Use Adobe Campaign for Predictive Modeling and Propensity Scoring

Today demands on mar­keters for audi­ence acqui­si­tion go beyond seg­men­ta­tion based on sole­ly on behav­ior and intent. They are search­ing for net new prospects who might act in accor­dance to the mar­ket­ing pro­gram goals. Tight mar­ket­ing bud­gets and faster deliv­ery to mar­ket, make it dif­fi­cult for mar­ket­ing depart­ments to hire and build a Quants Team, to deliv­er attri­bu­tion mod­els, and lead scor­ing algo­rithms.

Adobe Mar­ket­ing Cloud prod­ucts pro­vide attri­bu­tion mod­els based on indus­try best practices. Adobe Ana­lyt­ics has Pre­dic­tive Mar­ket­ing mod­ule that helps ana­lyst do every­thing from anom­aly detec­tion, clus­ter­ing audi­ence and assign­ing propen­si­ty scores with­in Data Work­bench. An audi­ence that is tai­lored with­in Work­bench can then be tar­get­ed using oth­er Adobe Solu­tions includ­ing Tar­get. More infor­ma­tion here.

If you want to use your own ML engine and data that is already in Cam­paign or if you don’t have or don’t want to con­nect Ana­lyt­ics to Adobe Cam­paign, then there is an option with­in Adobe Cam­paign the Mar­ket­ing Ana­lyt­ics mod­ule. Adobe Cam­paign is an automa­tion tool that helps you tar­get, seg­ment, deliv­er com­mu­ni­ca­tions to cus­tomers across mul­ti-chan­nel jour­neys and allows mar­keters to work on pre­dic­tive audi­ence infor­ma­tion with­in Adobe Cam­paign using a sim­ple tem­plate approach.

Follow these steps

To cre­ate pre­dic­tive mar­ket­ing mod­els, go to the Pro­files and Tar­gets > Jobs > Pre­dic­tive mod­els node of the Adobe Cam­paign con­sole tree.

Predictive Model

Predictive Model creation

In the Mod­el­ing tab, you can insert data to enable the tem­plate to pre­dict cus­tomer and prospect behav­ior.


The algo­rithm lets you build a mod­el to define a tar­get val­ue based on a set of vari­ables. The engine uses the fol­low­ing analy­sis tech­niques: Weka and Kxen engines. Like any oth­er cam­paign tem­plate, the pre­dic­tive tem­plate works on tar­get dimensions. Use the Add but­ton to select each vari­able you want to insert into the mod­el. These are the fields of the table which the analy­sis will con­cern


Once the mod­el is designed, you can see the pre­dic­tion cal­cu­lat­ed by the mod­el.                                             



Once the Mod­el tem­plate is in place, you can to use the pre­dic­tive intel­li­gence in  cam­paign work­flows with an out of the box com­po­nent called Pre­dic­tive Split. The Pre­dic­tive Split activ­i­ty lets you split a defined pop­u­la­tion using a pre­dic­tive mod­el that we designed ear­li­er.

The pre­dic­tive work­flow should look some­thing like this:predictive-workflow

And after the Pre­dic­tive Split, you can add any deliv­ery that you want. Real world cus­tomer jour­neys are com­plex but pre­dic­tive split can be help­ful and sim­ple way of upgrad­ing your cam­paign work­flows to have pre­dic­tive intel­li­gence.