In my past few posts, I’ve sug­gested that Unique Vis­i­tors is a flawed met­ric, prone to sig­nif­i­cant inac­cu­ra­cies. I real­ize this posi­tion is con­tro­ver­sial. In fact, one blog­ger sug­gested that it was down­right “irre­spon­si­ble”. That’s fine — there was a time when every­one thought the earth was flat. It takes time, often a great deal of time, to accept that the tra­di­tional view may have been wrong. It’s also fas­ci­nat­ing to me that despite their push­back, these naysay­ers have yet to offer any rea­son why Unique Vis­i­tors should be perserved as an accu­rate met­ric. Any­one? Buehler?

Iron­i­cally, even though I’ve heard the most push­back from ven­dors who regret­tably base much of their ana­lyt­ics on Unique Vis­i­tors, these inac­cu­ra­cies are not vendor-specific; rather, they are largely man­i­fes­ta­tions of the Inter­net itself as I high­lighted in my recent post, 15 Rea­sons why all Unique Vis­i­tors are not cre­ated equal.

I’ve also sug­gested that due to these lim­i­ta­tions, I often rec­om­mend that web ana­lyt­ics pro­fes­sion­als use Vis­its or Ses­sions in their base­line analy­sis, as it pro­vides a more accu­rate and depend­able view of con­ver­sion, per­sua­sive­ness, and ulti­mately Return on Invest­ment. And to clar­ify some recent feed­back I’ve received, this includes both macro and micro conversions.

For exam­ple, if you oper­ate a high-touch retail site where many vis­its often occur before pur­chase, you should look to not only macro site con­ver­sion (orders/visits), but also micro con­ver­sions such as prod­uct con­fig­u­ra­tor con­ver­sion (suc­cess prod­uct configurations/visits). For jew­el­ery sites this con­fig­u­ra­tor could be a “ring selec­tor”; for Travel sites it could be a “saved itin­er­ary”, for Auto­mo­tive sites it could be a “vehi­cle con­fig­u­ra­tion” or “dealer quote”, etc.

All of these rep­re­sent micro-conversions; which are effec­tively sub­sets of your macro site con­ver­sion and rep­re­sent key mile­stones or deci­sion points in the cus­tomer life­cy­cle.

In turn, I’ve acknowl­edged that Vis­its also has its own lim­i­ta­tions when it comes to mar­ket­ing opti­miza­tion ini­tia­tives like clus­ter­ing and seg­men­ta­tion, remar­ket­ing, latent response, life­time value, cam­paign ROI, and other “person-centric” analyses.

Due to these lim­i­ta­tions with both Unique Vis­i­tors and Vis­its, I’ve hinted at the con­cept of Unique Cus­tomers. In my opin­ion, Unique Cus­tomers addresses many of the sig­nif­i­cant inac­cu­ra­cies inher­ent to Unique Vis­i­tor mea­sure­ment, but also com­pen­sates for the lack of unique­ness when using Vis­its as a met­ric. So what is a unique cus­tomer and how do you mea­sure one? Read on…

What is a Unique Cus­tomer exactly?
Unique Cus­tomers can be derived using sev­eral dif­fer­ent approaches — but almost always relies on an indi­vid­ual iden­ti­fier like user “123456” or “mbelkin”. These iden­ti­fers are typ­i­cally assigned to your vis­i­tors dur­ing a pur­chase, reg­is­tra­tion, sub­scrip­tion, lead sub­mis­sion, or sup­port inquiry process.

For exam­ple, let’s assume you run a retail web­site called Elec­tron­ics Plus. When new cus­tomers pur­chase from Elec­tron­ics Plus, they are asked to cre­ate a new pro­file with their per­sonal infor­ma­tion. This infor­ma­tion, while not always required, typ­i­cally aids in cus­tomer sup­port and order exe­cu­tion, among other ben­e­fits. When the cus­tomer has com­pleted their per­sonal pro­file, Elec­tron­ics Plus assigns them a unique iden­ti­fier like “mbelkin”. This login is usu­ally secured with a pass­word so the cus­tomer can pro­tect their pur­chase his­tory, order sta­tus, or expe­dite pur­chases in the future. At the same time, the web­site itself typ­i­cally encrypts this unique iden­ti­fier so it can­not be eas­ily com­prised — for exam­ple, “mbelkin” would become cus­tomer “17894653”.

