Tra­di­tional retail­ers focus on com­mon met­rics, such as click-thrus, con­ver­sion, and cart aban­don­ment. Often, how­ever, too many retail­ers focus their efforts around only these met­rics, defin­ing elab­o­rate plans to get more traf­fic, more peo­ple into the fun­nel, and hope­fully more con­vert­ers. So, rather than step­ping back and ask­ing some basic ques­tions — such as “Who is my cus­tomer?” and “Why does my site mat­ter to him or her?” — they blindly move for­ward with a strat­egy focused around met­rics rather than customers.

I’ve done a great deal of work with media and enter­tain­ment com­pa­nies, which tend to develop cus­tomer– or audience-focused strate­gies. If your rev­enue stream is sell­ing adver­tis­ing, get­ting more peo­ple to your site, get­ting them to engage more for longer peri­ods of time, and get­ting them to repeat in a given month, then an audience-focused strat­egy is key. Although cer­tain met­rics such as page views, unique vis­i­tors, and video starts still make up a key part of the online mar­ket­ing plan, grow­ing audi­ence seg­ments that are valu­able based on behav­ior, repeat vis­its, and time spent on the site are cru­cial to exe­cut­ing on any effec­tive mar­ket­ing plan.

So, where to begin?

Attribute Selec­tion

Start with your most engaged vis­i­tors and dis­cover which attrib­utes dif­fer­en­ti­ate them from other visitors.

Build­ing a customer-focused strat­egy means under­stand­ing what makes valu­able cus­tomers valu­able. For instance, per­haps an engage­ment with a par­tic­u­lar offer or prod­uct cat­e­gory will show value across a cus­tomer seg­ment, but that engage­ment has a sharp cor­re­la­tion with cus­tomers under 50; or per­haps dif­fer­ent behav­ioral attrib­utes — such as social likes and tweets — cor­re­late with a strong affin­ity with a par­tic­u­lar prod­uct line. Devel­op­ing cus­tomer seg­ments with a focus on the behav­iors and met­rics that define those seg­ments will help you cre­ate cus­tomer per­sonas that you can tar­get through spe­cific cam­paigns, mar­ket­ing chan­nels, or prod­uct offers.

As an anal­ogy, think of a sales­per­son in a store. A good sales­per­son can spot a cus­tomer and size him or her up pretty quickly. The sales­per­son can deter­mine whether a per­son is likely to buy, can make a sug­ges­tion about par­tic­u­lar prod­uct that might be of inter­est, and can spot peo­ple who are just brows­ing or killing time. We are try­ing to use dig­i­tal data to size up the same thing in our online world by pick­ing up on behav­ioral cues and cre­at­ing a vir­tual “gut instinct” based on known correlations.

Cor­re­la­tion and Cau­sa­tion
Don’t assume cor­re­la­tion equals causation.

Seg­ments are seg­ments. They are help­ful for under­stand­ing a cus­tomer set and what might moti­vate those cus­tomers, but in the end, they are just assump­tions. You would never assume in a face-to-face sales sce­nario that you know what a cus­tomer wants; you might make an edu­cated guess and ask ques­tions accord­ingly, but in the end, you are mak­ing assump­tions and expect the inter­ac­tion to dis­till more data.

Per­haps in your data you iden­tify a par­tic­u­lar prod­uct cat­e­gory that seems to excite a set of cus­tomers, so you develop a pro­gram to tar­get that prod­uct cat­e­gory across that cus­tomer seg­ment … but it shows no mean­ing­ful lift. Does that mean that the seg­men­ta­tion strat­egy was wrong? Not nec­es­sar­ily. All that the seg­men­ta­tion shows is com­mon­al­i­ties and dif­fer­en­tia­tors. It doesn’t show true cus­tomer moti­va­tions. Your tar­get­ing pro­gram made some assump­tions about what the data showed, and your assump­tions were wrong — not the data.

So, the first thing to do is to imple­ment the seg­men­ta­tion strat­egy across the entire tar­get­ing pop­u­la­tion and see which seg­ments respond to which offers. This will give you a bet­ter under­stand­ing of moti­va­tors on a segment-by-segment basis.

Tar­get the Past — and the Future
Don’t tar­get based only on past behav­iors, but also on the next page.

If you knew what cus­tomers wanted when they arrived at your site, you would be golden. I want a watch. I land on Zap­pos, and presto: all its watches appear, already sorted to my inter­est. But, unfor­tu­nately, this intent-based mind read­ing is not pos­si­ble, so we have to focus on what we do know. We know from the data what cus­tomers have looked at pre­vi­ously. We know how their inter­ests are sim­i­lar to and dif­fer­ent from what oth­ers have looked at and con­verted. So, what is the “tip­ping point behavior”?

If you are clear about the tip­ping point behav­iors that would put me in the watch-buyer seg­ment, you can iden­tify the behav­ior and auto­mat­i­cally assign me to that seg­ment when I exhibit inter­est. For watches, per­haps it’s as easy as click­ing through from a paid Google search; you see that my refer­rer includes watches or pop­u­lar watch brands, and imme­di­ately you pro­mote me to a mem­ber of the watch-buying seg­ment. Or per­haps I get there by view­ing more than three watch brands across two vis­its. What­ever the deter­mi­nate behav­ior, lever­age what you know now based on where I am in my visit and tar­get me immediately.

Bet­ter under­stand­ing your cus­tomers, their inter­ests, and their moti­va­tions should have a direct impact on your bot­tom line. Dri­ving con­ver­sion, after all, is key, but it may take time to have a mean­ing­ful and sta­bi­lized impact.

In Con­sult­ing, we’ve devel­oped a num­ber of tools and tech­niques for eas­ily iden­ti­fy­ing what makes cus­tomers valu­able and which activ­i­ties draw them in. Ana­lyz­ing those valu­able cus­tomer seg­ments and how they change over time can help you devise a strat­egy that will sup­ply a wealth of effec­tive, tac­ti­cal tar­get­ing efforts that will ulti­mately help you achieve that goal.