Meme Mapping: Learn to Run Better Creative Tests by Reverse-Engineering Hollywood

Advertisers are losing control. As platforms like Meta and Google automate most targeting, attribution, and optimization decisions with machine learning, soon the last growth lever left will be creative testing: Which concepts, copy, colors and artwork drive the best results?

Most every marketer is familiar with A/B testing, but few have a consistent framework for prioritizing what to test. As experiment backlogs spiral out of control, the process devolves to just do whatever the CEO says. 

But you can avoid this fate by breaking creative ideas down to their component parts, or “memes,” and systematically remixing creative elements to drive growth.

Find the right combination of memes – visual cues, catchy tunes, key phrases – and you win.

In this post, I’ll make the case for the rising importance of creative testing. Then dive into how Hollywood uses memes to help people find the right movie to watch, to introduce you to the concept of memetics. Finally we’ll run through a step-by-step example of how to use Meme Mapping to get a more consistent return on investment from experimentation.



About the Author:

Mike was a co-founder at Ladder, a 50-person growth marketing agency with $50m spent and 8,000 experiments run, for clients like Booking.com, Monzo Bank, and Facebook Workplace. He is currently writing a book on Marketing Memetics.


“Creative” is the Last Performance Battleground

When I first started doing digital marketing in June 2010, most of the job was done in spreadsheets: whole rooms full of analysts creating pivot tables, calculating KPIs, and optimizing bids for Google Adwords campaigns.

The startup I worked for, Efficient Frontier, was founded by a coworker of Jeff Bezos’ at hedge fund D.E. Shaw, who sought to apply algorithmic trading algorithms to optimize digital ad campaigns. It was later acquired by Adobe for around $400 million and lives on as Adobe Media Optimizer, though these days it’s more likely that Google (and now Meta) are doing the optimization for you.

It’s rare today to find a media buyer that advocates for manual bidding or budget optimization in Excel, or even a 3rd-party bid optimization platform; Meta and Google’s solutions are too good (and free!):

  • Lookalike targeting audiences automatically find the most likely users to convert.

  • Campaigns automatically optimize to a specific cost per action or budget cap.

  • Budgets are automatically re-allocated towards your best performing ads.

Most of the work I did manually a decade ago is now handled by Meta and Google’s machine learning algorithms.

Not that creative never mattered: A Nielson study in 2017 found creative contributed 47% of campaign performance, with a further 15% from brand. But as targeting and optimization get automated, soon creative will be the only growth lever left.

In a world where creative assets are all Meta and Google want us to upload, creative becomes the last battleground for marketers who need an edge against the competition.

(I saw this coming a while ago, so before I left in March 2020 I re-oriented my agency fully towards data-driven creative testing — which it’s still doing successfully today.)

Artificial intelligence algorithms like GPT-3 and DALL-E are coming for us there too, with early results showing that machine learning can generate mind-blowingly cogent language and imagery. However the ability to instantly produce any creative you can imagine — for next to zero cost — only makes creative testing strategy even more important.

When you can create anything, the bottleneck becomes deciding what to create.

How the Right Creative Attracts The Right Customers

As Steve Jobs famously said: “People DO judge a book by its cover.”

Imagine you’re taking your significant other to the movies for Valentine’s Day. You’re hoping to pick something romantic, and a movie poster catches your eye.

It has Ryan Reynolds — a comedic actor you both like. He’s smiling in the poster, embraced by an attractive actress who has her head resting on his shoulder. The background of the poster looks like fireworks, and the tagline is “True Love Never Dies.”

This looks like the ticket.

Credit: Marvel/Fox via The Verge

If you’re familiar with “Deadpool,” you’re already in on the joke. It’s not a romantic comedy at all: It’s a superhero movie, and definitely not a family-friendly one.

In the very first scene, one of the bad guys gets skewered with dual samurai swords by our hero, who breaks the fourth wall to acknowledge the on-screen carnage: “Some of the best love stories start with a murder.”

