nowucca.com - personal software technology blog
Let’s expose the benefits of experiments (AB or “multivariate” testing) to the consumers of technology apps. The idea is: enable customers to get deeper insights beyond straight data-amalgamation insights (e.g. “20% of your profile views came from Great Britain”). This could be a fruitful avenue to help customers discover insights using tech platforms.
This really works well for technology companies that both store some of your information and make that information public.
I can think of quite a few things I’d like to set up AB tests for on social networks:
- LinkedIn profile headline : which headline generates more views, more interactions, more InMails?
- Facebook : Identify post sentiment and track groups of friends: which sentiments for which group generates more likes, more comments?
- Twitter : Schedule the same tweet for release at different times to different populations: which generates more favorites, re-tweets?
The main problem with this idea is that one usually needs many samples (e.g. many page views) to allow conclusions to be drawn with confidence. As long as these details are honestly exposed, people may still discover interesting results and have a shot at applying results carefully, with the appropriate amount of “belief” in their results.
This idea of using experiments is already heavily exploited by companies that display information to you from other sources (e.g. Netflix, PayPal, other news sites). There are even companies like Optimizely to make it easier for enterprise customers.
So what do you think? Let’s get the power to define and run experiments into consumer hands.