Don’t Shrink to Fit

"With the right targeting, this could be really effective"

This comment has festered. It came up in a meeting this week while reviewing some creative provided to us by YouTube as an example of best practice for a new format.

Now, to my eye - the creative was...not great.

Bland.

Nothing distinctive.

No meaningful or memorable story.

A mishmash of product demonstration shots.

Even now, I can't articulate anything more about the idea behind the asset. We tried to give it some post-rationalization, but the best we came up with was quite a stretch (that it acted almost as a well-produced product review/rating to give you an extra jolt of confidence to finally hit "buy").

The team's point was that media could make the creative better. That if you thin-slice the audience small enough, this ad was exactly what was needed to get someone to buy. That if you know enough about the audience and context through data signals, creative will breakthrough even more because it's just that relevant to that person, in that moment. It's an argument I've heard hundreds of times over the past decade from the reps at Meta, Google, or insert-any-digital-data-owning-media-platform-here. I've made it myself.

But that math doesn't math.

Creating relevance by tweaking the audience - e.g. restricting the media delivery to only show the creative to people who fit the bill of understanding/being likely to respond to the creative - operates with a false assumption. That only makes sense to do if the media costs hold steady as you scale down. That's not the world we work in. Targeted media costs more - the more restrictive you target, the more you're paying in marginal data costs. So while you may have a lower total cost of media to reach 100 people than 100,000, you're likely paying 10-15x more per person reached. So even if you increase your response rate (which works, up to a point - see below), you have a much smaller pool of people to respond in general, and the marginal cost of each response in terms of increased effectiveness and total profit returned becomes astronomical as you flood the segment.

Research published last year demonstrates that the response curve needed to make that make sense is just too steep - there's a natural limit to how much/how quickly people respond to advertising, no matter how relevant. Essentially, you need an exponential increase in response rates (e.g. post-view conversion to purchase) with every linear step increase in marginal CPM.

Just not possible in the real world.

So, relevance is a false promise? A windmill tilted at?

Not quite. You just have to pull the right lever. If you instead aim to increase relevance through creative that resonates deeply - even with only a small subset of the people you reach - you can achieve more significant and profitable growth in campaigns.

Call it the IYKYK strategy. Or my personal favorite "Coded Creativity."

And it works better, you'll still reach the inner circle of people who will react as you hope, and even more at the margins. The "wasted impressions" of people who don't fit the narrow target can potentially react at a lower rate, or even better, slowly become intrigued by the coded signals of the creative, thus joining and growing the in-group over time.

There are plenty of examples of brands doing this well (if only partially) today. The most common shortcuts to coded creativity: celebrity and influencer partnerships (think Ice Spice + Starry in the Super Bowl; Tag Team + GEICO), meaningful sponsorships (Ally Bank and USWNT/Women's Sports), dissonant timing or media placements (Change the Ref; IBM; IKEA; CK), and broadcasting a narrowcasted interest (McDonalds; AmEx).

So, let's follow the example and make small big, don't shrink-to-fit.