The misleading movie poster was a prank to generate social media buzz for the movie, but it serves to highlight something important about consumer behavior: People look for sensory cues to determine if they’re buying the right product, and to regulate how they feel post-purchase.

Pull quote: "People look for sensory cues to determine if they're buying the right product, and to regulate how they feel post-purchase."

If a business transmits the wrong sensory cues, it could attract the wrong customers. When marketing sets misleading expectations, customers might feel disappointed by their purchase, leading to poor reviews (and a decreased conversion rate) and impaired word of mouth (the primary factor behind 20–50% of all purchasing decisions).

And it all happens quickly: Consumers form design opinions in as little as 17 milliseconds, so first impressions matter. 

What “Meme” Really Means

You probably associate the term “meme” with internet memes: those image macros with white text superimposed on a funny image that go viral on social media. But that’s actually a hijacking of the term, which was coined by a British evolutionary biologist, Richard Dawkins, in his 1976 bestseller “The Selfish Gene”:

“Examples of memes are tunes, ideas, catchphrases, clothes fashions, ways of making pots or of building arches. Just as genes propagate themselves in the gene pool by leaping from body to body via sperms or eggs, so memes propagate themselves in the meme pool by leaping from brain to brain via a process which, in the broad sense, can be called imitation.” 

The word meme comes from mimeme, the ancient Greek word for “imitated thing.” Dawkins meant it as an analogous to biological genes, a way to describe how our culture evolves in ways not explained by our DNA.

Pull quote: "Essentially, memes are just associations in our brain — and are often used, unconsciously, by our consumer brains to direct us towards the 'right' products to buy."

Essentially, memes are just associations in our brain — and are often used, unconsciously, by our consumer brains to direct us towards the “right” products to buy.

Unlike genes which randomly mutate, memes can be edited by human creativity.

Viewed through this lens, marketers are memeticists — the meme equivalent of geneticists — engineering the right combination of memes that resonate with their target audience.

When a marketer gets this right, they increase their customer’s cognitive fluency — the human tendency to prefer things that are not only familiar, but also easy to understand — which is scientifically proven to improve conversion rate.

Reverse-Engineering Hollywood: How Movie Posters Use Memes to Build Trust and Drive Revenue

It can be hard to grasp just how prevalently memes are used to communicate what type of movie you’re about to see.

It’s a critical trick Hollywood marketers employ to keep viewers coming back satisfied. As Walt Hickey, a pop culture expert at FiveThirtyEight, told ABC News: "Risk aversion has really paid off, and studios and production companies have noticed this.”

It can cost hundreds of millions of dollars to make a modern movie, and they need to convince millions of consumers to spend 2+ hours of their lives and $10+ (with popcorn) to see it. Studies show that in conditions of uncertainty, rather than optimize for maximum utility (in the case of a movie: utility = entertainment or enjoyment), consumers actually act to minimize regret.

Our brains evolved over millions of years, so it’s no surprise that the same familiar combinations of memes show up again and again, to tell us what to expect. 

As a consumer, marketer, or entrepreneur, you might rarely register what a red dress, yellow background, or blue tint means. But your brain is a meme-recognition machine, so it just happens naturally. It’s only when you zoom out and observe meme usage across multiple titles that patterns become clear.

Let’s examine five tropes from Hollywood marketing, accompanied by movie poster composite images shared by creative director Lee Stafford on Twitter, and movie distributor Christophe Courtois on his website.

1. Blue and Orange = Action

Why is everything in Hollywood blue and orange? In the days before digitization, sepia was used to preserve old film. When digital editing made color correction easier, everything migrated to blue. Most flesh tones (those featured in Hollywood anyway) are in the orange range, and blue is the opposite end of the color spectrum, which according to color theory makes people “pop” from the background. This combination fits the bill for action because this particular pair of complementary colors are associated with strongly opposing concepts — fire and ice, earth and sky, land and sea, dawn and dusk, natural and scientific.

