It was sup­posed to be a nice relax­ing day of sort­ing through his email and get­ting caught up from a long yet enjoy­able hol­i­day break. Unfor­tu­nately for Arte­mus Brown, Ana­lyt­ics Man­ager for Geometrixx, his expec­ta­tions were not to be met on that par­tic­u­lar day.

Arte­mus Brown approached the first mar­ket­ing team meet­ing of the year like all the oth­ers, he was ready to speak to the most recent sales cycles and pre­pared to answer detailed requests with “Let me look into that and get back to you.” When asked by his direc­tor “What were the gross sales totals in rev­enue as well as orders for the week lead­ing up to Dec 25th?” he was armed and ready with his num­bers to which he said, “We wit­nessed 92,000 sales that equal $2,281,000 in gross rev­enue. This shows an increase in aver­age order value of 26 per­cent for the period.” Given that his answers were above the pro­jec­tions he and all the oth­ers in the room smiled and relaxed a bit. His direc­tor responded with, “Fan­tas­tic it looks like our prod­uct rec­om­men­da­tions soft­ware along with our hol­i­day mar­ket­ing blitz was effec­tive. Can you please break down those num­bers to expose how much of this increase can be assigned to our efforts? Also, given that we were net­ting larger sales than pre­vi­ously, are those sales from return­ing cus­tomers or new cus­tomers? Can you also tell me what per­cent of our traf­fic is from cus­tomers and what per­cent is from prospects? What are the dif­fer­ences in user behav­ior in our cus­tomers when they were prospects and prospects that never con­vert? If we can iden­tify cus­tomer behav­ior that we can apply to prospects via tar­get­ing, then we can raise KPI’s.” Arte­mus fired off his stan­dard “Let me look into those requests and put some­thing together for us to review” then sat qui­etly for the rest of the meet­ing try­ing to stay focused on what was being dis­cussed in the meet­ing and not on build­ing his analy­sis frame­work in his head.

The abil­ity to divide datasets into cus­tomers and prospects has been a chal­lenge for many Web Ana­lyt­ics prac­ti­tion­ers since the early days of our rapidly evolv­ing field. Ana­lyz­ing the customer’s behav­ior when they were prospects has been even more of a chal­lenge. Mar­keters want to under­stand the dif­fer­ences in acqui­si­tion pat­terns as well as behav­ior that exist between these two key groups. Under­stand­ing and expos­ing those dif­fer­ences allow them to manip­u­late prospects via con­tent tar­get­ing and per­son­al­iza­tion tech­niques to achieve their spe­cific busi­ness objec­tives. Manip­u­lat­ing the out­come by rais­ing aver­age order value, increas­ing aver­age order size or some other unstated objec­tive is made pos­si­ble through seg­mented analysis.

Lever­ag­ing Adobe Insight these chal­lenges have been over­come. The trans­for­ma­tional capa­bil­i­ties of Adobe’s Multi-Channel solu­tion, allow the ana­lyst to iden­tify the point in time when a vis­i­tor, or prospect, becomes a cus­tomer. This opens the doors for dif­fer­ent kinds of insight­ful analy­sis. The ana­lyst is eas­ily able to divide the dataset into the two groups, cus­tomers and prospects and wit­ness the changes in size of those groups over time. From there, addi­tional seg­ments can be cre­ated and ana­lyzed within the par­ent seg­ment clas­si­fi­ca­tions of cus­tomer and prospect.

Below we see how Arte­mus has divided his multi-channel dataset into two major vis­i­tor types: cus­tomers and prospects. This enables him to quickly trend the break­out of these two groups over time as well as to apply any addi­tional met­rics against each group for fur­ther analy­sis. What is revealed is inter­est­ing. In the weeks lead­ing up to the hol­i­day break, Arte­mus notices that the per­cent­age of known cus­tomers vis­it­ing the site remains rel­a­tively con­stant. This is very encour­ag­ing as he sees this as an unan­tic­i­pated ben­e­fit from their reten­tion efforts. The week of 12/10/2012 high­lights this in a unique way, although over­all Web Vis­i­tors decline week over week, the per­cent of customer’s remains con­stant. This leads him to make a note in future analy­ses to focus on multi-purchase cus­tomers and com­pare their buy­ing cycle with that of sin­gle pur­chase cus­tomers to expose the most com­mon dura­tions and apply those to known sin­gle pur­chase cus­tomers in sub­se­quent visits.


