One of the largest pushes in our indus­try over the last year is to cre­ate a mas­sive per­son­al­iza­tion scheme on your web­site. Appar­ently peo­ple missed the point of Minor­ity Report, because the movie mocked this behav­ior and people’s assump­tions about what peo­ple are going to do. Time and time again, I hear about clients who are sold on “per­son­al­iza­tion” who have built out mas­sive 35 and 48 point schemas about where and what they are going to tar­get their per­son­al­ized con­tent, only to have them be shocked when I talk about how inef­fi­cient that is. With­out fail, the clients who get the worst return on their opti­miza­tion efforts are the ones that push full steam down this boon­dog­gle with­out apply­ing dis­ci­pline or feed­back into their efforts. Dynamic expe­ri­ences built to meet user needs are an amaz­ingly effec­tive tool, but in order to get that value, you must first tackle your own assumptions.

You are mak­ing three mas­sive assump­tions when you just run down this road:

1) Assum­ing you know WHO mat­ters most
2) Assum­ing you know WHAT type of changes mat­ter to them
3) Assum­ing you know WHERE to make those changes

I would add a 4th one:

4) Assum­ing that the full cost of per­son­al­iz­ing your web­site is much less then it really will be.

Who are you tar­get­ing to?

Often groups want to push out based on a site behav­ior or behav­iors to try and cre­ate a more “dynamic per­son­al­ized expe­ri­ence based on what the cus­tomer has already declared their intent is”. Things like chang­ing the main ban­ner based on the num­ber of vis­its or an engage­ment scor­ing sys­tem. Groups will sit around in giant meet­ings try­ing to come up with the per­fect scheme, which is counter pro­duc­tive to get­ting the very value that those meet­ings are hop­ing to achieve. Just decid­ing on what you are going to do com­pletely misses the point in that the same per­son can be looked at 100 dif­fer­ent ways if you so choose. That same per­son came from some­where, looked for some­thing, used some sort of browser, dur­ing some day of the week, at some time, with some sort of prior his­tory… You can choose to tar­get to ANY of those things, which means that you need to fig­ure out a way to mea­sure the value of each one AGAINST each other.

This gets us to the point of what I usu­ally refer to as an “exploitable seg­ment”, one that has shown through CAUSAL data that it cre­ates a change in user out­come based on a change in the user expe­ri­ence. This is one of the many dif­fer­ences in the dis­ci­plines of ana­lyt­ics and opti­miza­tion, since in the world of cor­rel­a­tive data; you are look­ing for groups that have a dif­fer­ent behav­ior. In opti­miza­tion, we are look­ing for groups who CHANGE their behav­ior based on a test. This means that we don’t care that peo­ple who come from search spend half as much time on your site as peo­ple who come straight to the site, we only care if those two groups have a dif­fer­ent “win­ner” for var­i­ous test results. If the same thing wins for both, they may both have a dif­fer­ent propen­sity of action, but you gain noth­ing from chang­ing the user expe­ri­ence rel­a­tive to each other; the same thing helps both.

In the worst case, groups start out by talk­ing together or let­ting one per­son come up with a con­cept or schema that they are just sure will work. In the next step up we often we deal with groups strictly rely on CORRELATIVE data to try and answer this prob­lem. Unfor­tu­nately, cor­rel­a­tive data can’t tell you the value of an action nor the effi­ciency of an action. The best it can do is give you some insight into the rate of action and the like­li­hood of an out­come as long as you do not fac­tor in cost or effi­ciency. Ana­lyt­ics data can tell you what the rate of action is for the search peo­ple, but it can­not answer if their behav­ior will change based on a dynamic expe­ri­ence. You need to use causal data, the value of the changes rel­a­tive to each other, to really dive in and dis­cover what groups are actu­ally going to change their behav­ior and that you can lever­age to improve site performance.

What Mat­ters?

I can­not stress this enough… a user expe­ri­ence is more than ban­ner con­tent or copy. Usu­ally the largest and the longest last­ing changes to sites are based on real estate, or chang­ing the spa­tial dynam­ics of a page either through lay­out, inclusion/exclusion of items, rel­a­tive posi­tion­ing of items. Your other options are changes to pre­sen­ta­tion (how some­thing looks) or func­tion (how it inter­acts or how it is pro­gramed) and then copy/content. All of these things can be changed, and fig­ur­ing out the order of value is vital to effi­ciency. One of the very first things any pro­gram needs to do is fig­ure out the value of dif­fer­ent types of changes rel­a­tive to each other. You have to think in terms of the entire user expe­ri­ence and then fig­ure out what changes, rel­a­tive to each other, are most valu­able. Even if we are lim­ited to one con­tent area, we can look for effi­cien­cies of scale by focus­ing on the com­po­nents and rules of the con­tent over the indi­vid­ual con­tent item, in order to gain returns on oth­er­wise per­ish­able changes.

Where should I personalize?

Most teams walk into per­son­al­iza­tion with the want to tackle the entire site in one blast, to have the same type of per­son­al­iza­tion on the front door, land­ing pages, prod­uct pages, in cart, in the right gut­ter, every­where where they can fit it in. In most cases, even if one of those or mul­ti­ple of those spots is pos­i­tive, the over­abun­dance is inef­fi­cient to the point of coun­ter­act­ing a lot of the good that may come from putting per­son­al­ized expe­ri­ences up on the site. It also fails to account for the fact that maybe those spots shouldn’t exist? Or maybe dif­fer­ent func­tions on the site need dif­fer­ent types of “per­son­al­iza­tion”? Or maybe the most per­son­al­ized thing you can do is move items on the page to increase the effi­ciency of the user flow. Like every­thing else, it requires study and dis­ci­pline to fig­ure these things out.

So often the first thing I do when work­ing with groups is ask them to prove out the value of what they are tar­get­ing; with­out fail what they are doing has no pos­i­tive value, and can often times be hurt­ing per­for­mance. You have to be dis­ci­plined to get real, long term value, from your “per­son­al­iza­tion” pro­gram. Opti­miza­tion gives you the abil­ity to mea­sure effi­ciency and the value of items rel­a­tive to each other, which means the same tool you want to tar­get with can give you so much more, with almost no effort, sim­ply by being will­ing to ask this fun­da­men­tal ques­tion. We are try­ing to learn about what mat­ters most, not try to be “right”. Noth­ing is more valu­able then when you are “wrong” about your assumptions.

Like so many things in the past and that will be, per­son­al­iza­tion is the next buzz­word bingo item that has caught the atten­tion of the online world. It doesn’t mean that the con­cept is not valu­able or is not some­thing you should strive for. Some of the sites that get the most value get it from dynamic tar­geted user expe­ri­ences. What it does mean how­ever is that you can­not jump in with­out under­stand­ing the dis­ci­pline it takes to achieve real long term suc­cess. With­out being will­ing to go down the path of learn­ing, almost all efforts are doomed to failure.

I fully encour­age you to find mean­ing­ful, exploitable, dynamic user expe­ri­ences. You need to work to make your site a liv­ing breath­ing thing that shifts to meet new needs and is some­thing that dif­fer­ent peo­ple get dif­fer­ent things from. What has to hap­pen though, is you need to tackle each step with the atten­tion that it requires and to apply dis­ci­pline to reach mean­ing­ful results. You can­not just guess your way to vic­tory, but you can get there eas­ily if you are will­ing to answer key ques­tions through action and allow the results to dic­tate the path you are on, not your own ego.