Segmentation: The party game

Every Decem­ber, at the Omni­ture hol­i­day party, spouses, part­ners, and dates of employ­ees get a crash course in the value of some of our prod­ucts, cour­tesy of a “seg­men­ta­tion game.” Every­body stands up, and the emcee (CTO Brett Error) reads from a list of cri­te­ria and instruc­tions, such as, “If you have a five-dollar bill on your per­son, please sit down,” or “If you have ever been told that you look like a celebrity, remain stand­ing; every­body else, sit.” (When you’re asked to sit down, you’re out.) The idea is that even­tu­ally, through this some­times bizarre and always humor­ous seg­men­ta­tion, the group of a cou­ple thou­sand is whit­tled down to one winner.

My wife is not a web ana­lyst, mar­keter, devel­oper or IT pro­fes­sional. For her, the game is a wel­come intro­duc­tion to the idea that busi­nesses need to be able to break apart their online data to focus on key groups and demo­graph­ics. Site­Cat­a­lyst offers seg­men­ta­tion func­tion­al­ity, often through the use of cor­re­la­tions and sub­re­la­tions, but the real pow­er­houses of seg­men­ta­tion in the Online Mar­ket­ing Suite are Data Ware­house, Dis­cover, and ASI. Since most read­ers of this blog do under­stand the con­cept of seg­men­ta­tion, and why it is crit­i­cally impor­tant to online busi­ness opti­miza­tion, I’d like to dive a lit­tle bit deeper here and explain exactly how to build a cou­ple of com­monly used seg­ments to illus­trate how this process works.

Some back­ground

If you haven’t read Adam Greco’s post on seg­men­ta­tion, I highly rec­om­mend it. Espe­cially impor­tant is the fol­low­ing expla­na­tion of seg­men­ta­tion “containers”—Page Views, Vis­its, and Visitors—

1. Page Views — Drag­ging a Page Views con­tainer to the seg­ment can­vas will allow you to define which Page Views you would like to include or exclude from the seg­ment. When eval­u­at­ing the Page Views con­tainer, Omni­ture is, in effect, scan­ning through each page view it finds within the spec­i­fied time frame and decid­ing whether it should be included or excluded. There­fore, it may be the case that two dif­fer­ent page views from the same visit may or may not be included in the seg­ment. For exam­ple, let’s assume that you are build­ing a seg­ment where you only want pages where the lan­guage was Span­ish. It may be the case that a vis­i­tor viewed ten pages dur­ing their visit, but only two of those ten were viewed in Span­ish. Using a Page Views con­tainer, would mean that only these two page views would be included in the segment.
2. Vis­its — Drag­ging the Vis­its con­tainer to the seg­ment can­vas will allow you to define which vis­its you would like to include or exclude from the seg­ment. When eval­u­at­ing the Vis­its con­tainer, [the tool] is, in effect, scan­ning through each visit it finds within the spec­i­fied time frame and decid­ing whether the entire visit should be included or excluded. There­fore, if any of the cri­te­ria are met within the Visit, all data from that visit will be included (or excluded if using the exclude tab) in the seg­ment.  Using the pre­ced­ing exam­ple, if the seg­ment look­ing for pages viewed in Span­ish were built using a Visit con­tainer, the entire visit would be included since at least one of the pages was viewed in Span­ish (even though the major­ity were in English).
3. Vis­i­tors — Drag­ging the Vis­i­tors con­tainer to the seg­ment can­vas will allow you to define which Vis­i­tors you would like to include or exclude from the seg­ment. When eval­u­at­ing the Vis­i­tors con­tainer, Omni­ture is, in effect, scan­ning through all data it has for each Vis­i­tor within the spec­i­fied time frame and decid­ing whether at any time the vis­i­tor met the cri­te­ria. If it finds that the vis­i­tor has met the cri­te­ria, all Vis­its and Page Views for that vis­i­tor will be included in the seg­ment.  Con­tin­u­ing the pre­ced­ing exam­ple relat­ing to pages viewed in Span­ish, if a vis­i­tor had six site vis­its within the spec­i­fied time frame and in one of those vis­its viewed at least one page in Span­ish, data from all six vis­its would be included in the segment.

Here’s a visual rep­re­sen­ta­tion of this:

(I made it myself; can you tell?) If I were to seg­ment on value A using the Page Views con­tainer, I would get the two page views con­tain­ing “A” in the first visit shown above. If I were to seg­ment on value A using the Vis­its con­tainer, I would get the five page views in the visit where “A” exists. And if I were to seg­ment on value A using the Vis­i­tors con­tainer, I would get both vis­its, because the vis­i­tor did have a value of “A” at one point.

