This Super Bowl Ad Generated Millions in Earned Media

Meta spent $7 million on a 30-second Super Bowl spot.

Then they made Chris Hemsworth accidentally eat a piece of art worth $6.2 million.

The result? Coverage in Fast Company, AdWeek, Hollywood Reporter, TechEBlog, and dozens of other outlets. Chris Hemsworth's TikTok teaser alone pulled 190,000+ likes before the ad even aired.

Most Super Bowl ads disappear 48 hours after the game. This one became a case study in how to make product demos entertaining enough that people actually want to watch them.

Chris Pratt walks through what appears to be an art gallery, wearing Ray-Ban Meta smart glasses. He uses the AI feature: "Hey Meta, what is this?" The glasses identify Maurizio Cattelan's "Comedian"—a banana duct-taped to a wall that sold for $6.2 million at Sotheby's in November 2024.

Chris Hemsworth enters, also wearing the glasses, casually eating a banana. Pratt does a double-take. The banana is gone from the wall—just duct tape remains.

"That's a $6.2 million banana!" Pratt yells.

They scramble through a refrigerator looking for a replacement. Reality TV icon Kris Jenner appears: "Gentlemen." The reveal—they're in her home, not a gallery. When she asks "Who eats art?", the two Chrises remind her she forgot to say "Hey Meta" first.

Directed by Matthew Vaughn (Kingsman), shot entirely with practical effects. Zero CGI.

Super Bowl placement: $7-8 million for the 30-second spot

Earned media coverage: Featured in Fast Company, AdWeek, Hollywood Reporter, WWF, TechEBlog, SlashGear, Today Show exclusive preview

Social proof: 190,100+ TikTok likes on Hemsworth's teaser post alone

Context: First-generation Ray-Ban Meta glasses had poor user adoption. This campaign specifically designed to overcome that barrier and position the product for mainstream acceptance.

Broadcast reach: Super Bowl LIX (February 2025)—most-watched TV event of the year

The ad did something most product demos fail at: it made you forget you were watching a product demo.

Three demonstrations of the "Hey Meta" AI feature happen in 30 seconds, but you're too busy laughing at Hemsworth's panic to notice you're being taught how the product works.

The banana wasn't random. Cattelan's "Comedian" was already internet-famous when it sold for $6.2 million three months earlier. Meta didn't create a cultural moment—they hijacked one that was already trending and put their product in the middle of it.

Here's the element that separated this from every other celebrity Super Bowl spot: casting two actors who already have on-screen chemistry. Hemsworth and Pratt weren't just famous faces reading lines. They were playing off established Marvel Universe dynamics. That relationship did more work than the $7 million media buy.

One more thing worth noticing—this ad for AI-powered glasses was shot with zero AI or CGI. Everything's practical. Real donuts, real smoke, real duct tape. In a moment when audiences are skeptical of AI-generated everything, showing physical reality becomes a trust signal.

What you're about to get:

This ad cost $7M to place. Here's how to steal the strategy for $0.

What they did: Opened with someone using AI glasses to identify expensive art, then immediately had another character accidentally eat it.

Why it works (psychology): Your brain notices two things: the unexpected and the high-stakes. A $6.2 million mistake triggers instant anxiety. Absurdity paired with consequences creates memorable tension.

The banana art was already culturally loaded. When something's been memed into collective consciousness, referencing it feels like an inside joke between the brand and the audience. You're not explaining the reference—you're activating existing knowledge.

Pattern interruption is the mechanism. We tune out "normal" ads. Bizarre scenarios demand processing. Once your brain engages to figure out "wait, what?", you're already paying attention through the actual product demo.

How to steal it:

Find what your audience is already talking about. Not what you think they should care about—what's actually trending in their conversations. For B2B, that's LinkedIn comment sections, industry Slack channels, subreddit threads. For D2C, it's TikTok sounds, Twitter discourse, Reddit AMAs.

