This post was originally published on saxifrage.xyz before relocating here.

In The Selfish Gene, Dawkins coined the word “meme”, to mean a “unit of cultural information spread by imitation”. An idea gets stuck in your head, sometimes remixed with a random mutation, and then you spread it to others – think of it as a ‘virus of the mind’.

Unlike genes, memes replicate by jumping from brain to brain; but ‘survival of the fittest’ still applies: if an idea doesn’t reproduce, it dies out (gets forgotten). In marketing we’re in the business of spreading ideas, so it’s important to know how the mechanism works.

Every marketing campaign can be broken down into its component parts, or memes. What adcopy, creative or targeting techniques did they use? If you do this enough, you’ll start to notice patterns: tactics that work again and again, across multiple campaigns.

Most marketers try what worked for them before, or what they heard worked elsewhere. They either get lucky, or they get it wrong. Natural selection happens over millennia: but you can cheat – systematically test your way to a winning combination of memes.

The good news: although a lot has changed, our brains still operate the same as they did in the Stone Age, implying the list of dominant memes is finite, and it’s possible to learn them all. For that you need a system – use mine and save yourself 10 years and $50m.

Here’s how it works: 1) collect a swipefile of campaigns, 2) break them down into their memes, 3) identify what patterns work best, 4) always be testing new combinations. In short: do more of what works, less of what doesn’t... simple, but not easy.

Meme Engineering

In memetics terms, a successful brand is a “memeplex” – a collection of memes that have evolved into a symbiotic relationship. The competitors in your space have (intentionally or accidentally) found a combination of memes that successfully replicate in customer’s minds.

You need to learn what works, then choose where you’ll differentiate. Customers are already conditioned to expect certain memes in a category, so you can’t innovate on everything. Instead the goal is to find a handful of distinctive, memorable assets that you can own.

1) Swipefile

Collect examples of competitor campaigns, and save them centrally to a folder. The larger the swipe file you build, the better you will be able to spot patterns in the designs, writing and placements used. Include campaigns from other industries you think might work in yours.

2) Memes

Review the assets in your Swipefile and break each down to its component parts. Tag any creative, copy and targeting tactics that you identify. Build a reference dictionary of these tags so you can collaborate with your team to correctly categorize campaigns.

3) Patterns

Like genes, successful memes replicate – are any patterns showing up consistently over time or across the industry? Formalize the definitions and categories of the tactics you’re seeing. Quantify which memes are more popular than others.

4) Testing

Develop a testing strategy from your analysis of what memes are working. What memes will you copy to meet expectations, and where will you differentiate to win? Systematically test memes going from high-level concepts, down to smaller variations on successful themes.

Modern Memetics

None of this is new! Roll back to the 1960s and you’d find Marketers keeping ‘Swipe files’ of successful campaigns, studying trends in popular culture with Semiotics, and Ogilvy saying “Never stop testing, and your advertising will never stop improving”.

What’s new is our ability to collect, analyze and categorize campaign data at scale. Advances in machine learning let us automate identifying creative memes, and we can test new ideas with instant feedback on digital channels, and eventually in VR.

Download every competitor campaign with Web scraping and APIs. Automatically tag assets with OCR, image recognition and NLP. Identify patterns then systematically test and learn what works. Good luck: brain space is limited and there’s a war for attention.