During eharmony’s sign up, subscribers take a 200 page questionnaire that determines who they’re matched with. People are matched based on 18 psychological affects, or “dimensions of compatibility.” Subscribers see two to three matches per day from that pool.
In 2016 we lost nearly 50% of our subscribers. Because our subscriber-base shrank, people stopped recieving matches, or received no matches at all.
Executives assigned a UXR (Morgan) a designer (me) and a PM (Francine) to solve this problem.
Morgan (UXR) and I started with qualitative interviews and learned that subscribers weren’t happy with their matches because they weren’t getting enough (of course) but also because they didn’t match superficial traits like age, height and employment. These led to HMWs (how might we’s?) that guided a cross-functional brainstorm.
From the brainstorm, the team came up with the idea of showing a compatiblity score. Christina could see members who aren't compatible with her, but do fit her superficial preferences. She could also see why she's matched with the less attractive people. This would solve the problem of “not enough,” but also, our hypothesis was that this could improve the perception of quality.
To validate that subscribers were interested, I worked with Morgan (UXR) to run a feature desirability survey.
For the first iteration, I showed Christina how she aligns on each dimension by doing comparison bar graphs.
There were some problems with this approach, as you could accidentally insult someone by revealing that they are emotionally unstable, or confuse them. For example, if you are both ranked “low” in intellect, this is a good thing because you are both fine being ignorant.
It took a lot of explaining to communicate what was “good” or “bad.”
Compatibility launched on iOS, Android, mobile and desktop web in December, 2016. It was a huge success for the company, growing the matching pool by 30% and decreasing the number of users with 0 matches by 20%. In our feedback tool, people said they “loved the compatibility read-outs.”
I wish we had done post-launch usability with real data and people. We knew from our feedback tool that people liked the section – not totally sure how it affected their perception of quality.
I also wonder if the single score was too obscure. I wanted to keep tweaking the experience, but because the project was deemed successful, the company wanted us to move on.