If You Want to Design Effective Ads, Learn to Play Battleships

This article was co-written with 

You've probably played Battleships. You know, the game where two players place a set of battleships on a grid, and take it in turns to sink one another's ships by guessing their location.

For a game that ostensibly has nothing to do with ads or marketing, it can teach us a hell of a lot about both of them. To see why, let's consider a simplified game of Battleships.

Hunting Battleships

Let's say you're playing a game of Battleships with a 10x10 grid, and (to keep things simple) we'll say that each player has one ship that's 5 grid squares long. You and the other player place your ship on the grid, and the game begins.

You go first. You reckon that your first guess doesn't really matter much so you choose it at random: D3.

The guns fire and the shells start ringing, but it's a miss. No hit. Your opponent plays their turn, they miss too.

It's now your second go, and you have to choose where to take your next shot. You've got one key piece of information from round 1, that there's no ship in D3. But why exactly is this useful?

It's useful because it tells us that the opponent's ship is less likely to be in one of the grid squares immediately adjacent to D3, than it is to be in a square far away.

In case it's not obvious why, consider the range possible locations your opponent's ship could be in before and after you found out that it wasn't in D3. At the start of the game, their ship could've been anywhere on the board. Once you learned it didn't have any part of it in D3, there were fewer possible locations near D3 that it could have been in.

For example, you now know that it couldn't have been in grid squares D1 to D5, because if it would then your shot in D3 would've hit it.

There are a number of different possibilities in which the ship is in D3 that have been ruled out by the outcome of your first turn. Because we know the ship is five squares long, this decreases the probability that the ship is in any of the surrounding squares too.

So, what do you do?

If you're clever about it, you won't try D4 on your second go. Or D2, C3, or E3.

It's not impossible that the opponent's ship could be in these squares, but it's less likely to be in any of them now you know it's definitely not in D3.

The smart second move would be to try somewhere on the other side of the board, maybe an F6 for instance. Because this is sufficiently far away from D3, knowing that your opponent's ship is isn't in D3 doesn't decrease the likelihood that their ship is in F6.

And if their ship isn't in F6? Maybe on your third go you try somewhere you haven't explored yet; perhaps the top left or the bottom right of the board.

This process will keep going, where you continue to look for parts of the board you haven't tested yet, until you get a hit. Once you get a hit, you know that their ship will be in at least one of the immediately adjacent squares to where you got the hit.

At this point, your strategy switches. Instead of looking for unexplored bits of the board, you look at the squares right next to where you've just explored. You keep doing this until you've found the grid squares that their ship occupies, and the game is over.

Where do ads come in?

Often when it comes to designing ads, newer brands are guilty of putting out a ton of ads all with only very minor differences. The idea seems to make sense; test small differences in the styling or the messaging on the ads, in order to learn what works best.

The problem though is that this approach only makes sense for well-established brands, who already know with a high degree of accuracy which of their ads resonate best. When you have heaps of historical data about what works, you can afford to go granular on what you test.

Newer brands, or brands launching new products, are staring at the equivalent of an empty Battleships board at the start of the game. They don't have data on what resonates; they haven't taken any shots yet.

Putting out batch after batch of creatives with a similar look and feel is like taking the opening moves A1, A2, A3, and so on.

You'd never do this in a game of Battleships. As we saw earlier, if your first shot misses you move to the other side of the board, and so on. And yet most brands do the complete opposite of this, churning out near-identical creatives in the hope that shifting one square at a time will get them their first hit.

Sure, this could work. You might find the best ad ever on your third narrow iteration, but it's unlikely. You're much more likely to find your best ad by casting your net wide initially, and only then narrowing down more and more on what works.

The space of possible ads is much more complex than a 2-dimensional Battleships grid. If you wanted to massively simplify it though you could imagine it as being 2-dimensional. You could label its axes as something like creative style and messaging style:

These labels are just examples, you can think of them however makes most sense to you.

What's important when you're producing creatives is to think how closely they all sit together in this space. If they're all very close together, you're unlikely to learn much from your test.

This is fine if you already know which area of the graph works for you; if you're already well-established and know exactly what resonates with your target audience.

Realistically though, this isn't true for most advertisers. Most advertisers should be casting their net much more broadly, and testing all around the space to find what works for them.

It's only by really testing all areas of the graph that you can have the confidence to really start narrowing down your testing. If your shot at D3 was a miss, you shouldn't try D4 until you've tried the rest of the board.

This piece was co-written with 

Thanks for reading

I'd love to hear your thoughts; come say hi to me on Twitter.

If you want to join 400 other growth marketers in hearing about when I post new stuff, drop your email below. No spam, I promise.

Nice one!
Oops! Something went wrong while submitting the form.