How AI Is Changing the Role of the Digital Marketer

We live in an increas­ing­ly auto­mat­ed world. Every month, machines take on more opti­mi­sa­tion deci­sions once made by humans. This trans­for­ma­tion has many people—marketers and con­sumers alike—wondering whether we’ll reach a point where we sim­ply set the machines free to do their job. Giv­en that many see this trans­for­ma­tion as inevitable, what will the role of the dig­i­tal mar­keter be in five years’ time?

In truth, many machines are already being set free. Recent­ly we’ve seen major break­throughs in speech recog­ni­tion from sys­tems like Amazon’s Alexa voice ser­vice. We’ve seen rapid growth in the pow­er of machine trans­la­tion in sys­tems like Google Trans­late. And tech­nolo­gies like auto­mat­ed image recognition—not to men­tion self-dri­ving cars—wouldn’t be pos­si­ble with­out major advances in the way com­put­ers see the world.

Learn­ing from feedback

What all these tech­nolo­gies have in com­mon is that they’re appli­ca­tions of arti­fi­cial intel­li­gence (AI)—more specif­i­cal­ly, neur­al net­works and deep learn­ing. We’ve had these tech­nolo­gies at least since the 1990s, but com­put­er sci­en­tists have final­ly har­nessed both the vast com­pu­ta­tion­al pow­er and the enor­mous store­hous­es of data—images, video, audio, and text files strewn across the Internet—that, it turns out, are essen­tial to mak­ing neur­al nets work well.

But neur­al nets, like many mod­ern AI tech­nolo­gies, don’t work in a vac­u­um. As projects like Google Trans­late and self-dri­ving cars have shown, AI can only do its job effec­tive­ly if it receives ongo­ing feed­back from the real world.

In 2007, many high street banks in the Unit­ed King­dom were using TouchClar­i­ty (a com­pa­ny I used to work for, now part of Adobe)—an auto­mat­ed behav­iour­al tar­get­ing platform—to deter­mine which prod­ucts to show to which cus­tomers. The results of machine learn­ing at the time indi­cat­ed that a hero ban­ner, along with three small­er cre­ative slots, was the best way to use the home­page real estate, and most banks’ home­pages real­ly did end up look­ing like this fair­ly rapidly.

Over time though, banks realised that the home­page had to be seen as more than a sales oppor­tu­ni­ty; it’s an effec­tive tool for sign­post­ing spe­cif­ic cus­tomer jour­neys. Banks arrived at this real­i­sa­tion through a cre­ative process and by con­sult­ing with customers—not as the result of a machine learn­ing exer­cise, which is why today their home­pages look very dif­fer­ent from that for­mu­la­ic hero ban­ner-led approach.

Even so, machines are hav­ing an impact on opti­mi­sa­tion in a num­ber of ways.

Advances in analytics

Gen­er­a­tive design uses com­put­erised algo­rithms to explore entire solu­tion sets. In oth­er words, you state your goals and con­straints, then allow the com­put­er to gen­er­ate designs and iter­a­tions for you that you might nev­er have thought of—a kind of accel­er­at­ed arti­fi­cial evo­lu­tion, as some have called it.

We often use the con­cept of gen­er­a­tive design in dig­i­tal mar­ket­ing, under the names audi­ence seg­men­ta­tion, clus­ter analy­sis, and pre­dic­tive ana­lyt­ics. These tech­nolo­gies analyse data in an auto­mat­ed way to dis­cov­er insights about our audi­ences more quick­ly, and even pre­dict future behav­iour, by under­stand­ing when and why a cus­tomer might get in touch. This com­put­erised “fore­knowl­edge” enables us to deliv­er enhanced per­son­al­i­sa­tion and allo­cate our resources more efficiently.

Now that gen­er­a­tive design can pre­dict cus­tomer behav­iour more accu­rate­ly than human mar­keters can, should the mar­ket­ing indus­try pre­pare for a mas­sive wave of redun­dan­cy notices? After all, as the Gov­er­nor of the Bank of Eng­land, Mark Car­ney famous­ly said, “Every tech­no­log­i­cal rev­o­lu­tion mer­ci­less­ly destroys jobs well before the new ones emerge.”

The truth, as usu­al, is a bit more nuanced and complex.

Trans­for­ma­tions in marketing

Mar­keters’ jobs will indeed under­go mas­sive changes over the next few years, but it’s not all gloom and doom. The changes will, on the whole, be positive.

Over the last few years, we’ve seen mar­ket­ing become more account­able for a broad­er range of dis­ci­plines, includ­ing direct rev­enue con­tri­bu­tion, and, in many cas­es, even cus­tomer expe­ri­ence. This set of roles, skills, and require­ments is too big for us to mas­ter with man­u­al process­es, which means we need to think about how machines can help.

We need to start by fos­ter­ing con­ver­sa­tions about how we can bring data sci­ence into our mar­ket­ing activ­i­ties, and, ide­al­ly, how we can work with data sci­en­tists with­in our own organ­i­sa­tions to pro­vide more of that all-impor­tant real-world feed­back to the machines that are becom­ing cen­tral to our work.

But even more than this, as Econsultancy’s CEO has pre­vi­ous­ly explained, the mar­ket­ing indus­try needs to start replac­ing its “T‑shaped people”—those with a spe­cif­ic, deep expertise—with “pi-shaped peo­ple,” who have a broad­er knowl­edge base and pos­sess skills that span both left and right brain dis­ci­plines. In oth­er words, we need mar­keters who can be ana­lyt­i­cal and data dri­ven, while also under­stand­ing brand image, sto­ry­telling, and expe­ri­en­tial marketing.

As mar­ket­ing under­goes these shifts in skills and think­ing, our rela­tion­ship with machines will also change. In many cas­es, we’ll move from being mere­ly oper­a­tors of dig­i­tal mar­ket­ing tools to being curators—choosing the best pos­si­ble solu­tion and work­ing along­side the com­put­er to co-cre­ate the ide­al design. And when machines advance to the point that they’re offer­ing their own opin­ions, which, strange as it sounds, may be as lit­tle as five years away, we’ll move into the role of men­tors, pro­vid­ing input in the form of human-defined goals, val­ues, and parameters.

In short, the idea is not to replace peo­ple with robots, but to “remove the robot from the per­son,” as Aviva’s CFO, Tom Stod­dard superbly expressed it in a recent inter­view. In oth­er words, AI will help free mar­keters from mechan­i­cal tasks and empow­er them to use their cre­ativ­i­ty to tell pow­er­ful, cre­ative sto­ries about their brands. When we set the machines free to do their jobs, we’re like­ly to find that mar­keters, too, are more free to do theirs.