Some industry folklore is repeated so often, so widely, that we collectively just take it for granted. It’s always been that way and always will be.
It’s these types of conventions that my team, Google’s Media Lab, likes to challenge. When put to the test, some of them hold true. Others turn out to be myths that, once busted, enable us to move our marketing to the next level. Here are three marketing myths we’ve busted in the past year.
Marketers understand that they should be adapting and optimizing creative assets to the specific features of the platform where they’ll run.
The industry myth is that producing video in this way is both expensive and takes lots of time. A director and large crew have to be hired, and they have to shoot somewhere glamorous, taking their expensive equipment with them. By the time you get through post-production, it’s taken many months and a lot of dollars. The final piece of art, beautiful though it may be, now has to run everywhere — one size fits all — with no time or resources to adapt or customize.
As we’ve learned here at Google, though, it doesn’t have to be that way. We put together a team of people tasked with carrying out digital-first video experiments. Their goal was to find ways to make video faster, cheaper, and more effective. And they’ve already had some successes.
For example, for the launch of the Google Home Hub, we took one base video and ran it through a tool called Directors Mix that allows you to create custom videos on a massive scale. We ended up with 80 versions of the ad, each tailored to a different context. So if someone had just looked up a recipe on YouTube, they might have seen this version, which showed how to make stuffing.
Whereas someone searching for some background music for their holiday party might have instead seen this one, showing how easy it is to look up a music playlist on the Google Home Hub.
The best part? Each of the videos ended up costing just $1,800 to create. We spent $144,000 to build 80 contextually relevant ads, proof that video doesn’t have to be slow or expensive.
In digital marketing, we gather all sorts of data points to understand whether our creative and media strategies are going to plan. We can see how long someone spent watching a video, how far someone scrolled down a page, or how many of our website visitors are bouncing. The list goes on.
But just because you can measure something, does that mean you should? We’ve realized that, when it comes to data, less is more.
It all started when we audited the analyses being shared with our leadership by teams across Google Marketing. We discovered that, collectively, we were reporting on 70 different metrics globally. How did we expect our CMO and VPs to make coherent decisions, to compare one campaign or strategy to another, when our teams weren’t speaking the same language?
Instead, as my colleague Avinash Kaushik wrote on Think with Google earlier this year, we’ve whittled down all those data points to just six metrics that matter. Why that number? Because we run two types of campaigns: brand and performance. Across those campaigns, we care about three things: whether we’re capturing people’s attention, how they’re behaving in response, and what the outcome is. So now, rather than drowning in metrics, we have just one for each of the things we’re interested in measuring.
“As advertisers in the age of machine learning and artificial intelligence, it’s easy to think of ourselves in an epic faceoff with these machines,” Ben Jones wrote in a Think with Google piece last year. This fear that machines will displace us is a normal one, and it’s certainly not limited to the marketing industry. But the fear is unfounded. Instead, as we’ve discovered through our experiments this past year, it’s about understanding what machines do better than us and letting them get on with it, freeing up humans to lean into what we do uniquely well: insights, inspiration, and creativity.
Here’s an example. This is the equation for calculating customer lifetime value (CLV) — a way of identifying who your most valuable customers are, something all marketers need to know.
Now I’m no mathematician, so it would literally take me a lifetime to figure out what this means. But even the most analytically minded people would take a while to work this out manually, which is why we used to get an updated CLV only every six months.
Then we turned to machine learning. By handing off our data — like traffic sources and previous campaign performance — to a tool called TensorFlow. We went from having access to one CLV every six months, that we then had to use across all our bids, to having 2,000 predicted customer value (pLV) calculations a day. Being able to use a real-time pLV versus a six-month-old CLV has allowed us to better optimize and regularly refresh our Google Ads bidding strategy.
Machines can also free up time in the area of ad creative. For example, we’ve been able to use smart creative technology to optimize display and search ads in real time based on how people respond to them. The format has greatly outperformed the legacy static display and search ads we’d been using for more than a decade.
Machines are well cast for these highly manual, low-level decisions. So we let them do it, freeing us up to focus on the things machines can’t do as well as us — like identifying the next marketing myth we need to bust on our journey to becoming smarter, more effective digital marketers.
By: Joshua Spanier