Case study · Lemmings
Data-Driven Ad Creatives
Made ad creatives for Lemmings, iterating until the best-performing one had the player failing.
The game
Lemmings is Exient’s mobile remake of the 1991 puzzle classic, on Android and iOS, and has since passed 10 million downloads on Google Play.
- Role
- User Acquisition Artist
- Date
- July 2018 – December 2020
- Skills
- Storytelling
- Video Editing
- Brand Identity
- Visual Direction
- Content Ideation
- User Experience
- Tools
- Premiere
- After Effects
- Photoshop
The challenge
I made the adverts that brought new players to Lemmings: social creatives for Facebook and Instagram, and interstitials that ran inside other games. Each had three seconds to earn an install. I worked in a loop with the analysts, shipping variants and iterating on what converted. The winning format wasn’t one I’d have predicted.
A Year of Showing the Game
The role covered more than one title. For the first few months I was cutting creatives for Dancing with the Stars alongside the Lemmings work. Both campaigns assumed the same thing: show the product, assume it sells.
Lemmings got the same. Character parades, festive scenes, a level with a Go! prompt and a logo. All built on a guess: if people saw the game clearly enough, they’d want it.
They converted, but barely. Every round came back from the analysts looking like the last. The ceiling was obvious. Why it was there wasn’t.
What Actually Converted
In August 2019 I tried asking instead of telling. The ad laid out a level with three solutions: A, B, or C. No answer.
A few days later I made the one that changed things. An animated finger places tools ahead of the Lemmings, moving through a level: umbrella here, dig a shaft there. It reads like a rescue until the last Lemming walks into the trap anyway. Everything’s placed right. They still die.
It doesn’t ask you to admire anything. It makes you want to fix it. The gameplay’s real. Those are the tools you’d actually use in the real game. Only the ending’s staged.
It won because of what came next. Within a few months I was rebuilding that same format as a Christmas version, and it kept earning its slot. In January we tried the inverse: a rescue that actually works. The numbers stayed with the analysts, but the decision didn’t: we kept making the fail version, not the win one. Budget follows what converts.
What I Took From It
It changed what I ask. Used to be what looks good in three seconds. Now it’s what does the person watching want to do about it.
That applies everywhere. A good tutorial lets you fail and you remember it. A near-miss teaches you more than watching someone else do it right.