Online recommendation – can the algorithm replace the seller?

As we know, per­son­al­iza­tion is one of the major chal­lenges of e‑commerce, in order to boost cus­tomer sat­is­fac­tion, to improve loy­al­ty and to gen­er­ate addi­tion­al sales. Regard­less of their indus­try or the size of their struc­ture, 75% of mar­keters ques­tioned as part of the sur­vey con­duct­ed in 2014 by Adobe on dig­i­tal mar­ket­ing opti­miza­tion are con­vinced that per­son­al­iza­tion plays a key role in the achieve­ment of long-term busi­ness objec­tives. In this con­text, the rec­om­men­da­tion is one of the most pow­er­ful mar­ket­ing tools and is becom­ing more and more essen­tial on retail web­sites, putting back the cus­tomer at the cen­ter of the sales process. Dis­cov­er in this arti­cle the objec­tives of the rec­om­men­da­tions and their ben­e­fits when they are imple­ment­ed on a retail website.

The rec­om­men­da­tion in mar­ket­ing, yes­ter­day and today

The aim of the rec­om­men­da­tion is to help the client to choose, by offer­ing prod­ucts match­ing their needs or their tastes. Tra­di­tion­al­ly, this rec­om­men­da­tion func­tion is pro­vid­ed by the sell­ers in the stores. Let’s look at a sim­ple exam­ple and let’s assume that I need a tent for my next camp­ing hol­i­day. I go to a spe­cial­ized super­mar­ket, in the tents sec­tion: if no ven­dor is avail­able to direct me, I will choose a tent by myself, and then will quick­ly be head­ing to the cash desk. On the con­trary, in the pres­ence of a sell­er, I would be able to receive per­son­al­ized advice, a rec­om­men­da­tion adapt­ed to my needs, and I would also prob­a­bly be sug­gest­ed addi­tion­al pur­chas­es such as a sleep­ing bag or a floor mat. The ben­e­fits of the rec­om­men­da­tion are clear: it increas­es the aver­age bas­ket while increas­ing the per­cep­tion of a suc­cess­ful cus­tomer experience.

At the moment, in-store sales are still much high­er than those con­duct­ed online, part­ly because of the pos­si­bil­i­ty of per­son­al advice. The intro­duc­tion of dynam­ic, auto­mat­ed rec­om­men­da­tions on your retail web­site is there­fore a real need, in order to accel­er­ate sales and prof­itabil­i­ty. I have already said it, pro­vid­ing qual­i­ty cus­tomer expe­ri­ence is para­mount. And in a world where sup­ply is more over­abun­dant than ever, where the slight­est research can take hours and where imme­di­a­cy has become essen­tial, mak­ing user’s pur­chas­es eas­i­er online by immers­ing them with­in a spe­cif­ic prod­uct uni­verse is a great way to enhance their cus­tomer expe­ri­ence. Today, tech­nol­o­gy allows mar­keters to, through the seg­men­ta­tion of audi­ences, the track­ing and the analy­sis of behav­ioral data, offer always more per­son­al­ized rec­om­men­da­tions, there­fore always more rel­e­vant. This is where the algo­rithm can, if not replace, at least become the equiv­a­lent of the sell­er, mix­ing cus­tomer knowl­edge, test and dynam­ic rec­om­men­da­tions of prod­uct, con­tent, and services.

The dif­fer­ent types of recommendations

Rec­om­men­da­tions can achieve dif­fer­ent objec­tives: dis­cov­er prod­ucts, offer alter­na­tive prod­ucts (such as a dif­fer­ent mod­el of tent if we use again the exam­ple of the camper client), offer prod­ucts bet­ter suit­ed (such as a fam­i­ly tent I have chil­dren), or lead towards com­ple­men­tary prod­ucts. The aim is to improve the cus­tomer expe­ri­ence by putting him in con­tact with prod­ucts that he wouldn’t have thought of on his own.

To achieve this, one of the most effec­tive solu­tions is to rely on behav­ioral analy­sis, using an algo­rithm that will ana­lyze vis­i­tor behav­ior and will either cre­ate audi­ence seg­ments based on this analy­sis, or will enrich pre-exist­ing audi­ence seg­ments (cre­at­ed via Adobe Ana­lyt­ics, for exam­ple). With a tool like Adobe Tar­get and its Rec­om­men­da­tions mod­ule, you will thus be able to link the behav­ior of audi­ence seg­ments with the var­i­ous exist­ing prod­ucts and ser­vices, in order to dynam­i­cal­ly rec­om­mend the prod­uct best adapt­ed to a spe­cif­ic audience.

Adobe Tar­get is used by Sony to escort its cus­tomers dur­ing their jour­ney on the PlaySta­tion Net­work (using a PS4 or PS3), but also on his mobile and web­site. Elliot Dumville, Direc­tor of Prod­uct Man­age­ment E‑Commerce at the Sony Enter­tain­ment Net­work, dis­cussed at the Adobe 2015 Sum­mit of Salt Lake City how the Adobe Tar­get rec­om­men­da­tions fea­ture allows them to con­sid­er­ably increase the aver­age bas­ket of their vis­i­tors, and there­fore their glob­al rev­enue. The cus­tomer vis­it­ing one of its spaces is then offered each time a com­ple­men­tary prod­uct, an extra ser­vice, or a prod­uct that they are like­ly to help giv­en their ini­tial choice (Peo­ple Who Bought This Also Bought). If you are inter­est­ed to learn more about how PlaySta­tion is using Adobe Tar­get to rec­om­mend rel­e­vant con­tent through­out the cus­tomer jour­ney, I can only advise you to have a look at this video:

Video: Rethink­ing rec­om­men­da­tions: A look at the NEW recs in Adobe Tar­get Premium

In con­clu­sion, the rec­om­men­da­tion should be based on a sol­id under­stand­ing of cus­tomers and a care­ful data analy­sis, hence the need for an extreme­ly pow­er­ful algo­rithm, a reli­able tech­nol­o­gy and effi­cient tools such as Adobe Expe­ri­ence Man­ag­er or Adobe Tar­get. How­ev­er, when all these ele­ments are com­bined, the results are gen­er­al­ly sig­nif­i­cant, and allow the retail site to increase its aver­age bas­ket, to boost its con­ver­sion rate, and to improve cus­tomer sat­is­fac­tion. Again, it comes back to the per­son­al­iza­tion of the cus­tomer rela­tion­ship, which is fun­da­men­tal for suc­cess in e‑commerce today: com­pa­nies using vis­i­tors tar­get­ing mul­ti­ply on aver­age their con­ver­sion rate by 2 (White Paper Adobe — Results of the 2014 study on the opti­miza­tion of dig­i­tal mar­ket­ing).

What about you, what do you think of online rec­om­men­da­tions? Have you had the oppor­tu­ni­ty to test their per­for­mance on your web­site? Feel free to share your expe­ri­ences in the com­ments section.

Cov­er Image — © goir —