This func­tion­al­ity also allows the ana­lyst to break­down the sep­a­rate prospect and cus­tomer behav­ior. Under­stand­ing the behav­ior of your cus­tomers in their vis­its prior to their first con­ver­sion visit is impor­tant. This analy­sis yields insight into how much time elapses between the ini­tial visit and the pur­chase visit as well as what con­tent aids in the sub­se­quent con­ver­sion visit. Under­stand­ing the behav­ior of cus­tomers when they were prospects also allows for mar­keters to bet­ter under­stand the impact of mar­ket­ing efforts, that may not be tied to last click con­ver­sion and thus unknown in most analy­sis sets. Iso­lat­ing “assist­ing” con­tent as well as iso­lat­ing behav­ioral nuances can be exploited dur­ing tar­get­ing and user design test­ing. Adobe Insight allows for these seg­ments as well as addi­tional an more refined seg­ments to be lever­aged in Adobe Test&Target thus com­plet­ing a full multi-channel opti­miza­tion system.

Drilling into the behav­ior of cus­tomers dur­ing non-conversion vis­its that occurred prior to the first con­ver­sion gleans great insight into

  • The amount of research cus­tomers need prior to conversion
  • The quan­tity of vis­its the aver­age cus­tomer makes prior to conversion
  • The types of con­tent a cus­tomer con­sumes prior to conversion

Under­stand­ing these aspects and the cus­tomer buy­ing cycle opens the door for con­tent tar­get­ing and also user design enhance­ments to decrease the amount of time between the ini­tial non-conversion visit and the first con­ver­sion visit. These tech­niques can also be applied to known cus­tomers in efforts to increase addi­tional purchases.

In order to under­stand some quick insights into the dif­fer­ences between cus­tomer and prospect behav­ior, Arte­mus needs to under­stand some basic infor­ma­tion on entry and exit con­tent. Below he sees that most prospects come in through the home page, which is in line with the land­ing page strat­egy tied to most of the field mar­ket­ing efforts. How­ever, an unex­pected dis­cov­ery is that Face­book is account­ing for the top two entry pages for cus­tomers by per­cent­age dur­ing the period being ana­lyzed. This is exactly the small nugget he can report and track to sup­port the mar­ket­ing efforts of the Social Media team. The hid­den gem he finds is that the Women’s sec­tion title page is the third most pop­u­lar entry page by % for cus­tomers. This inter­est­ing insight reveals a demo­graphic trend that is counter to non-seasonal trends.

The most glar­ing take-away for Arte­mus is not some­thing that he wants to report up but rather a con­cern that he needs to address with the user design team. The top exit page for prospects is a retrieve pass­word page, which is a page that a prospect should never be pre­sented with. Arte­mus is curi­ous how prospects are reach­ing an authen­ti­ca­tion point in their vis­its and makes a note to run path analy­sis includ­ing this page on the prospect seg­ment. Addi­tion­ally, he wants to under­stand what prospects are search­ing for that is caus­ing the no results pages to be the sec­ond high­est exit page. The final take­away he sees is that cus­tomers exit at a very high rate on the order sat­is­fac­tion sur­vey pages, this is typ­i­cal but he makes a note to dis­cuss any pos­si­ble incen­tives to decrease this as an exit point post con­ver­sion with the added bonus of hav­ing more com­pleted surveys.


The next step for any ana­lyst wit­ness­ing these trends in the data would be path analy­sis to deter­mine the most com­mon paths for prospects to the authen­ti­ca­tion point as well as look­ing at spe­cific inter­nal key­words to iden­tify what key­words are return­ing No Results. This small exam­ple high­lights how dif­fer­en­ti­at­ing between cus­tomers and prospects can be lever­aged in mul­ti­ple ways. With the unlim­ited cor­re­la­tion and seg­men­ta­tion capa­bil­i­ties of Adobe Insight, ana­lysts are empow­ered to quickly drive the analy­sis in any direc­tion the data is suggesting.

Under­stand­ing your customer’s behav­ior is the first step towards enrich­ing all future com­mu­ni­ca­tions. Cre­at­ing high-level seg­ments to dif­fer­en­ti­ate cus­tomer groups allows the ana­lyst to expose the hid­den nuances of those seg­ments behav­iors. Once cus­tomer seg­ment behav­ior has been defined the doors for user design enhance­ments, con­tent tar­get­ing, per­son­al­iza­tion and reten­tion efforts are thrown wide open. Adobe Insight allows for the rapid query of vast arrays of data over mul­ti­ple online as well as offline data sources. This enables the ana­lysts to begin their analy­sis broad and shal­low and then quickly go deep and nar­row to unearth action­able data points.

In the com­ing months will be writ­ing more about the pow­er­ful analy­sis capa­bil­i­ties your busi­ness can take advan­tage of today. We will explore, in depth, Adobe’s ana­lyt­ics tech­nolo­gies and solu­tions. We will lay out the process for lever­ag­ing these solu­tions in Adobe Insight and offer some action­able strate­gies for imme­di­ate results in your busi­ness. Stay tuned for the ongo­ing cul­ti­va­tion of our hero Arte­mus Brown as he evolves from Ana­lyst to Web Ana­lyt­ics Action Hero.