The Customer Persona is Dead! Long Live the Customer Profile!

The era of cus­tomer per­sonas is over.

Try­ing to fit your audi­ence into pre-defined cat­e­gories isn’t just rudi­men­ta­ry, it’s inef­fec­tive. That’s why per­sona-based mar­ket­ing, which is an inside-out approach to under­stand­ing audi­ences, increas­ing­ly leads to irrel­e­vant expe­ri­ences. Brands are so used to seg­ment­ing cus­tomers as “males aged 18–24” or “females in their ear­ly 30s” that they risk miss­ing the big pic­ture. And in the worst case, these unfor­tu­nate bias­es can have an effect upon their rela­tion­ship with customers.

Con­text is every­thing, espe­cial­ly when try­ing to make sense of people’s activ­i­ty across so many chan­nels. It’s time to kill per­sona-based mar­ket­ing and embrace a more pro­gres­sive approach to per­son­al­i­sa­tion that reflects this real­i­ty. The chal­lenge is that there is no lin­ear way of dri­ving the cus­tomer expe­ri­ence today. Cus­tomers have so many ways of access­ing con­tent and inter­act­ing with brands that tar­get­ing them based on data from just one of these chan­nels is not enough.

The sec­ond chal­lenge is the frag­men­ta­tion of media today. TV, online, mobile, walled gar­dens like Face­book or YouTube – each of these chan­nels isn’t just used dif­fer­ent­ly, it also gen­er­ates dif­fer­ent forms of cus­tomer data. It has there­fore become cru­cial to col­late data from all of these sources onto a unique sys­tem and devel­op a com­plete view of each customer.

A com­plete cus­tomer pro­file is the holy grail of per­son­al­i­sa­tion, but it is a com­plex beast that requires the right mix of tech­nol­o­gy, process, and cul­ture to get right. Let’s tack­le each of these fac­tors one by one.

Be mindful of the ecosystem you’re serving

Dif­fer­ent chan­nels require a dif­fer­ent approach to cam­paigns and tar­get­ing. It’s not uncom­mon that a brand’s ana­lyt­ics set­up works well for desk­top audi­ences but isn’t suit­able for under­stand­ing mobile users. Or for an audi­ence to scale well in a web envi­ron­ment but not in an app.

It’s cru­cial to first define the ecosys­tem of peo­ple and chan­nels you’re serv­ing, and reverse engi­neer the right tech stack from there. Again, cus­tomer behav­iour should gov­ern every aspect of your deci­sion-mak­ing, even at the tech­nol­o­gy lev­el. The ide­al tech stack con­nects all these chan­nels togeth­er so you can adapt your approach based on the needs of spe­cif­ic cam­paigns and audi­ences. This becomes par­tic­u­lar­ly impor­tant today, with data man­age­ment becom­ing a major pri­or­i­ty – with­out a link between chan­nels, data leak­age and loss become high­er risk issues.

Get granular with data, but only when you need to

The gen­er­al mantra around data today is “the more, the mer­ri­er”. With more data sources than ever to help them under­stand audi­ences, brands are build­ing more com­plex pro­files than ever to help them per­son­alise experiences.

This isn’t a bad approach per se, but it’s not nec­es­sar­i­ly the right approach every time. How deep you dive into your data ulti­mate­ly comes down to your cam­paign objec­tives. For exam­ple, if you’re chas­ing a spe­cif­ic KPI for a high­ly spe­cialised audi­ence (i.e. gen­er­at­ing leads among CIOs work­ing in your top 25 B2B accounts) it makes sense to get gran­u­lar, but in the case of an always-on cam­paign a broad­er approach might be best, so you can broad­en the net and dri­ve awareness.

Get­ting gran­u­lar also rais­es the chal­lenge of work­ing with too much data, which fur­ther com­pli­cates things when try­ing to refine your approach for indi­vid­ual cam­paigns. This isn’t just about work­ing with many data sources, but also about under­stand­ing where that data has come from when work­ing with third-par­ty part­ners and DMPs. Even the dif­fer­ence between prob­a­bilis­tic data (which is col­lect­ed auto­mat­i­cal­ly) and deter­min­is­tic data (which peo­ple sub­mit will­ing­ly) will affect how that infor­ma­tion is analysed and fac­tored into targeting.

Lead­ing brands tend to opt for flex­i­bil­i­ty, choos­ing a tech stack that allows them to draw on the right DMP (or DMPs) for each cam­paign. They see the val­ue of work­ing with a part­ner that under­stands their data needs, is con­nect­ed to a broad range of DMPs, and most impor­tant­ly under­stands the sto­ry behind all the data these plat­forms collect.

Test, learn, and be flexible

The top­ic of flex­i­bil­i­ty bring us to the impor­tance of test­ing. It might sound like I’m going retro and revert­ing back to how we under­stood audi­ences in the 90s, but there is no sub­sti­tute for tak­ing a step back and test­ing your hypothe­ses with real cus­tomers to under­stand what makes them tick.

Ear­ly in my career, a client asked me to run a Face­book cam­paign only tar­get­ing pilots and cab­in crew, who they saw as their only rel­e­vant per­sonas. My instincts told me the brief was too restric­tive, so I ran a par­al­lel test to see how their mes­sage would res­onate with oth­er audi­ences. The tests revealed that sport lovers in par­tic­u­lar were excep­tion­al­ly recep­tive to our mes­sage, dri­ving 50% more con­ver­sions week on week than the expect­ed audi­ence. As a result, our client reduced their spend on pilots and cab­in crew and dif­fer­en­ti­at­ed with invest­ment in more poten­tial audiences.

We live in a dig­i­tal­ly col­lec­tive world where peo­ple change their ideas about prod­ucts and brands almost dai­ly, and in which there are no longer lines between how we con­sume con­tent in our pro­fes­sion­al and per­son­al lives. There is no way a brand can accu­rate­ly define the breadth of peo­ple who will care about their mes­sage, and this is where they are miss­ing out.

The AI effect

I couldn’t dis­cuss the top­ic of tar­get­ing and audi­ence test­ing with­out men­tion­ing Arti­fi­cial Intel­li­gence. From Nor­we­gian tele­coms leader, Telenor, to London’s Heathrow Air­port, brands every­where are using AI to col­lect, analyse, and act on cus­tomer data more quick­ly and for a broad­er audience.

The beau­ty of AI is that even as we con­tin­ue to refine algo­rithms man­u­al­ly (up-weight­ing and down-weight­ing var­i­ous sig­nals, improv­ing its approach to tar­get­ing, and so on) the tech­nol­o­gy is also devel­op­ing on its own. Today’s algo­rithms are con­stant­ly test­ing and learn­ing about cus­tomer behav­iour at speed and on an incred­i­bly large scale.

AI is like a new employ­ee you need to train before they take on more respon­si­bil­i­ty. And it’s learn­ing fast. It will become the marketer’s ulti­mate assis­tant, pre­dict­ing what we’re going to do next and pro­vid­ing us with added insight to ensure we make the best pos­si­ble deci­sion every time. The tech­nol­o­gy is by no means per­fect yet, but judg­ing from the inno­va­tions on dis­play at CES we’re already much clos­er than we were just a year ago.

Read more about how Adobe Sen­sei AI is bring­ing new speed and insight to cus­tomer ana­lyt­ics. And click here to learn how Adobe Audi­ence Man­ag­er is help­ing brands build com­plete, accu­rate cus­tomer profiles.