2. Red Dress = Seduction

Red is the color of Valentine’s Day, and women have used red lipstick since at least 10,000 BC, so this isn’t an association Hollywood invented. The “Red Dress Effect” is hypothesized to be linked to biology: Higher oxygenation in blood is associated with higher levels of estrogen and fertility in women. The color red has also often been associated with impulse purchases, for example data from ebay shows a red background triggers more aggressive bidding in auctions.

3. Eye = Anxiety

As social animals, we’re always vying for attention. Your eyes act as a spotlight for your attention. Even babies use the eye information of others and follow their gaze. In positive encounters, eye contact enhances our self-awareness and elicits prosocial behavior. However in a negative context, eye contact can cause embarrassment, shame and guilt; we can punish breaking of social norms with a glance. Amongst animals, staring is interpreted as a display of dominance, which kicks in the fight or flight response. This is why holding direct eye contact for longer than 2–5 seconds is unnerving, and can overstimulate the brain, leading to anxiety.

4. Yellow Background = Indie Film

Why do all indie films use the color yellow? Could it be that yellow pencils are awarded for creative advertising? Or perhaps they’re riffing on the name of the Sundance Film Festival, where being awarded is a prerequisite to breakout indie success. It’s more likely due to yellow’s position as the most visible color on the spectrum, which is why it’s used for taxis and warning signs. Small films with limited budgets must grab attention, but also act as a clear warning for action fans who may be disappointed with the lack of explosions. Interestingly, studies show that yellow is one of the least popular colors for both men and women, which explains why that part of the spectrum is left uncontested.

5. Black, Blue, and Fiery = Warner Brothers

At the opposite end of the spectrum to indie films are the blockbusters from Warner Brothers. Despite vastly different content, from Batman to The Hobbit, Mad Max and Harry Potter, all of the posters from this studio are eerily similar. Predominantly black and blue, there’s always a source of light (usually fire) to bring in the complementary orange to complete the set. By consistently repeating the same memes, they’re reinforcing their distinctive brand assets, so the audience can instantly (and subconsciously) recognize the studio’s next movie.

IPSOS found the highest performing creative employed distinctive brand assets 34% more than average. Consistency breeds familiarity. Familiarity breeds trust. Trust makes purchase decisions easy.

“Meme Mapping” Isn’t Just for Hollywood

Movie posters all look alike because directors consciously or subconsciously imitate previous works in their field. However tech companies like Netflix and Amazon have made great gains by reverse-engineering Hollywood’s memes to produce content that is destined to succeed.

So when it comes to crafting creative for your marketing: You could guess and get lucky, or you can borrow meme mapping from Hollywood to systematically identify recurring patterns in competitive or complementary products. 

Meme mapping will help you discover the commonalities — across concept, image and language — that are actively employed and recognized by the audience in your category.

One example you’ll see everywhere once it’s pointed out, is the use of “Corporate Memphis” style on websites in the B2B software space, including brands such as Intercom, Slack, and LinkedIn. For this industry demonstrating trust is important, and one (subsconscious) way to do so, was to commission custom artwork for your website (to show you can afford it), as opposed to using free stock photos. 

Now this style has become so prevalent, it has lost its initial signaling value, but at this point it’s so ubiquitous any brand that not using it would “feel wrong” and therefore “risky” to mainstream corporate buyers. You don’t have to go with Corporate Memphis for your new B2B software brand, but it’s important to be aware of the meme, so you can make the strategic decision between going with the default, or attempting to start a new trend. .  

Meme Mapping doesn’t have to be just about design, it applies to ad copy as well. Some words or phrases resonate significantly more than others, and knowing these “magic words” is one of the things that distinguishes a great copywriter from an average one. For example the meme “X made simple” has been used successfully across a number of industries:

  • Low cost airlines (Ryanair: “Low fares. Made Simple”)

  • Utility price comparison (uSwitch: “Switching. Made Simple”)

  • Security software (Aluria Software: “Security Made Simple”)

  • Wireless technology (Linx Technologies: “Wireless Made Simple”)

  • Banking (National City Bank: “Banking Made Simple”)

Of course, once everyone in an industry A/B tests their way to knowing what works, there’s a danger they’ll all converge on the same words and phrases, which can create an opportunity for differentiation; to zig when others zag. Meme Mapping can help you spot these convergences so you can either go with what works, or decide to distinguish yourself from the competition. 

