The over­ar­ch­ing goal of test­ing and opti­miza­tion is to effec­tively seg­ment vis­i­tors and uncover prof­itable tar­get­ing oppor­tu­ni­ties, elim­i­nat­ing the guess­work involved in the per­son­al­iza­tion of the cus­tomer expe­ri­ence in order to max­i­mize con­ver­sion and return on invest­ment (ROI). Adobe Tar­get, the opti­miza­tion and per­son­al­iza­tion solu­tion in Adobe Mar­ket­ing Cloud, pro­vides real-time mar­keter access, con­trol, and insight in test­ing and opti­miza­tion pro­grams. Tar­get assists mar­keters in not only see­ing and cap­i­tal­iz­ing on real-time tar­get­ing oppor­tu­ni­ties based on best-performing con­tent at a gran­u­lar level, lever­ag­ing all avail­able data, but also man­ag­ing and grow­ing a pro­gram effi­ciently based on Adobe’s proven matu­rity model, indus­try suc­cesses, and best practices.

Even in a full-service model, where Adobe Con­sult­ing man­ages the entire pro­gram (deliv­er­ing rec­om­men­da­tions and results from the start), the con­cept is to edu­cate and pro­vide your team mem­bers the oppor­tu­nity to have own­er­ship over their pro­gram. Ini­tial quick wins are evan­ge­lized, fur­ther pro­mot­ing a test­ing and opti­miza­tion cul­ture through­out the busi­ness and even­tu­ally, when team mem­bers under­stand the strat­egy, tech­niques, and exe­cu­tion of test­ing and opti­miza­tion, they can assume a self-service model and run the pro­gram themselves.

Con­trol is a key fac­tor in dri­ving the right opti­miza­tion based on your spe­cific busi­ness ini­tia­tives, as well as react­ing in real time to the trends you see in vis­i­tor data. This is why you should be skep­ti­cal when a point solu­tion pushes a full-service model as its sole option.  You should be wary of what appears to be, or is pre­sented as, “mar­keter con­trol” in a solu­tion when it really is not. For exam­ple, true mar­keter con­trol is not  pro­vided when a point solu­tion has you fill out a sim­ple form field request in the test design that then requires a ven­dor to code and to exe­cute doc­u­ment object model (DOM) manip­u­la­tion behind the scenes (essen­tially relin­quish­ing con­trol and exe­cu­tion to an out­side ven­dor for ques­tion­able return).  The real rea­son why point solu­tions will often push a full-service model is that their prod­uct and user inter­face (UI) are cum­ber­some and con­fus­ing. These “solu­tions”  often rely on cus­tom cod­ing and devel­op­ment work for any­thing above a basic but­ton color change, text swap, or sim­ple test on a sta­tic ele­ment.  Even more mar­keter con­trol is lost in this sce­nario because most web­sites are not built or coded cleanly enough to enable easy test cre­ation with solu­tions that pro­vide lim­ited DOM manip­u­la­tion in their “what you see is what you get” (WYSIWYG) editors.

These providers also lack expe­ri­ence and are not con­fi­dent in their abil­ity to teach or to train you in the best meth­ods for test­ing, opti­miza­tion, and pro­gram growth.  Adobe Tar­get is designed for grow­ing a pro­gram, and is built on a foun­da­tion of our proven matu­rity model, which guides busi­nesses from basic inci­den­tal test­ing to advanced per­son­al­iza­tion and an embed­ded opti­miza­tion pro­gram.  Our Adobe Tar­get Solu­tion also pro­vides guardrails within the visual onsite edi­tor n that allow for less reliance on cod­ing for more dynamic ele­ments, with easy instruc­tions on how to exe­cute more advanced test­ing and tar­get­ing sce­nar­ios with­out nec­es­sar­ily requir­ing cod­ing or devel­op­ment resources.

