How we Designed for Reducing Duplicates Tweet #7
jayceeday
started this conversation in
Case Studies
Replies: 0 comments
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
-
This was originally a Twitter thread
This seems like a pretty boring small project, but it actually started pretty big and scoped it down a ton. I think this story is interesting because it’s not some beautiful dribbbly UI with a linear process.
A thread
We work with our community managers to improve how Feedback in Discussions works, here: https://github.com/github/feedback/discussions/categories/discussions-feedback
The problem was: there were a lot of duplicates.
Our (me and the wonderful at evi’s, my pm) initial reaction was to do something like this (pic) and show matching existing discussions when they typed the title, and maybe when they typed in the description.
We thought that this was probably an issue with issues (lol) at some point also, so we spoke with their team, and learned they had tried this before. The ux wasn’t ideal because of the moving text box, and the quality of the results were low, because our search isn’t amazing.
So bummer, but we decided that we couldn’t do it for these reasons.
I realized that because our search isn’t great, that even before this step, on the overview, users might not be finding the results they want in the first place. I suggested we improve search but this would be a HUGE increase in scope, so this is something we are looking into separately and not tackle in this project.
Another thing, is that this might be pretty intrusive when we don’t have that many duplicates, so how might we do something more subtle?
So that was our proactive approach, what if we try reactive? Evi and I met and she did these beautiful (but genius) sketches.
The idea was that you could merge duplicate discussions. This would have been quite a bit of work to build, combining upvotes, what to do with labels, which one is the “main” one, there was a lot of work. This would also be extra work for the maintainer. We decided against this, for these reasons. So double bummer.
The team came up with another idea: what if we just had a checkbox here with a link to search for similar discussions on the previous screen? We also would add a “Friendly reminder” to explain it. The checkbox would only appear on their first post.
I had a hunch that people would be annoyed by this, so I did a quick 5-10 minute user test on usertesting.com to see how users reacted to this and these were the results:
Users did not notice the checkbox right away or at all.
Only 1 user mentioned the "Friendly reminder" text.
A couple people didn't realize the check box was mandatory
People liked the nudge to suggest this!
All users expected it to bring them to a search w keywords
So there you have it, within 24 hours I ran two user tests and iterated quickly, and had designs ready. Thanks to usertesting.com. This is what we are shipping:
It seems like a very small improvement but we went through a lot of winding roads to get there, and sometimes design isn’t super elegant and visually exciting.
Our next steps are to see how this performs, and if it in fact reduces duplicates while we look into improving search. If this works well, it might be something we implement in issues also!
The Blinks/tldr; for you to take away:
Ask other people if they’ve done something similar and get their learnings before taking off. And document your learnings.
Quick user tests are incredibly impactful with little effort.
You have to make trade-offs for technology sometimes.
Design isn’t always pretty.
You can find this in blog-like not-tweet-like form, in my Discussion here! Smash that starrr button for more.
Beta Was this translation helpful? Give feedback.
All reactions