Put your product in the middle of that conversation. Don't force it. The banana worked because identifying expensive art is a legitimate use case for AI glasses. Find the natural intersection between trending topic and product capability.

Budget version: User-generated content challenges. Ask customers to show your product in the wildest possible scenario. The absurdity comes from reality, not production budget.

SMB application: Create social content around the "mistake" your product prevents. Film someone almost sending the wrong email, catching it with your tool. The setup is normal, the save is your product, the result is relief.

What they did: Cast two actors with existing on-screen chemistry (Marvel co-stars) plus an unexpected authority figure (Kris Jenner) who shifts the power dynamic.

Why it works (psychology): Casting matters less than chemistry. Two strangers reading a script = expensive talking heads. Two people with established rapport = authentic reactions. Your brain recognizes genuine interaction.

The psychology here is relationship transfer. Audiences already liked the Pratt-Hemsworth dynamic from Marvel films. Seeing them together activates that positive association, then transfers it to the new context (Ray-Ban glasses).

Kris Jenner's appearance works because she shifts from background to authority. The power reversal—these two action stars are suddenly caught by a reality TV icon—adds unexpected hierarchy to the scene.

How to steal it:

Don't hire influencers in isolation. Pair people who already create together, riff together, or have existing audience overlap. Micro-influencers who frequently collaborate will outperform solo mega-influencers who've never met.

Look for complementary dynamics, not identical audiences. Pratt and Hemsworth aren't competing—one's the straight man, one's the chaos agent. Find pairs where roles are clear.

Budget version for SMBs: Use your own employees who genuinely like each other. The junior marketer and the account manager who always joke in Slack. Film them using your product to solve a real work problem. Authentic chemistry beats polished strangers.

B2B application: Customer panels, not solo testimonials. Get two clients who implemented your solution together on a call. Let them talk to each other about it, not to the camera. The conversation is the content.

What they did: Demonstrated the product three distinct times, each showing a different feature, without repeating or feeling like a feature list.

Act 1: "Hey Meta, what is this?" (Object identification) Act 2: "Hey Meta, play my gallery music" (Voice commands) Act 3: Jenner's corrected command (Reinforcing the activation phrase)

Why it works (psychology): Repetition without redundancy. Each demo advances the plot while teaching a different capability. Your brain doesn't register "I'm being taught" because you're focused on "what happens next in the story."

The rule of three is cognitive load management. One demo = you might miss it. Two demos = you start to notice the pattern. Three = pattern recognition locks in. But beyond three becomes repetitive.

How to steal it:

Structure your content as narrative with embedded education. Don't say "here are three features." Show a character facing a problem, using feature A to partially solve it, discovering they need feature B, and feature C being the unexpected solution that resolves everything.

Each demonstration should reveal character or advance stakes. In the Meta ad, the first demo shows competence (Pratt knows how to use the glasses). The second shows calm (trying to fix the situation). The third shows authority shift (Jenner corrects them). Product demo + character arc = engagement.

SMB application for product videos: Film a customer onboarding in story format. Show them struggling with the problem (pre-product), discovering your first feature, having a complication that requires your second feature, and the resolution that shows your third capability. Same structure, tiny budget.

SaaS demo structure: Build your sales demo as a narrative. "Here's how a lead enters your system [feature 1], gets qualified [feature 2], and converts [feature 3]." Same information, story structure instead of feature list.

What they did: Made the mistake happen BEFORE the audience knows it's a mistake. We discover the $6.2 million banana detail at the exact moment Pratt does, creating shared discovery anxiety.

Why it works (psychology): Narrative transportation. When we discover problems alongside characters rather than being told about them, we experience their emotional state. Shared discovery creates empathy. Empathy creates engagement.

Information timing is the mechanism. If the ad opened with "there's a $6.2 million banana," we'd be watching Hemsworth eat it with detachment. By revealing the value after he's already eaten it, we feel the panic with Pratt.

