Marketers obsess about every detail of their campaign, spending hours defining the audience targeting rules and picking out the perfect image to accompany our meticulously crafted copy. But how can we be confident that these steps are making a difference in our business’ top line?
Those that love analytics will suggest using data, running A/B tests and proving hypothesis with hard facts. However, marketing tools are often clunky, and we’re dependent on IT departments, which can make experimentation slow. According to eMarketer, marketing technology is still a broadly unresolved concern, so we get the choice of running campaigns based on our gut or endure the process of getting that A/B test launched. We often choose the former, and our biases remain unchecked.
Thus, we continue the same journey: invest time in creative and be extra careful with targeting. Our mindset is that each campaign should be different from the last, so any sort of reuse is wrong.
This is known as action bias, a built-in psychological mechanism that doesn’t allow us to sit still and tells us that any action is better than leaving things as-is. The same mechanism appears in soccer where goalkeepers jump on penalties when standing still would have resulted in better outcomes.
Marketers are just as susceptible. We don’t reuse our old visuals or targeting, as it would be humiliating or seem like we’re not trying enough. But what happens when we challenge this assumption? What if storytelling, not visual design, made the biggest difference? What if a machine was better at targeting than any human?
If you interview a customer about your last campaign, will they remember the image you used or the color of the call-to-action button? They probably won’t, but we spend time—or agency money—iterating on these minuscule aspects.
What if instead you created a portfolio of email templates and committed to making each campaign fit into one? What if you time-boxed the investment for customization? This is exactly what eBay did when they transitioned from an ad-hoc campaign process to templatized, scalable ones, which cost a fraction of the price. This allowed eBay to concentrate on the relevance of its message rather than superfluous aspects of the campaign.
Go wild with the templates, but then reuse them and minimize customization. Then you can apply them across mediums (e.g., display, social, video ads).
Marketers view 1:1 personalization as the holy grail, but 93% of U.S. consumers say they aren’t getting any relevant marketing communications. However, 40% of marketers report personalization efforts are producing incremental sales and customer satisfaction.
Is the root cause the tools we use? Most mar-tech tools force humans to do the heavy lifting by creating sophisticated segmentation rules to drive personalization. We can’t scale our operations, creating 100 highly targeted campaigns instead of 10 broad ones. Costs would skyrocket, and the complexity would explode. With so many small campaigns, how can we understand what our customer contact strategy is? There’s got to be a better way.
Machine learning for digital marketing
If we equip a machine with intuition on the factors that are most predictive for customer preferences, it’ll do a great job applying this knowledge at scale. It’ll automatically run dozens of statistical experiments at the same time and reliably predict what customers will like, allowing us to reach 1:1 personalization by treating each customer as a unique entity.
One example is how Netflix merchandizes each show differently for their customers, picking the artwork for each viewer based on machine learned recommendations that are fine-tuned over time and improved with each users’ action.
New mar-tech tools that enable this level of sophistication are already in-market and will change the game for digital marketing and free us up to do the kind of work that only humans are good at, like empathy, creativity and original thought.
Humans like to think that controlling every aspect of things we care about will maximize chances for success. However, in many situations, letting go of the wheel allows us to invest our time where we really excel. And the machine can now augment (not replace!) us to help scale beyond what was ever humanly possible.