead. Attribution 2.0 is about direction, not credit.”
[20] Sharp, B. & Romaniuk, J. (2016). “Mental Availability Is Not Awareness, Brand Salience Is Not Awareness.” Ehrenberg-Bass Institute blog. https://byronsharp.wordpress.com/2011/03/26/mental-availability-is-not-awareness-brand-salience-is-not-awareness/
[21] Built In (2025). “The Mere Exposure Effect in Marketing & Advertising.” Anthony Grimes, University of Sheffield. https://builtin.com/articles/mere-exposure-effect
[22] Alight Media (2025). “The Power of Psychology in OOH: The Mere Exposure Effect.” https://alightmedia.com/news/the-power-of-psychology-in-ooh-the-mere-exposure-effect
[23] KINESSO / Meta (2024). “Searches Originate From Somewhere.” 4% incremental paid search traffic, 39% increase in organic search visits. https://www.performancemarketingworld.com/article/1900904/searches-originate-somewhere
[24] Made Up Mind (2026). “What Is the Mere Exposure Effect?” Mere exposure effect peaks at 10-20 exposures, subliminal exposure stronger than supraliminal. https://madeupmind.org/blog/mere-exposure-effect-guide
[25] Human Performance (2025). “The Mere Exposure Effect.” https://humanperformance.ie/the-mere-exposure-effect/
[26] EBSCO Research (2025). “Mere-Exposure Effect.” https://www.ebsco.com/research-starters/psychology/mere-exposure-effect
[27] IPA (2023). “The Next Chapter for The Long and The Short of It.” Les Binet and Peter Field retrospective. https://ipa.co.uk/knowledge/ipa-blog/the-next-chapter-for-the-long-and-the-short-of-it
[28] Marketing Week (2024). “Les Binet: Avoid Pushing Brand and Performance at the Same Time.” 60% of profit payback comes from long-term brand building. https://www.marketingweek.com/les-binet-brand-performance/
[29] Brinker, S. (2026). The New Martech “Stack” for the AI Age. Databricks. Decisioning as Ring 4. https://www.databricks.com/resources/ebook/new-martech-stack-ai-age
[30] Gartner (2026). Magic Quadrant for Decision Intelligence Platforms. January 2026.