There is a fun­da­men­tal con­cern for the accu­racy of your tests when con­trol is relin­quished in your opti­miza­tion pro­gram. How­ever, this fear is unfairly extended to the use of automa­tion, per­haps because the algo­rithms pro­vided for automa­tion by most ven­dors are deliv­ered as a “black box.”  In this sce­nario, algo­rithms are merely selected by the mar­keter with lit­tle or no cus­tomiza­tion capa­bil­ity and run “beneath the hood,” obscur­ing mar­keter insight into how or why the solu­tion is ren­der­ing a spe­cific decision.

A major dif­fer­en­tia­tor for the Tar­get solu­tion, in terms of our out-of-the-box rec­om­men­da­tions algo­rithms, is our com­mit­ment to enabling the mar­keter to not only cus­tomize the loca­tion and data being lever­aged for real-time deci­sion­ing, but to run tests to deter­mine which cus­tom algo­rithm per­forms the best in dif­fer­ent sce­nar­ios. This abil­ity will be extended with major updates to our rec­om­men­da­tions UI and algo­rithms begin­ning in August.

Our mod­el­ing sys­tem, which was renamed auto­mated per­son­al­iza­tion with our Adobe Tar­get Pre­mium June 25 launch, also pro­vides insight into how it self-learns, within its resid­ual vari­ance frame­work, uncov­er­ing the most pre­dic­tive vari­ables to lever­age for per­son­al­iz­ing an expe­ri­ence to an indi­vid­ual based on a series of dis­tinct pos­si­ble con­tent vari­a­tions or offers. The reports auto­mat­i­cally gen­er­ated by the mod­el­ing sys­tem are also improved with the new Pre­mium release. They are much more graphic, and pro­vide detailed, advanced seg­ment dis­cov­ery by test­ing and deter­min­ing what are, or just as impor­tantly what are not, the rel­a­tive influ­ence or pre­dictabil­ity of hun­dreds of dis­tinct pro­file vari­ables. In addi­tion, we offer a detailed descrip­tion of how our mod­el­ing sys­tem self-optimizes by con­stantly test­ing its effec­tive­ness to ensure that it adapts to trends and changes in cus­tomer behav­ior for the most effec­tive per­son­al­iza­tion to the indi­vid­ual.  This level of accu­racy safe­guard is not found in point solu­tions that claim to offer “auto­mated behav­ioral tar­get­ing,” nor is there the depth of pro­file data and vari­ables pro­vided by other solu­tions that can be found with our uni­fied mas­ter mar­ket­ing pro­file (a mod­el­ing sys­tem is only as intel­li­gent as the data fed into it). This includes data syn­chro­niza­tion across Adobe Mar­ket­ing Cloud solu­tions along with a series of APIs that pro­vide ease of reg­u­lar data import and updating.

Adobe Tar­get Pre­mium lets users mea­sure and opti­mize its auto­mated per­son­al­iza­tion offer­ing in much the same way that they can test and opti­mize the out-of-the-box rec­om­men­da­tions algo­rithms.  You can now choose between resid­ual vari­ance and ran­dom for­est algo­rithms, which pro­vides an alter­na­tive auto­mated sta­tis­ti­cal tech­nique to fine-tune the best automa­tion for more sce­nar­ios across more of your dig­i­tal prop­er­ties to free you up for those oppor­tu­ni­ties where man­ual or rules-based tar­get­ing con­trol is war­ranted. More algo­rithms will be intro­duced shortly for even fur­ther flex­i­bil­ity and man­aged algo­rithm and automa­tion control.

When you can har­ness var­i­ous meth­ods of machine learn­ing now avail­able in Adobe Tar­get Pre­mium, you can iden­tify oppor­tu­ni­ties for per­son­al­iza­tion and max­i­miz­ing vis­i­tor engage­ment more quickly, increas­ing con­ver­sion and loy­alty while not relin­quish­ing the under­stand­ing, con­trol, and con­fi­dence over how these pow­er­ful tac­tics are being used to drive even quicker suc­cess for you and your program.

0 comments