Gamifying upsell at Yango Play

Yango Play is the streaming product I lead design for — video, music, and games for users across the GCC and beyond. The annual subscription tariff is one of the highest-leverage conversion goals on the product, and the team had been hitting a ceiling on it for months.

This case is design I directed but didn't author. Denis Shumov owned the project end-to-end as a Sandbox initiative — concept, exploration, motion, prototyping, and ship. He drove the entire visual production through Krea.ai, which is the part of this story that mattered most for delivery. My role was the framing, the call on what to ship, and unblocking. The craft below is Denis's.

Yango Play already runs five communication formats in production: Tour Updates, Fullscreens, Promo Bottom Sheets, Micro Promoblocks, and Notifications. They cover the full surface of how we talk to users — first-launch tour, fullscreen takeover, in-feed nudge, and out-of-app push.

Our annual tariff sat behind these surfaces, and the team had iterated on them hard. CTR, engagement, and conversion were all tracked. The number that mattered — annual tariff subscriptions per month — had plateaued.

Our existing interactive tool — Wheel of Fortune — was the proven lift driver. But it was on another track and would not be ready in time for Ramadan. Ramadan was four weeks out, and Ramadan is the conversion moment for our GCC audience. The annual tariff has to land then or it doesn't land at all that year.

Two converging facts: the existing static formats had stopped scaling, and the interactive tool we'd built before wouldn't ship in time. We were scrolling past our own promos at the same sale points users were — that's the truthful read of the ceiling. The ask was a worthy interactive replacement, designed and shipped in four weeks, on three platforms — Android, iOS, and TV.

Interactivity and motion would beat static because they'd buy the user's first three seconds. Three seconds is a lot in promo time — long enough to read the offer, long enough to convert. The lift wouldn't come from a lower price. It would come from keeping the user present.

The offer was sized for that frame. An annual tariff is roughly 265 paid days; the promo adds 100 free days on top. Same money in, more days delivered — and it's sold as a gift, not as a discount or a cheaper monthly. "You won 100 days" reads completely differently from "save 30%."

Denis pulled five principles out of our previous wheel-of-fortune learnings and the static-promo data. Everything below was in service of one of them.

Generation is endless. Anything can be turned into a mechanic. So we set four criteria for whether a mechanic was good enough to ship — anything that failed two of them was cut.

The working method matters more here than any single artifact. Denis ran the entire visual exploration through Krea.ai — image, motion references, and full-fidelity moodboards for every direction we tested. That's how we got from "we need a Ramadan mechanic in four weeks" to a finished motion-driven promo without a dedicated video team.

The catalog of mechanics tested:

For each direction, Denis generated a complete visual treatment in Krea — not a sketch, not a wireframe, a high-fidelity moodboard with motion references — in under an hour. Six directions in a day. That speed is what made the exploration possible under the timeline at all. With a traditional pipeline (concept → static moodboard → motion brief → AE comp → review) we'd have spent the four weeks just exploring.

A scratch over a generic surface is a discount. A scratch over a film genre is a media product. So we ran a round where the surface itself was a Yango Play title — Drama, Comedy, Fantasy. Same scratch action, same offer, but the prize is days of content the user can already see on the catalog.

Two production options on the table.

We first narrowed to two mechanics, both built on the same central image — a magic ball. Scratch the ball to reveal the prize, or shake/spin to trigger the reveal. Familiar fortune-teller object, instant read across markets.

Then our regional partner Maysam flagged it. Magic-call imagery isn't appropriate in the GCC market — culturally and religiously off. We had to find a central object that wasn't a magic ball, and we had a few days to do it.

Krea earned its keep here. Denis went back into exploration mode, generated dozens of alternative central objects across an afternoon, and landed on the Ramadan lantern (fanous) — a symbol that's iconic to the region, neutral on religion, and visually richer than the ball ever was. Two interactions then ran in parallel: spin the lantern and scratch the gold off it.

From all of that, two finalists earned the ship slot — both built on the same lantern, both ending with the prize emerging from inside it. The difference is the mechanical action and the surface state at frame one.

v1 — Spin the lantern. The lantern hangs intact in the centre, lit and animated. The user spins it; momentum settles into the reveal of "100 days" glowing through the glass. The reward feeling rides on the rotation arc.

v3 — Scratch the gold. The same lantern, but coated in gold leaf with no Yango logo on it. The user drags a finger across the surface; the gold erodes in real time and the lantern emerges underneath. The brand and the prize both surface from under the user's hand.

Scratch won the ship slot. It maps to a real-world action — everyone has scratched a lottery ticket. Shake/spin doesn't have the same payoff intuition; you don't know if you've spun hard enough, and the absence of clear cause-and-effect kills the reward feeling. Logo removed from the gold layer because the brand had to emerge from under the user's finger, not greet them at the start.

Two-line copy. One CTA. No ladder. The mechanic takes about three seconds end to end.

Shipped in four weeks. Three platforms. The results split cleanly by platform — and the split was the most interesting part.

Android — strong lift. CTR and activations both up in Egypt, the primary Ramadan market. The mechanic worked exactly as the principles predicted.

iOS — analytics still settling. A handful of tracking bugs masked the early numbers. Likely positive once corrected, but we don't quote a number until we trust it.

TV — the surprise. Test group clicked less than control.

Three things on the next sprint:

The Ramadan promo was the visible output. The reusable webview pipeline — the one Denis built around Krea.ai as the visual layer — is the meta-output, and the more valuable one.

The interesting story for me was watching how much of the production line could be operated through Krea.ai by a single designer with motion fluency. Concept, motion, prototype, and ship — one head, four weeks, three platforms.

That's the shape of the design work going forward. The team's job stops being "produce assets" and starts being "set the principles, choose the criteria, and decide what ships." This case is the first time I saw that work end-to-end on the team I run.