A Tale of Three Dataset Schemas [Analysis with Insight]

When the solu­tion now called Omni­ture Insight became part of the Omni­ture Suite last year, mar­keters gained the pow­er­ful oppor­tu­nity to inte­grate exist­ing web (Site­Cat­a­lyst) data with other data points, includ­ing offline data. Many Omni­ture cus­tomers have already taken advan­tage of this oppor­tu­nity and are prov­ing that the “sky is the limit” as far as data they can inte­grate and uti­lize in this fast, visual analy­sis tool.

Although there are some ready-to-use imple­men­ta­tions that meet com­mon Key Busi­ness Require­ments (KBRs), each Omni­ture Insight instal­la­tion is at least some­what dif­fer­ent, dic­tated by the unique busi­ness needs and sit­u­a­tions of the client.

Let’s briefly take a look at three inte­gra­tions of offline and online data and dis­cuss the dataset archi­tec­ture that was employed and busi­ness needs that were solved – we’ll refer to the rela­tion­ship between the lev­els of data as the “schema.”

1) Live Online Ad Integration

One pop­u­lar use of Omni­ture Insight is to com­bine Site­Cat­a­lyst click stream data with live online ad impres­sion infor­ma­tion. In this case, ad impres­sion infor­ma­tion can either be loaded as an offline data source (a data file com­ing on a recur­ring basis from your ad server) or col­lected live, in real-time.
Using one of sev­eral meth­ods to tie the Site­Cat­a­lyst cookie ID to the ad server cookie ID, ad impres­sion and click data can either be com­bined within the “Vis­i­tor”, or another level of data for a “Cus­tomer” can be added above the “Vis­i­tor” level.

The for­mer (com­bin­ing “Ad Visit” and “Ad Impres­sion” data) is the most com­mon, and is illus­trated below.

A “Vis­i­tor” could have one or many “Ad Vis­its” (business-defined peri­ods of activ­ity with online adver­tis­ing.) Each of these Ad Vis­its could have one or many “Ad Impres­sions.” Any attribute of an Ad Impres­sion can be col­lected and made avail­able for analy­sis and for selec­tion – attrib­utes like “ban­ners with red text” as opposed to those with blue text, ads on one site ver­sus another site, ads in one size or place­ment ver­sus another.

Also included in the dataset, as illus­trated in the schema, are the stan­dard Site­Cat­a­lyst ele­ments, like Vis­its and Page Views.

As a result, a selec­tion of an ad attribute to which a set of Vis­i­tors was exposed quickly makes avail­able the infor­ma­tion related to their on site Vis­its, Page Views, Con­ver­sions, etc.

Con­versely, a selec­tion of cer­tain Con­ver­sion attrib­utes, prod­ucts pur­chased, or other site-based attrib­utes yields the oppor­tu­nity to ana­lyze ad attrib­utes that con­tributed the most – or the least – to those events.

2) Offline Retail Data

Another grow­ing use of Omni­ture Insight is for offline retail data. At its sim­plest view, you might con­sider the web-related lev­els of data we’re accus­tomed to (“Vis­i­tor -> Visit -> Event”) and their retail-world par­al­lels (“Cus­tomer -> Trans­ac­tion -> Item”.) A Cus­tomer to your retail store or chain might have one or many Trans­ac­tions. Each of those Trans­ac­tions might have one or many Items.

As in the above exam­ple, each level of the dataset can have its own attrib­utes that can be selected or ana­lyzed. For exam­ple, you might have infor­ma­tion on a Customer’s loy­alty type, level or sta­tus, demo­graphic infor­ma­tion, geo­graphic loca­tion, and more. For Items, you might have infor­ma­tion on brand/supplier, price, cost, and mar­gin, and loca­tion in store.

Select­ing a seg­ment of Trans­ac­tions (for exam­ple, trans­ac­tions that included choco­late candy bars) allows quick analy­sis of what other Items were most com­monly in the bas­ket in those Trans­ac­tions (“bas­ket analy­sis”.) Select­ing the same seg­ment of Trans­ac­tions and then select­ing latent trans­ac­tions — before or after – allow insight into what Cus­tomers pur­chased in prior or sub­se­quent Trans­ac­tions, and how long it was between those Transactions.

Again, any attribute at any level can be selected or ana­lyzed for insight into its rela­tion­ship with or impact upon other attrib­utes at other levels.

3) Com­bined Offline Retail & Online Mar­ket­ing Data

One of the most excit­ing areas of inte­gra­tion has been at the inter­sec­tion of this online and offline data.

One such exam­ple includes a full com­bi­na­tion of online and offline data – with online Vis­i­tor IDs “keyed” to offline Cus­tomer IDs to allow a com­pre­hen­sive view of the customer.

Web data (green in the below schema) is col­lected in Site­Cat­a­lyst and avail­able to web chan­nel ana­lysts in that tool, but is also sent to Omni­ture Insight to be com­bined with all other offline & online data for the Customer.

In many of these cases, Offline/Online data includes:

• All Web site traf­fic (Vis­its, Events, Con­ver­sions, etc.)
• All Cus­tomer demo­graphic and loy­alty data
• All Online Ad impres­sions, attrib­utes, etc.
• All Offline (Direct Mar­ket­ing) impres­sions for the Customer
• All result­ing Trans­ac­tions (which, for this imple­men­ta­tion, occur offline) and Items

In one com­bined analy­sis tool, busi­ness users are able to ana­lyze on or make selec­tions based upon Items in the final pur­chase, mar­gin in that pur­chase, pro­mo­tions applied to the pur­chase, loca­tion of the pur­chase – as well as attrib­utes of the Cus­tomer: the ad or direct mar­ket­ing expo­sures the cus­tomer had, and what they expe­ri­enced and/or reserved on the web­site prior to the final purchase.

Imag­ine the power of full analy­sis and attri­bu­tion across chan­nels from first impres­sion through con­ver­sion, and beyond — all in one tool.

If you believe you could ben­e­fit from the offline data inte­gra­tion or high data vol­ume visual analy­sis capa­bil­i­ties by adding Omni­ture Insight to your suite, talk to your Omni­ture con­sul­tant or account man­ager today.

Have a ques­tion about any­thing related to Omni­ture Insight?  Do you have any tips or best prac­tices related to Omni­ture Insight you want to share?  If so, please leave a com­ment here or send me an e-mail at mhalbrook@​omniture.​com and I will do my best to answer it here on the blog so every­one can learn. (If you pre­fer, I won’t use your name or com­pany name.) You can also fol­low me on Twit­ter @Michael­Hal­brook.

Learn more about Omni­ture Con­sult­ing
Learn more about Omni­ture University