An extremely high per­cent­age of per­ceived issues that I’ve seen involv­ing the reports returned by Data Ware­house and ASI are based in a mis­un­der­stand­ing of con­tain­ers. Admit­tedly, they are com­plex and dif­fi­cult to explain. Adam did a great job clar­i­fy­ing exactly what scope of data is returned by each con­tainer. Users always need to con­sider this scope when build­ing a seg­ment; do you want the visitor’s entire his­tory, or just a rel­e­vant piece of it? Do you want all of the page views in a visit to be included, or just a por­tion of them?

Cam­paign Attribution

Pulling con­ver­sion data for a cam­paign track­ing code seems straight­for­ward enough. Most peo­ple prob­a­bly define the seg­ment as vis­its where track­ing code equals/contains [value]. This is cer­tainly a valid option. How­ever, some users are frus­trated when the result­ing reports con­tain data for track­ing codes that do not match the seg­ment. Some vari­a­tion on the ques­tion, “If I spec­i­fied track­ing code XYZ, then why am I see­ing con­ver­sions tied to track­ing code ABC in my report?” has been asked of me lit­er­ally dozens of times. The answer is that the visit con­tainer includes all data from any visit that matches the seg­ment. You may have spec­i­fied track­ing code XYZ, but both track­ing codes were ever passed dur­ing the same visit, then you will see both in your report. That might be really valu­able to you; it shows which cam­paigns are clicked dur­ing the same visit as your tar­get cam­paign. But in the event that you want to kick out any page views within the visit that don’t per­tain to your desired track­ing code, use the Page Views con­tainer to include page views where track­ing code equals/contains [value]:

You may be think­ing, “Won’t this only include the page views where s.campaign was set? What if I want to include page views where con­ver­sion occurred so that this data can be tied back to the track­ing code?” The seg­men­ta­tion tools are smart enough to include all page views where the given value was passed or where it per­sisted from a pre­vi­ous page view. So using the page view con­tainer will include all page views tied to the track­ing code, includ­ing those on which con­ver­sion happened.

Note that you can use “track­ing code is greater than or equal to [excla­ma­tion point]” to include any track­ing code in the seg­ment. The “greater than or equal to !” state­ment means “is not null” or “has any value at all.” (Con­versely, “less than !” returns only data where there was no track­ing code.) How­ever, keep in mind that the prin­ci­ples dis­cussed above are still rel­e­vant. If the user’s visit con­tained some page views where there was no track­ing code, and you seg­ment for vis­its where track­ing code is not null, the page views where there was no track­ing code will also be included.

Accu­rate visit/visitor counts in any container

Users often choose the vis­i­tor con­tainer think­ing that it is the only way to get an accu­rate unique vis­i­tor count (or the visit con­tainer to get an accu­rate visit count) for the seg­ment. This is untrue. Seg­men­ta­tion using the Page View con­tainer is often the most direct way to answer the given ques­tion, and it will return cor­rect vis­i­tor and visit counts. Data Ware­house doesn’t need all of the data from a user visit in order to attribute a visit to a vari­able value. For exam­ple, if you’re seg­ment­ing using the Page View bucket for rows where eVar1 equals “Blue,” and request the vis­its met­ric, you will get a cer­tain num­ber of rows where eVar1 equals “Blue.” Data Ware­house then exam­ines the vis­i­tor data for those rows and cal­cu­lates the num­ber of vis­its by find­ing the unique vis­i­tor ID and visit num­ber com­bi­na­tions. The num­ber of vis­its will not be higher using the visit con­tainer with the same seg­ment def­i­n­i­tion, because the num­ber of unique com­bi­na­tions will not be any dif­fer­ent (and the same holds true with the vis­i­tors metric).

Con­clu­sion

Let’s call this an early chap­ter in the book on seg­men­ta­tion in the Omni­ture Online Mar­ket­ing Suite. I rec­og­nize that this post may intro­duce a num­ber of ques­tions for some of you, and that’s okay. Please let me know what ques­tions or con­cerns you have, and I’ll address them in follow-up posts. As nerdy as it may sound, I love seg­men­ta­tion, and I would be happy to explain any of this fur­ther. As always, you can con­tact me via Twit­ter, Friend­Feed, LinkedIn, or by e-mailing omni­ture care [at] omni­ture dot com.

And for the record, I’ve never been told that I look like a celebrity, so if any­one can sug­gest one to whom I bear even the slight­est resem­blance, please let me know; it might help me win the seg­men­ta­tion game this year.