4 Steps to “Meme Mapping” Creative to Grow Your Business

The goal of “meme mapping” is to understand what memes are associated with your chosen product category, so you can then meet — or challenge — consumer expectations.

The insights you gain from meme mapping are a great source of new hypotheses for conversion rate testing, or optimizing your ad campaigns with creative testing. In all scenarios selecting memes systematically will lead to more consistent results from experimentation. 

The example I’ll use is finding the right listing image for a rental property (just to make it simple to illustrate with public data). However the same process can be used for creative testing on social ads, conversion optimization on websites, or really any scenario where you need to find the right creative asset to drive performance. 

1. Collect examples and build a swipe file

Back in the 1960s, copywriters would clip copies of ads they liked and keep them in a file for later when they needed inspiration. It’s far easier today with modern note-taking apps and cloud storage: Rather than clipping magazines, we can now take a screenshot and store it along with our notes in a productivity app like Notion for free.

The aim is to collect examples of competitive marketing assets, so you can later understand what patterns of memes are being used. This is usually an “always-on” task – as you go about your day you’ll collect inspiring examples of designs, listings, or ads that resonate – but can be done more systematically ahead of producing new creative or undergoing a rebranding exercise.

Choose a source of competitor information, for example the Facebook Ad Library, marketplace listings, or just the content of your competitor’s homepages. 

Example: If you were promoting a rental property in Miami, you might collect screenshots of popular listings on Airbnb in a Notion database.

2. Tag each example with relevant “memes”

Use inductive coding, a data analysis process where you interpret what common themes are shared across a set of assets, to tag memes from the text and images contained in your swipe file. Start reviewing the assets in your swipe file and tagging them with any patterns that you see. You can start high level, and drill down into more detail later as you spot more interesting patterns.

If that seems like a daunting task because you have built up a large list of assets, you can break up the task by selecting say 10% of your assets, and creating codes that broadly cover the sample. Select the next 10% and apply your codes, adding new ones, drilling down or consolidating where useful.

It's important that labels aren’t too sparse (appearing less than 3 times) but also aren't so broad that they become meaningless (covering more than 80% of the assets). You can use this method to tag objects in images, or phrases used in text, to see what common memes emerge.

By deconstructing what’s in each asset in your swipe file, you can spot patterns across these assets to give you clues as to what memes are working.

Example: We have tagged each listing with what the main image shows, either a bed, a living room or the beach, so we can see what most people are using.

3. Identifying Patterns

In this step, you’re looking for patterns that emerge that weren’t previously on your radar, or to confirm hypotheses you already suspected.

For most public data sources, there aren’t publicly available performance figures. For example, the Facebook Ad Library only shares performance data for political ads, and we’re unlikely to know how our competitor’s web pages convert.

Instead you should count what memes appear most or least often, or what has been running the longest. Frequency often serves as a good enough proxy for success, because when marketers actively manage their campaigns, elements that don’t add to performance get dropped.

However if you do have any form of performance data, you should try to incorporate it where possible, because we can’t always rely on other marketers to do their jobs and drop poor performing memes.

For example, on Airbnb a meme might be widely used, but if all the listings using it have a small number of reviews, that’s an indicator it’s not good for performance. In this case we don’t know how many bookings the listings have, but we can use the number of reviews as a “good enough” proxy for performance.

When you do have performance data, you can use advanced techniques such as Message Mix Modeling. This technique uses linear regression analysis to estimate the relative impact of each meme on performance.

Example: We found that listings that show the living room have half as many reviews (a proxy for number of sales) on average versus those that show a bed or the beach, so we’ll avoid using this meme.