So let’s con­sider this for a moment. This sim­ple check­out process has just yielded an indi­vid­ual ID that in most cir­cum­stances, can be used to iden­tify a unique cus­tomer. And impor­tantly it does not suf­fer from the same inac­cu­ra­cies of Unique Vis­i­tors. In most cases, the ID will not change if vis­i­tors delete cook­ies; it will not change if mul­ti­ple vis­i­tors use the same com­puter; it will not change if one vis­i­tor uses mul­ti­ple com­put­ers; it will not change if vis­i­tors reject cook­ies; it will not be impacted by IP pool­ing nor will it be impacted by ded­i­cated cor­po­rate IP addresses.

This indi­vid­ual ID also does not suf­fer from the same lim­i­ta­tions of Vis­its; namely, the lack of unique­ness. Rather, each Unique Cus­tomer is truly unique — con­nected to one per­son, irre­spec­tive of time, place, or technology.

In short, this approach is among the best you’ll find for iden­ti­fy­ing truly unique peo­ple and prospects that visit your web­site. Fur­ther­more, this method­ol­ogy lends itself to much more robust mar­ket­ing analy­sis, cus­tomer seg­men­ta­tion, and life­time value analy­sis because it is con­nected to an indi­vid­ual. And per­haps most impor­tantly, since remar­ket­ing to indi­vid­u­als is a pri­mary strat­egy after you’ve com­pleted mar­ket­ing analy­sis or cus­tomer seg­men­ta­tion, this unique iden­ti­fier is crit­i­cal to exe­cut­ing those campaigns.

What if I’m not a retailer?
Many of you do not actu­ally sell prod­ucts and ser­vices online — and that’s OK. Unique Cus­tomers is not a retail-specific con­cept. Media pub­lish­ers can lever­age this strat­egy thru sub­scrip­tions and/or mem­ber reg­is­tra­tions. Lead gen­er­a­tion com­pa­nies in the auto­mo­tive, finance, tele­com, and tech­nol­ogy ver­ti­cals can lever­age lead forms, cus­tomer logins, mem­ber reg­is­tra­tions, prod­uct reg­is­tra­tions, or any num­ber of other events. Sup­port web­sites can also lever­age the reg­is­tra­tion process to estab­lish this iden­ti­fier, and most already do today.

A word on pri­vacy
No mat­ter how you intend to estab­lish your unique cus­tomer iden­ti­fiers, pri­vacy is a crit­i­cal con­sid­er­a­tion. Before you cap­ture any­thing, make sure your pri­vacy pol­icy prop­erly reflects the cus­tomer infor­ma­tion you intend to col­lect and is in com­pli­ance with your cor­po­rate stan­dards. In most cases, if you are using pseudo-anonymous iden­ti­fiers like “17894653”, your exist­ing pri­vacy pol­icy *may* suf­fice or require a lit­tle mod­i­fi­ca­tion. How­ever, if you intend to cap­ture the actual cus­tomer email address — or some other form of sim­i­larly per­son­ally iden­ti­fi­able infor­ma­tion — it is more likely you’ll need to mod­ify your pri­vacy pol­icy sig­nif­i­cantly to reflect this. Pri­vacy prac­tices and poli­cies are often very unique to each com­pany, so exam­ine this closely with your Legal team before pro­ceed­ing any further.

Wait, what about…?!
At this point, you prob­a­bly have a myr­iad of ques­tions about unique cus­tomers. Should this met­ric replace unique vis­i­tors? Should it supercede it when­ever avail­able? What about the mil­lions of vis­i­tors that I can’t col­lect a unique ID for? What about micro-conversions? What about behav­iors that occurred before the vis­i­tor iden­ti­fied them­self? How does this impact my KPIs? What new KPIs should I look at? What new opti­miza­tion pos­si­bil­i­ties does this offer?

These, and many oth­ers, are all good ques­tions with good answers. If you’d like to dis­cuss this topic fur­ther, please do not hes­i­tate to con­tact me and the Omni­ture Best Prac­tices Group.