Tech­nol­ogy is ad­vanc­ing fast and mak­ing per­son­al­i­sa­tion eas­ier and more af­ford­able, but there’s no point cus­tomis­ing your prod­uct or ad if it makes no sense to the re­cip­i­ent

Campaign UK - - PROMOTION - David Fletcher is chief data of­fi­cer at MEC UK

March 2012: Star­bucks has just in­tro­duced the much vil­i­fied “Can I get a name for your or­der, please?” thing and one wag fur­ther back in the queue shouts: “Don’t tell him, Pike!”

Five years on (and at least one Star­bucksmis­spell-names-on-pur­pose-to-drive-so­cial­me­dia-ex­po­sure con­spir­acy the­ory later): from “Share A Coke” to ecom­merce rec­om­men­da­tions and car con­fig­u­ra­tors to lo­cal-dealer end­frames, per­son­al­i­sa­tion is start­ing to look like mar­keters’ mass­par­tic­i­pa­tion sport.

Those agile enough to de­liver made-to­order pur­chases in high-value cat­e­gories – cars, travel and so on – have cus­tomi­sa­tion em­bed­ded in the way they cre­ate value, so cus­tomi­sa­tion in the ad­ver­tis­ing part of the chain is a nat­u­ral ex­ten­sion.

Me­dia-own­ers across the spec­trum are cus­tomis­ing (editorial) de­liv­ery to main­tain and de­velop their own value. Au­di­ence busi­nesses are be­com­ing cus­tomer busi­nesses al­low­ing per­son­al­i­sa­tion of ad­ver­tis­ing, too – re­gard­less of whether the prod­uct it­self needs per­son­al­is­ing.

Not only do we all have ac­cess to ev­er­in­creas­ing vol­umes of ever-more types of data, the tech­nol­ogy to store and sift it is tak­ing com­mer­cial shape. And trans­form­ing this is ar­ti­fi­cial in­tel­li­gence, which does for messy, mas­sive, un­struc­tured data sets what busi­ness logic and a pro­gram­mer can do with neater data: au­to­mate tasks where de­scrib­ing pre­cise rules is much harder.

You too can ac­cess Lex, the speech-to-text thing in­side Alexa, and text-to-speech sib­ling Polly for a princely 0.0004¢ per char­ac­ter. Vis­ual-recog­ni­tion tech is there, too. And, while it might be too early to de­clare that that’s the bots sorted, progress is ex­po­nen­tial. One FS client user noted that speech-recog­ni­tion ac­cu­racy has moved from 80% to 99% within just the past 18 months.

But whereas the op­por­tu­ni­ties have ex­panded greatly, the es­sen­tial rule re­mains the same: per­son­al­i­sa­tion – like al­most ev­ery­thing in ad­ver­tis­ing – is about con­text.

The eas­ier ex­am­ples to get ex­cited about are those like Star­bucks, where per­son­al­i­sa­tion ap­pears out of con­text or away from the norm. When ac­ti­vated in a way that makes sense to the re­cip­i­ent, these are un­doubt­edly pow­er­ful and likely to be re­ceived with more up­side than down­side.

Chan­nel 4 VOD ser­vice All4’s re­cent launch of cus­tomi­sa­tion by au­dio saw the trailer for Alien: Covenant use view­ers’ names in au­dio as well as the end­frame: “David. Run.” It’s out of con­text be­cause we still don’t ex­pect our names to ap­pear in TV ads, even on logged-in plat­forms like All4. Chan­nel 4 reck­ons on maybe only one cam­paign in 30 us­ing some form of per­son­al­i­sa­tion, keep­ing the norm… the norm.

In con­trast, so­cial-me­dia plat­forms are, by def­i­ni­tion, per­son­alised ex­pe­ri­ences, so a lack of ad­ver­tiser cus­tomi­sa­tion could be ar­gued as detri­men­tal to the ex­pe­ri­ence. But per­son­alised here doesn’t mean us­ing peo­ple’s names or as­sumed pref­er­ences like an overly fa­mil­iar sales­per­son. Rather, it lies in cus­tomi­sa­tion of mes­sage around life­style and in­ter­ests that are more likely to fire a “Sys­tem 1” re­sponse to your thumb to pause and play than a generic ad would.

Face­book points out that there’s a bal­anc­ing act to do be­tween reach and the ben­e­fits to rel­e­vance that per­son­al­i­sa­tion brings. Pur­su­ing per­son­al­i­sa­tion re­duces achiev­able au­di­ence size. So there’s a com­mer­cial ar­gu­ment in there, but most mar­keters also want im­pact at scale.

And this is prob­a­bly the cue for all of us. Achiev­ing rel­e­vance at scale doesn’t nec­es­sar­ily re­quire a deep un­der­stand­ing of in­di­vid­u­als. But chang­ing au­di­ence and copy for the con­text of a wide va­ri­ety of data sig­nals – of which in­di­vid­ual data might be a part – sounds scaleable.

De­liv­er­ing on this is an im­per­a­tive, as Sarah Gold­ing has pow­er­fully ar­gued for her ten­ure as IPA Pres­i­dent.

There’s magic in the ma­chines which is up to us to find.

“The op­por­tu­ni­ties have ex­panded, but per­son­al­i­sa­tion is [still] about con­text”

Newspapers in English

Newspapers from UK

© PressReader. All rights reserved.