4. Creative Testing

With these memetic patterns identified, you now have a map of the territory in which you can operate as a brand. You should first choose what memes you’re simply going to copy to establish familiarity, but also think about innovation that can help you differentiate your product from competitors.

  • Copy: Users expect certain memes in a category to orient themselves, and the default should be to give them what they want. If there is a commonly accepted default, and you don’t have a strong opinion on it, copy. 

  • Innovate: If you copied every meme from competitors, there’d be no reason to choose your product or service! When there’s a meme you can use, that you know resonates with the audience, and that other competitors can’t easily copy, then there’s space to innovate. .

How do we ultimately decide when to copy and when to innovate? It really depends on your vision: innovate on what matters to your strategy, copy anything that’s not an integral part of it. Let’s take one more lesson from Hollywood: “The comfort of familiarity should not limit the filmmaker’s vision, but should act as a starting point to get the consumer’s attention and play to their risk aversion so that they sit and watch the film, instead of passing it by.” – Jason Cherubini, Academic, Entrepreneur & Film Producer.

Now you’ve selected the memes that matter to you. What next?

Take ideas from your analysis and build on them with meme-jacking: starting with a proven concept rather than the dreaded blank canvas when writing marketing content or making a new design. Even when copying, you don’t have to copy exactly: extract the underlying theme of what’s working for competitors so that it’s familiar to consumers, but give it your own unique twist.

This exercise should generate more ideas than you can possibly test, so the biggest job is actually prioritization. 

Each idea you generate throughout this process can be consistently tested as concepts, themes, or variants, until you find a winning combination.

  • Concept: High-level attributes (product features, jobs to be done, value propositions)

  • Theme: Implementations of a Concept (subject of image, artistic styles, adcopy routes)

  • Variant: Variations within a Theme (border colors, similar photo with different people, phrasing of words)

This Concept-Theme-Variant framework comes from Eric Suefert of MobileDevMemo, who provides a great example of testing Ogres vs Dragons for advertising a mobile game.

The delineation between these categories can be fuzzy and unique to your brand, but the purpose is to separate your testing across different magnitudes of change: Conceptual level changes should happen rarely and move slowly — these are the bigger strategic choices that should take a lot of evidence to shift. But there are many thematic implementations within each concept to iterate on. And each theme, in turn, has multiple small variation changes that can be made.

If you’re an established brand, you should already have a good idea of what Concepts resonate with customers, and your Themes might be already dictated by your brand style guide. However for smaller start-ups, or older brands in need of a refresh, you’ll want to begin  at the Concept level. Then work your way down to Themes and Variants as the memes prove themselves.

You don’t always have to run a test to validate or eliminate a creative route — sometimes you can select or rule out options based on intuition and the analysis you did on your swipe file.

Example: The research showed that “bedroom” and “beach” both had potential to be successful, so we need to A/B test them to measure their relative performance. We continue testing Themes and then Variants as we find winners. 

If you find “beach” works significantly better, you can try different themes within that concept: Should we have just a shot of the sand, or people playing volleyball?

When you find a winning theme, double down: There are thousands of potential variations of people playing beach volleyball that you can try. Eventually any creative application can saturate from ”creative fatigue” as an audience gets tired of seeing it — so you start again from the top at the Concept level, or go back to the research phase to map more memes and generate more ideas for testing.

Meme-Mapping Gives You a Competitive Advantage

Memetics works subconsciously most of the time, so you often have no idea just how prevalent a meme is in your industry before you start diligently cataloging examples and see the patterns emerge.

Because you’re doing the work to map a broad selection of memes, you’re casting a wider net than competitors relying on experience alone — and you’re more likely to find obscure memes that give you a performance advantage. As advertising great John Hegarty says: “The originality of an idea depends on the obscurity of sources.” ...or was it Albert Einstein that said it? Nolan Bushnell? C. E. M. Joad?

Every test is related back to the meme map, so know what to test next if this wins or loses. Ultimately this approach translates to more consistent results and a greater ROI from your testing process.