What do you do when you know someone on your friend request list looks…suss?
Spammy friend requests are a thing, and our team’s Product Specialist collected user feedback that showed:
1) People didn’t know how to remove suspicious or unwanted friend requests.
2) These requests made users feel uncomfortable.
Our team knew we had to serve some sort of friction that could help people make better decisions before accepting suspicious friend requests that we were able to identify on the backend — but as a Content Designer and user advocate, I also wanted to make sure that we didn’t misrepresent these potential “spammers” if our machine learning got it wrong and mislabeled innocent people.
We’re able to identify potential spammy friend requests on the backend through a Fake Index Score (FAI) — at a certain threshold, the person and hence their friend request would be labeled as “potential spam.”
So when a user tapped the “Confirm” CTA on a person who was a potential spammer, we’d pop up a bottom sheet that served as “friction” to make the user stop and think if they really wanted to accept this request.
I took content in so many different directions at first: should our tone be informative since this is an integrity issue, or should we be more helpful and reassuring since it’s unexpected UX behavior to serve a bottom sheet when all you wanted to do was accept a friend request?
It was weird to ask the user, “do you want to accept this friend request?” when their actions clearly stated otherwise; they tapped “Confirm” to begin with, so of course they wanted to friend this person, right?!
Facebook’s standard messaging has always been to friend “people that you know.” But what if that messaging was too outdated? I’ve made friends over Tumblr and Discord who I’ve never met IRL, but who I still consider to be “friends.”
This project made me realize it was time to catch Facebook’s guidance up to speed on online friending.
I decided to lean more into Facebook Friending’s value props and use them as a way to let people know that we were looking out for them.
For our final design, I chose to lean into safety: because at Facebook, we want to connect you with others in a safe way. It felt more like a gentle nudge vs. a generic and cold auto-confirm message.
THE RESULTS
And our content + design paid off. When we tested this iteration, we were able to reduce abusive friend request accepts globally on Android by -2.09%.
Of those users who filed an friend request user report after seeing the bottom sheet, there was a 0.59%+ improvement for reporting fake users and a -22.2% reduction in selecting “Uncomfortable” as the reason for the report — this could be due to a reassuring content approach.