Couldn’t find event details
Basic event info was buried in the navigation. Users had to hunt for what should have been obvious.
Problem
Musicians backed out at payment
Slove books musicians into venues, often for over $1,000 a gig. The flow said pay now, sort out the details later.
My role
Mobile designer, freelance
Ran discovery, card sorting, IA, prototyping and testing for the mobile app; teammates covered desktop and research. Delivered in 3 months, then stayed on a year.
Outcome
60% → 100%
Task success across three rounds of moderated tests. In the live app, conversion rose 15 points and abandonment fell 18 points.
Committing $1,500 on that little information felt like a gamble. So people didn’t.
What the research surfaced
Basic event info was buried in the navigation. Users had to hunt for what should have been obvious.
Without upfront equipment needs, timing, or responsibilities, paying felt like a leap of faith.
The insight
I expected the fix to be fewer steps. The research pushed me the other way: users wanted more to go on before committing.
Every interview landed on the same thing: people would happily spend longer if it meant understanding what they were agreeing to. So I designed for clarity, even where that meant adding steps, and tested each fix with users before the next.
Round 1: swipe to scroll navigation
Swiping felt too much like a dating app. Scroll and dark mode gave a more professional feel.
Round 2: “likelihood” became “match %”
Made cards fully clickable, cut the info density, and rewrote confusing booking-probability copy.
Round 3: credibility signals on profiles
Venue profiles felt unreliable, so I added proof of played gigs, upcoming events, and social links.
A deliberate trade-off
Booking time went up. Everything else went up too.
The new flow takes almost twice as long. Task success, confidence and conversion all went up anyway, because users finally understood what they were agreeing to before hitting “pay.”
Decisions
Four calls, what each replaced, and what it cost.
Scroll and dark mode
cost · A more playful feel
instead of Swipe (read as a dating app)
Four facts per card
cost · The score explanation sits one tap deeper
instead of The full match explanation on the card
“Match”
cost · A slightly less precise word
instead of “Likelihood”, the model’s word
Showing “0 gigs played”
cost · New venues look untested
instead of Hiding empty proof
In testing, the tasks people failed in round 1 were the ones the research flagged first: finding basic event details. By round 3, 12 of 12 got through.
Moderated tests, 12 users per round. Confidence from a one-question survey after the flow. Conversion and abandonment from the client’s analytics, old flow vs new.
Reflection
The obvious fix would have shipped in a week and made things worse. The research is what stopped me.
This is the project that taught me to trust the research over the obvious answer. If you’re designing for trust in a high-stakes flow, payments, bookings, anywhere people hesitate before committing, I’d love to compare notes.
More work