I’m a guy and I work in the tech indus­try, so it’s prob­a­bly not sur­pris­ing I like tech ‘toys’. When a new tech gad­get comes out, (i.e. Ama­zon Kin­dle Fire, Sam­sung Galaxy Tab, iPad 3, etc.) it’s hard for me to resist the urge to go out and buy it because it’s new. Do I need it? Prob­a­bly not since I already have a Nook Color and it suits most of my needs right now. But the other devices are excit­ing, shiny and NEW! ‘It’s going to change my life,’ I say to myself. I feel like I need it! Thank­fully for my bank account, my wife helps keep my early-adopter ten­den­cies to a minimum.

Not unlike my urge to go out and pur­chase another gad­get, some retail com­pa­nies find the allure of devel­op­ing an advanced mar­ket­ing attri­bu­tion model too much to resist. An exam­ple of an advanced attri­bu­tion model is cre­at­ing a weighted scor­ing sys­tem to give a per­cent­age of credit to a mar­ket­ing chan­nel based on how far away that mar­ket­ing chan­nel was from an order.

The devel­op­ment of an advanced mar­ket­ing attri­bu­tion model can be incred­i­bly insight­ful for some retail com­pa­nies, but less so for oth­ers if they aren’t in a posi­tion to take advan­tage of it. Before going down the path of invest­ing resources in devel­op­ing it, I’d encour­age you to answer two questions:

  • Ques­tion 1: What per­cent­age of our users will we gain added insight into?

If you decide to use an advanced attri­bu­tion model, do it because it makes sense for your busi­ness, not because it’s the fun, shiny and NEW thing you hear of other retail com­pa­nies doing. Some spend con­sid­er­able time, effort and money in devel­op­ing an awe­some attri­bu­tion model only to real­ize it applies to a small por­tion of users and the action and insight they end up with doesn’t jus­tify the orig­i­nal invest­ment. Before you jump in, make sure you know how deep the water is. Start by deter­min­ing what per­cent­age of orders occurs on a cus­tomers’ first visit vs. a return visit. You can eas­ily pull this infor­ma­tion from the Visit Num­ber report in SiteCatalyst.

Take the fol­low­ing two exam­ple companies:

Com­pany A receives a lit­tle more than half of all orders on users’ first visit to the site, leav­ing the remain­ing orders occur­ring on return vis­its. An advanced attri­bu­tion model may make sense, but don’t for­get the 53% of users for whom the model wouldn’t give any addi­tional insight.

Com­pany B is a slightly dif­fer­ent story. Three-quarters of orders involve mul­ti­ple vis­its and so an advanced attri­bu­tion model would pro­vide under­stand­ing to a greater per­cent­age of the site’s users. The devel­op­ment of an advanced mar­ket­ing attri­bu­tion model should be a higher pri­or­ity for Com­pany B to Com­pany A.

This doesn’t mean Com­pany A should avoid going down the path of devel­op­ing an advanced attri­bu­tion model. How­ever, if Com­pany A delays dig­ging into the 53% and opti­miz­ing their mar­ket­ing chan­nels while they fig­ure out that advanced attri­bu­tion model, then they’ve missed the point. Iden­tify what data you can take action on now and then do it.

  • Ques­tion 2: What are we doing with what we cur­rently have?

One of the rea­sons I orig­i­nally pur­chased my Nook Color was that I liked the idea hav­ing a vari­ety of books always at my fin­ger­tips. I had grand visions of lots more read­ing and being mag­i­cally trans­ported to a place where I could eas­ily get through the ever-growing list of books on my to-read list. In real­ity, I’m still plug­ging away at about the same rate as I was before I pur­chased my Nook.

Sim­i­larly, the level you are using your exter­nal mar­ket­ing data before devel­op­ing an advanced mar­ket­ing attri­bu­tion model is the same level you’ll use it after devel­op­ing an advanced mar­ket­ing attri­bu­tion model. If we go back to our com­pany exam­ples, let’s look back at Com­pany A. They have a healthy por­tion of their busi­ness they can start opti­miz­ing imme­di­ately with­out an attri­bu­tion model. Even if they decide to incor­po­rate an advanced mar­ket­ing attri­bu­tion model, they shouldn’t delay dig­ging into the data they have now. At the end of the day, web ana­lyt­ics exists to help retail­ers make more money and if you aren’t using the data to take action, you’ve miss­ing the boat.

Advanced mar­ket­ing attri­bu­tion mod­els can pro­vide a wealth of insights into the rela­tion­ship between dif­fer­ent mar­ket­ing chan­nels and how to opti­mize your over­all mar­ket­ing spend (look for an upcom­ing post from my friend and col­league Michael Hal­brook). How­ever, make sure you are in the right place to take advan­tage of the insights and it’s the right time to incor­po­rate into your business.