I screenshot. A Jakarta user said it — and it became the product brief.
An Android app for Indonesian mobile users to capture, organize, and OCR-search their screenshots — built from a market we'd never touched, validated through 1,000+ survey responses, 8 in-field interviews, and a landing page test that unlocked executive investment.
In early 2018, Indonesia was a brand-new market for Mozilla Taipei. No prior product, no on-the-ground research, no relationships. Before naming a product, the team had to answer a simpler question: what do these users actually need?
The starting point wasn't a feature — it was a behavior almost nobody on the team had taken seriously. Indonesian mobile users seemed to lean on screenshots in ways we didn't fully understand. The research started there.
We used Jakpat, Indonesia's mobile survey platform, to run five rounds of research — ~200 respondents per round, around 1,000 in total. To validate user needs without engineering, I embedded storyboards directly in the surveys: hypothetical scenarios users could react to, no prototypes required.
A cross-functional workshop with PM, marketing, engineering, BD and UX used a risk-priority matrix to pick the top hypotheses to pursue. But the bigger finding was structural: Indonesians used screenshots for almost everything — faster than apps (no account, offline), more reliable than bookmarks on slow networks, and accumulating in galleries with no way to find them again.
Screenshots over apps. The behavior wasn't a workaround — it was a primary interface.
Persona 1 — synthesized from the 1,000+ survey responses
Persona 2 — synthesized from the 1,000+ survey responses
Persona 3 — synthesized from the 1,000+ survey responses
Two layers of validation followed: in-field testing for user need, a landing page test for market size.
Ricky (User researcher) and I flew to Indonesia to interview eight participants. The first four sessions validated unmet needs and shaped iteration — we updated the Framer prototype within a day. The next four ran usability testing on the iterated design. One offhand line from a Jakarta participant stuck with the whole team, and ended up framing the product brief itself.
User need was validated, but market size wasn't — and exec investment depended on it. I built a landing page describing the future product, ran paid social on Indonesian channels, and used the Playing to Win framework to argue why Mozilla should invest specifically in this market. The landing page converted at around 20%, well above the threshold the exec team set. The budget moved.
Landing page used to market-size demand before product investment — ~20% conversion
Two pain points anchored the design. Galleries overflowed with screenshots that were impossible to find again. And screenshots were static images — they couldn't trigger any digital action. The product had three core moves: capture fast, organize without thinking, and search by content rather than filename.
When I was uncertain about an interaction, we built a low- or hi-fi prototype, recruited Indonesian users for a fast usability sprint, and iterated. The OCR feature in particular went through several rounds to balance recognition accuracy against UI clarity — below 70% accuracy the search felt unreliable; above 85% it felt like magic.
Quick capture & sort — floating button triggers organize/share without leaving the gallery
Smart collections — visual states make it obvious which screenshots belong where
OCR text search — type any word from any screenshot, surface the image instantly
After launch, ScreenshotGo reached an average rating of 4.6 stars on Google Play across the Indonesian market. Download count is unavailable — Mozilla later sunset its Lite product line and the app was delisted, taking the install data with it. What survived was the user feedback itself.
The trick that mattered most wasn't the OCR or the floating button — it was embedding storyboards directly into Jakpat surveys to test concept-level hypotheses without engineering. Turning 200-respondent rounds into directional validation is now a default tool in my toolkit when budget is tight or engineers are unavailable.
User need was provable in weeks. Internal investment took months. The landing page test wasn't decoration — it was the lever that moved the budget. I've built market-sizing experiments into every early-stage project since, for the same reason: shipping a product is the easy half. Convincing the org to invest in it is the other half, and it's where most ideas quietly die.