As global creative director at Google’s Unskippable Labs, I’m always seeking to uncover what works best when it comes to ad creative. And when it comes to machine learning, I see all kinds of possibilities for it in every report we write and every experiment we run.
But I see trepidation among creative teams too. As a creative, you may wonder: If machines’ capabilities are growing so fast, how soon until they’re writing my ad copy? It’s a valid question, but I believe the role of creatives has never been more important. Machine learning is meant to work with creativity, not to replace it.
When we harness machine learning’s ability, we’ll make better, more relevant, and more effective ads.
Think of it like this: Using machine learning is like having a billion interns working for you, not a single Einstein coming up with the perfect solution. You have to figure out how to use them, which requires assigning them tasks and translating their output into something useful. Without you, the interns would be lost.
Rather than belabour the industry debate of creativity versus technology, I think we creatives are better off figuring out how to guide the interns. When we harness machine learning’s ability, we’ll make better, more relevant, and more effective ads.
The path to better advertising is teaching the interns how to win
Imagine your interns are playing a game of Breakout, that foundational computer game in which the player controls a paddle to keep a bouncing ball in play, aiming to destroy that pesky layer of bricks at the top of the screen.
For machine learning to work, it needs to:
The first thing the interns need to learn is what it means to “win.” In Breakout, this means highlighting the score and telling the interns that higher is better.
The next thing they need to learn is how to win. As they keep playing, they’ll discover that repeatedly attacking one column until the ball “breaks out” into the space above the bricks is the most efficient strategy. This gives them a stable set of rules to play within.
Finally they need a big enough data set to emerge victorious. If you give them 10 minutes, they’ll probably lose, badly. If you give them six hours at it, they’ll suddenly be destroying that top score I was so proud of in sixth grade.
The problem with ad creative now is that the interns are confused
If the interns can eventually handle Breakout on their own, why can’t they help us make better ads on their own?
Because, unlike with Breakout, there’s no consistent way to measure a win in the creative world. We haven’t all agreed on what “winning” is when it comes to great ads — nevermind how to achieve that win. Some argue it’s great storytelling; others say it’s the mental availability of the viewer. The measures and models of attribution are muddy and inconsistent. This is confusing for the interns.
On top of that, the rules keep changing because culture keeps changing, and creative and culture are intertwined. To make an ad that’s a hit, you have to strike the right note in culture at the right time.
There’s a moment in culture when the choices we make have extraordinary power. Make that choice too soon (say, by putting a catchy but unknown song in your ad), and you’re obscure. Make that choice too late (say, by putting that song in your ad a year after it became a ubiquitous smash hit), and you’re a cliché. And you can accelerate from obscure to cliché in a weekend. The interns are trying to hit a moving target in high wind to find that sweet spot in between.
The world of audience signals: There’s hope for the interns yet
Fortunately this fast-moving world provides us with a ton of audience signals. As that data set gets richer, the value of the data becomes clearer. This is very exciting news for your interns and for you as a brand storyteller.
Imagine the possibilities when you put those signals into the interns’ hands, and ask them: What patterns can we see? What value do those patterns have? And how much is customised ad creative worth?
We’re the ones who need to teach the machines what to look for and figure out what to do with the answers.
These are the answers my team is looking for now. We recently did an experiment with six-second bumper ads for CoverGirl. We found that customisation can make more effective ads than a one-size-fits-all approach. But that starts to open a new set of problems: How many ads do you need to make? How many iterations provide how much value?
The interns can help us find those answers. It’s the perfect time for them to grind through signal data, surface the best types of signals, and give us insight to write great lines that will resonate with each of many different audiences. But first, we have to experiment and ask questions so we know how to guide the interns appropriately.
Embracing the complexity: Creatives and interns, working together
If you, like me, are a relentlessly curious creative, you understand the importance of testing and experimenting with ads to find that sweet spot — the right audiences, paired with the right ads, at the right frequency and sequence to deliver maximum return on ad spend.
To make this work, we need to embrace the complexity of working with a billion interns. We’re the ones who need to teach the machines what to look for and figure out what to do with the answers. But if we simply hunker down on what we know, if we don’t become as fluent with data as we are with culture, if we don’t challenge ourselves, then we won’t be able to make our creative better.
The billion interns are coming. We need to start experimenting now to be ready for them. They can help focus our creative energy on where it can do the most good and take a lot of grunt work off our plates, freeing us up to spend our time where we can have the most impact. They will help isolate and establish the value of our creative work. This will make the work, and all of us, much better.
Explore creative inspiration, tools, and best practices on Google’s creative hub create.withgoogle.com.