Discovery to delivery,
end-to-end ownership
I was the sole designer on this one, which meant I owned every phase: diagnosing the real problem, designing the test, running iterations, and proving the impact. But sole owner never meant solo. I leaned hard on product and engineering at every step, because the best signal in the world is useless if the backend can't surface it. Here's how it actually unfolded.
A value signal system,
not just a badge
The fix wasn't a prettier label. It was a disciplined rethink of how deals get surfaced: one credible signal, backed by filter-level discoverability. My guiding principle the whole way was restraint. Fewer, stronger signals beat more noise every single time.
Three badge states tested (filled, outlined, ghost) alongside a contextual tooltip surfacing real-time demand and relative value signals, giving users both urgency and confidence.
Key design decisions
Contextual value, not universal
A "Hot Deal" is relative to similar inventory in that event, not some global standard. That kept the label honest even as pricing shifted hour to hour.
Selective application
I set a hard ceiling: no more than ~20% of listings could wear the badge. Scarcity is what keeps a signal worth trusting. Go past that and you're back to noise.
Filter integration
I turned "Hot Deals" into an active discovery tool, not a passive sticker. High-intent users could filter straight to value, basically a fast lane to checkout.
Eye-movement aligned placement
Placed it in the natural scan path, just left of price, exactly where eyes were already landing while comparing options.
Why I capped the badge at ~20%
This was the call I had to defend the most, so here's the argument as a toy you can play with. Drag the slider to badge more listings and watch the signal collapse into wallpaper. It's the whole "when everything is Best Deal, nothing is" insight, live.
Extending the signal to the seat map
The badge didn't stop at the ticket list. I mirrored Hot Deal sections straight onto the interactive seat map, so hovering or selecting a flagged listing lights up the matching section with the same 🔥 marker. That closed the loop between the list and the map and reinforced the signal at every touchpoint in the selection flow.
I built the spec as something
you could actually click
Static mockups tell the story, but they leave too much open to interpretation the moment engineering starts wiring things up. So I used AI to turn the Hot Deals spec into a working, interactive prototype in hours instead of waiting on a build cycle. It pulled double duty: it gave developers a behavioral source of truth, the real tab switching, the badge logic, the map overlays, so there was no guessing from a flat file, and it gave stakeholders something tangible to react to in review, which is a far faster way to reach internal alignment than another round of slides.
Same screen, three states. Developers could click each one to confirm exactly what changes between them, the listings, the map overlay, the badges, with nothing left to interpret.
What didn't go my way
(and how I handled it)
No project this fun is a straight line. The interesting part of being senior isn't avoiding the curveballs, it's how you catch them. Here are the three that tried to derail this one, and what I did instead of caving.
Scope creep: "can we badge everything?"
Once stakeholders saw the badge convert, the asks piled up fast. "Add Top Value." "Add Lowest Price." "Trending too." Everyone wanted their slice on the page, which is exactly how we got into the "Best Deal" mess in the first place.
I didn't say no with an opinion, I said no with a test. I pointed back to the early finding that high label frequency tanked trust, and reframed the ask: every new badge dilutes the one that's working. We agreed to ship one signal, measure it, and earn the right to add more later. Saying no got easier once the data was doing the talking.
Engineering: real-time "value" was expensive
My ideal version compared each listing against live, similar inventory in real time. Engineering flagged that doing it perfectly on every page load was a heavy lift, and I was not about to torch the timeline chasing a v2 dream.
I sat down with eng and co-scoped a leaner v1: a rule-based threshold against cached comparables that captured most of the value at a fraction of the cost. Good enough to validate the hypothesis, cheap enough to ship now. We parked the fancy version as a fast follow once the concept proved out.
My first badge styling tested worse
I went in loving a loud, filled-red badge. It was bold, it was confident, and in testing it read as an ad. Users glazed right over it. Humbling, but that's what testing is for.
I let go of the version I was attached to and followed the signal. I dialed the badge back, paired it with a contextual tooltip that explained why a deal was a deal, and tightened the application rules. The quieter, more honest treatment is the one that won. Design ego costs conversion; I'd rather be right than precious.
Perception changed.
Revenue followed.
Moving from noisy, overused labels to a disciplined value-signal system lifted conversion in a way we could measure. The takeaway I loved most: the bottleneck was never inventory, pricing, or intent. It was perception. Fix what people can feel, and the numbers move on their own.
Conversion funnel improvement
Beyond the numbers
MegaSeats as a test bed
Proved MegaSeats as a faster, lower-risk testing ground before rolling changes up to TicketNetwork. Quicker learnings, less exposure, happier stakeholders.
Scalable A/B framework
Walked away with a documented, repeatable experiment framework applicable to future merchandising and pricing initiatives.
Design as business strategy
Revenue grew without touching pricing or inventory, proof that design decisions move business outcomes at scale.
XFN trust
Led end-to-end with full ownership. Built credibility with product and engineering as a design partner who thinks in systems and ships results.