A/B testing -> permanent segmented UIs
A/B testing is used to experiment with two alternatives and gather empirical evidence about which version performs better. It aims to result in a singular decision through quantitative analysis. What happens if we shift the goal? Rather than a singular outcome, can our products exist in a state where a single choice does not have to be made for our entire customer set, but rather for a specific category of customers?
Note: This post isn’t all the well structured, and more of a collection of random thoughts / questions I had that are loosely coupled. This concept may already exist!
I was thinking about using A/B testing to decide how product features work, and what happens if there is an even split? What if, based on customer segments, specific designs are used - i.e., older customers may see a simpler layout where younger, more technically ready customers see a complex layout. I have to imagine that this is already done at places like Facebook, where they have a remarkable amount of demographic data and can likely slice their customer segments to run experiments against. Even to a point this is done in web development frameworks like Rails, where we can build out different internationalization translations for different customers - at it’s basic concept, this is modifying data based on who a customer is.
As a company thinks about going international, they need to worry about things like language direction, date presentation format, maybe expected UI designs based on other popular technologies in that area. For different parts of the world I’m sure that there are different technologies that are uniquely popular in that location. There are certainly different expectation when it comes to technology interaction, use, purpose and familiarity. Is it possible to build a robust application that can dynamically adjust the UI of a view, whether it be a web page or a mobile application view, based on the customer demographic or need? What issues what that cause for developers?
When dealing with a large enough user base it may not be correct to think that one size fits all for certain consumer products. Think tax software, or data aggregation software, or even social network feeds. Much work has probably been done in dynamic content generation (a la Facebook Newsfeed based on someone’s particular interests), but what about dynamic UI generation?