Ten years ago, most marketers relied on pretty basic testing and data science to make decisions about how to best reach their customers. A/B testing helped us compare two different offers or subject lines.
These days, there’s an exciting new world of personalization and customization for marketers to explore — and much smarter ways to test many different variables at once. So how can marketing teams learn about the newest technology and tools and make sure they’re driving intelligent campaigns?
I asked David Baker, co-founder of the adaptive messaging firm Cordial and a pro with more than two decades of experience in the adtech and marketing space, for his predictions and advice on the future of machine learning for marketers.
We’ll see a push for more relevant data — not more data
Most marketers don’t actually need more data, Baker says. Especially at big organizations, marketers have plenty of data. “What they need is to eliminate noise so they can quickly find insights and make decisions,” he says. Finding those quick insights, without having to sort through a lot of irrelevant information, will help agile marketing teams get a leg up on the competition. Sorting through a ton of data can be a slog that slows teams down. Baker says the reason machine learning and artificial intelligence are becoming more mainstream is simple: Marketers want access to insights now, not three months from now.
We’ll rely on more iterative testing
Baker believes many organizations will begin to look at testing as an iterative exercise, with small samples that exploit winning combinations quickly.
That means understanding your audience in context. “The fact is, you are a different person on your phone in the morning than you are at 7 p.m. at night on the couch in front of the TV with your tablet in hand,” he says.
So, ideal testing allows algorithms to alter tests as they run, instead of applying a blanket rule to the entire audience over time. “It’s the difference between chess and checkers,” he says. “Your thinking needs to evolve. The technologies are there if you can operationally scale it.”
We’ll get more sophisticated about our testing priorities
Baker says that many marketers miss a key point: Machine learning and AI shouldn’t be rolled out everywhere at once. “The value of machine learning and AI is that they increase accuracy and speed,” he says. To harness that speed, marketers should make decisions at what key problems they’re interested in, and find the right methods for solving those specific problems.
He suggests focusing on where testing needs to happen. “Few marketers can test effectively across email, web and mobile,” he says. Data is often stored in disparate systems, and testing everything at once almost always leads to wasted efforts.
Then, look closely at the different kinds of testing you could implement. Some AI technology is great at helping solve ‘What’s next?’ questions about products, types of customers, or promotion, he says. Other technology is better for optimizing interactions, and still others are primed to help marketers make real-time decisions — like in customer service interactions, conversational commerce and trigger responses.
“The key is to pick the low hanging fruit and run as fast as you can,” he says.
We’ll be more discerning about our Martech vendors
Baker suggests spending some time understanding the marketing tech landscape, including the capabilities of each technology. The industry is complex, with more than 1,400 vendors, so it’s worth investing some time in getting to know what’s out there.
Every marketing team will have to build their own use case to understand which tools are best for their needs. “So many tools are easy to do single tests at a time, but few offer operational scale when you have to run many experiments in a single campaign,” he says. He says marketers should look for:
- Design flexibility to customize what results you’re testing for (for example, clicks, tests or conversions).
- Audience size flexibility.
- Self-running optimization that can learn and run autonomously.
- Good reporting and insights to show you what’s working.
The best marketing organizations will make strategic decisions about the technology that powers their work. In other words, it’s not just about snatching up the latest shiny tools. “So many platforms have testing tools, but a lot of your success is in how you approach it and where you focus first,” he says.