Unicorn ads

Quality over quantity is a false choice. Quantity equals quality. If you focus on quality, instead of quantity, you’ll get neither. If you focus on quantity, you get both. On the first day of class, a professor at the University of Florida divided his film photography students into two groups. The first group would be graded on the quantity of work produced: 100 photos gets an A, 90 photos a B, and so on. The second would be graded on the excellence of their work: they need only to submit one image, but it had to be near perfect. At the end of the term, he was surprised to find that all the best photos were produced by the quantity group. They were busy taking photos, experimenting with different methods, and learning from their mistakes, while the ‘quality’ group speculated and theorized, with little to show for their efforts but one mediocre photo.

All human creative pursuits are dominated by the Pareto distribution, stating that 80% of outcomes are due to 20% of causes. Movies. Books. Music. Art. Technology. The ratio isn’t always 80:20, but the vast majority of value is always generated by a tiny percentage of the activities. It’s rarely possible to predict ahead of time what will work. When you can’t predict outcomes, the right strategy is to try as many things as possible without expending all of your resources, then double down when you find something that works. It’s unicorn hunting, not horse grooming. The ends justify the means, because one great ad is enough to ‘return the portfolio’ – pay back everything you invested in the losing variations – and all sins are forgiven when you find one. If you fail to find a unicorn ad, no amount of ‘doing the right thing’ will save you.

Analysis shows there’s only a 5-15% probability that a new ad will outperform the previous best creative in your account. However when you do find a winner they tend to outperform by 20- 500%, which makes up for all the budget wasted. It can take 2-3 weeks for a winning ad combination to reach scale. After 4-6 weeks even winning combinations reach their peak due to ad fatigue – users get tired of seeing it in their feed – and by weeks 10-12 the next unicorn hunt should be well under way or performance will significantly decline. This shelf life can be extended by executing lots of small variations of the same winning ad, but eventually that strategy gets exhausted too.

While we’re fighting the Pareto distribution to find a successful ad, we can use it to our advantage to bring down the cost of testing new ones. With 20% of the work of creating the final ad, we can make a ‘good enough’ version that gets us to 80% of the final performance figure. Closing the gap on the final 20% of performance will take a lot more work, but we only need to do that if the rough version performs. In this manner we can test 5x as many ad concepts as if we created a fully polished version for each idea.

I recommend testing from the concept level down: a good idea will work regardless of implementation details, and it’s better to iterate through lots of concepts then waste cycles on small variations. Once a big winner has been found you can explore different themes within the concept, and then variations on winning themes to extend the shelf life of the asset. Take a portfolio approach to hedge your bets: even when performance on your core ads is good, you should always be allocating 5-15% of your budget towards testing new ideas.

We don’t have to search the whole grid. Copying something that worked before is a good place to start, because it’s likely to work again. This is why most competitors in an industry end up looking the same: none of them want to take the performance hit of abandoning what works. To avoid this trap, look for things that work in a different industry, but haven’t yet been tried in yours, because then you avoid looking the same as the competition.

About the Book 📕

← Back to Essays 📝