Bite-sized UX
Vidsy · 2023–24 · Bite-sized · 2 min read
Client feedback on Vidsy videos was often vague and tangled: one comment holding three requests, or a note like “make it funnier”. Each unclear note meant another round with the motion designer. We used AI to break feedback into clear, specific action items, without getting in the way of clients whose feedback was already good.
01 · The problem
Deciphering client feedback took time. Comments were vague, entangled and short on specifics, which led to extra touchpoints and revisions.
How might we turn client feedback into clear, concise, actionable items for motion designers?


02 · AI scoring
We needed actionable feedback for designers, but didn’t want to add friction when a comment was already good enough. And some feedback was so ambiguous (“make it funnier”) that any suggestion would be a shot in the dark.
So the AI scores each comment from 1 to 5, and we only step in at 2 or 3: feedback that needs work, but isn’t so vague we can’t suggest a direction.
03 · Keeping clients in control
We wanted clients to use the AI so we could watch how it performed and improve it. But adding specificity to ambiguity means guessing, so the suggestions wouldn’t always be right.
I let clients keep their original comment, edit any suggestion directly, or dismiss the ones they didn’t want.


04 · Continuity
Breaking a comment into separate action items made it hard to see how each piece fit the whole, especially when part of it needed no change.
After many variations I landed on highlighting the part being worked on, with a carousel to step through each point in context.
05 · What happened
Average revisions per video, nearly halved.
Sooner that clients returned their feedback.
Screens are from the original Vidsy prototype.