This is the difference between watching a horror movie where you know the killer's location (suspense) versus not knowing (surprise). Both work, but surprise + discovery creates stronger memory encoding.

How to steal it:

Restructure your content to delay the "why this matters" reveal. Show the action first, explain the significance second. Demo the feature, then reveal what it prevented. Show the before/after, then explain the mechanism.

For case studies: Don't lead with "Client X increased revenue 300%." Show the client's pre-solution struggles, the implementation, the result, THEN reveal the 300% metric. Numbers hit harder after emotional investment.

For social content: Film the process of discovering something, not the presentation of what you discovered. People watch "I found this weird tool" differently than "Here's why this tool is great." One's a journey, one's a lecture.

Landing page structure: Testimonial videos should show the customer's problem-solving process with your product, ending with "and that saved us $50K." Not "I saved $50K with this product because…"

What they did: Hired Matthew Vaughn (Kingsman director), shot everything with practical effects, zero CGI or AI generation. For a product about AI, they deliberately avoided AI in the creative execution.

Why it works (psychology): Production quality signals product quality. Your brain makes assumptions about price point and reliability based on execution. Ray-Ban Meta glasses cost $299-$379. Cheap production would have signaled cheap product.

The practical effects choice is strategic irony. In early 2025, audiences are drowning in AI-generated content. Real smoke, real stunt driving, real banana-eating becomes a trust signal. "We make AI products but we're not cutting corners with AI in our marketing."

How to steal it:

Match production quality to price positioning, not budget reality. A $5,000/month SaaS tool should have cleaner video production than a $50/month tool. That doesn't mean expensive—it means intentional. Lighting, framing, audio quality matter more than camera cost.

Budget-conscious execution: iPhone 15 Pro with good lighting beats RED camera with bad lighting. Invest time in pre-production (shot list, lighting setup, audio testing) over expensive equipment.

The practical effects lesson for smaller teams: Show real results, not rendered graphics. B2B SaaS demos should show actual software interfaces with real data, not animated "concept" versions. Service businesses should show actual work being done, not stock footage of "teamwork."

When to go lo-fi deliberately: If your brand is about accessibility, raw production can signal authenticity. Recording on Zoom, visible home office, casual framing—these work if your positioning is "we're marketers like you, not a giant agency." Production style is a positioning choice.

Two principles drove everything in this ad:

Principle 1: Humor as Educational Trojan Horse

People don't resist entertainment. They resist being "taught to." By making the banana panic the A-plot and the product demo the B-plot, Meta taught usage without triggering resistance.

You laugh at Hemsworth's shock, remember the $6.2 million detail, and absorb "Hey Meta identifies objects" as background information. The education happens while you're entertained, not after.

Principle 2: Multi-Layer Social Proof

Three celebrity tiers = three audience segments. Marvel fans came for Pratt/Hemsworth. Kardashian viewers came for Jenner. Tech enthusiasts came because it's Meta. Each celebrity brought their fanbase, creating earned media across different communities.

The structure: Show your product solving a high-stakes mistake in a humorous way. Film a "day in the life" of a customer where your product prevents escalating disasters. Let the humor come from the situation, not forced comedy.

The celebrity equivalent: Use industry-recognized voices (podcast hosts, thought leaders, LinkedIn creators) who already collaborate together. Don't hire them individually—hire their existing dynamic.

The three-act demo: Build your explainer video with three distinct use cases showing different features, but tell them as one continuous story of increasing stakes.

Budget: $500-$5,000 for production if using micro-influencers, or $0 if using customer stories with good smartphones.

The structure: Create content around "expensive mistakes" your service prevents. HVAC company? Film "what happens when you skip the annual inspection" with escalating disasters (slightly comedic, mostly real). Each disaster is a different service you offer.

The celebrity equivalent: Partner with local business owners who are already friends. The restaurant owner and the boutique owner who both use your accounting services. Film them talking about tax season disasters they avoided.

The three-act demo: Show the same service solving different problems. Same HVAC company, but Act 1 is comfort (temperature), Act 2 is air quality, Act 3 is energy bills. Different value props, same offering.

Budget: $0-$500 (smartphone, basic editing)

The structure: Build social content around cultural moments that relate to your product category. Trending TikTok sound about busy mornings? Show your coffee maker in that context. Viral joke about gift-giving fails? Show your product as the solution.

The celebrity equivalent: Micro-influencers who already create together, not solo partnership deals. The two friends who do "trying weird products" videos. Send them both your product and let them create together.

The three-act demo: Product reveals on social should show: (1) Problem without product, (2) First use/reaction, (3) Unexpected use case they discovered. Three value props, one story arc.

Budget: $50-$2,000 for micro-influencer gifting/partnerships

The structure: Create case study content that treats the client's challenge like a thriller. Problem = inciting incident. Research/strategy = investigation. Implementation = climax. Results = resolution. Make solving boring B2B problems feel like solving a mystery.

The celebrity equivalent: Co-create content with complementary service providers. The SEO expert and the conversion rate optimizer doing a joint analysis of a client's funnel. Let the dynamic between specialties create the content.

The three-act demo: Portfolio pieces should show three distinct capabilities through one client story. Strategy, execution, optimization—but told as continuous narrative of escalation and resolution.

Budget: $0-$1,000 (just time and editing)

Mistake #1: Copying the absurdism without the cultural reference

The banana worked because it was already internet-famous. Random weirdness for weirdness' sake just confuses people.

You can't manufacture a cultural moment - you can only hijack existing ones. If you try to make up an absurd scenario with no cultural grounding, people tune out because there's no reference point to grab onto.

The fix: Track what your target audience is already discussing. Use listening tools (SparkToro for audience research, Reddit keyword tracking, LinkedIn hashtag monitoring). When you see conversation momentum building around a topic, that's your signal to create content intersecting with it.

Mistake #2: Hiring celebrities without chemistry

Star power without relationship = expensive talking heads. Two influencers who've never met will feel like two separate ads spliced together.

The Marvel connection between Pratt and Hemsworth did more work than their individual follower counts. Chemistry is the multiplier.

The fix: Research who already collaborates in your space. Which podcasters frequently guest on each other's shows? Which LinkedIn creators tag each other in comments? Which micro-influencers already make duet content? Hire the existing relationships, not individual reach.

Mistake #3: Making the product demonstration too obvious

The glasses enabled the story - they weren't the story. Product should be the mechanism, not the message.

When your demo feels like a demo, people check out. When your demo feels like plot development, people stay engaged.

The fix: Ask "what problem does my product solve?" Then film someone solving that problem, with your product as the tool they use. The product should feel inevitable to the solution, not forced into it.

Mistake #4: Skipping the practical effects (for what you can control)

AI-generated or cheap CGI would have destroyed the credibility of an AI product. The irony was intentional.

In early 2025, showing something real is a differentiator. If you're claiming your product delivers real results, your creative should show real execution.

The fix: Whatever's within your control, make it real. Can't afford real stunt driving? Don't fake it with CGI - do a smaller stunt practically. Can't afford actors? Use real customers showing real results. Constraint breeds creativity if you commit to authentic execution.

Mistake #5: Missing the platform-timing window

Super Bowl = peak attention moment. But the strategy works on smaller platforms with timing precision.

Posting generic content on Tuesday afternoon reaches no one. Posting culturally relevant content when conversation is peaking reaches everyone.

The fix: Use Google Trends to identify when topics are accelerating (the slope matters more than the peak). Create content in the acceleration phase, not after peak. Set alerts for keywords related to your space, create rapid-response content when trends emerge.

by CH
for the AdAI Ed. Team

We have moved comments to LinkedIn! 👈 This platform has its limits for communication, so click the article link below to comment, talk, like, or repost to colleagues.