The Video Pivot: Killing My Own Feature to Find the Real Use Case

Context

Indus AppStore is India's first homegrown Android app store. I owned Discovery — the consumer storefront, with conversion, engagement, and retention as my outcomes. Short-form video was exploding in India, and we bet on a full-screen video feed inside the app store: TikTok-style browsing for app discovery.

Even getting it built required de-risking: engineering was rightly skeptical of video infrastructure derailing the roadmap, so V1 shipped in five weeks with no ML on the critical path — videos ranked by category popularity and manual tags — while we instrumented every interaction (watch time, completion rate, scroll depth, tap-through) from day one. To beat the cold start, my team hand-curated ~500 videos covering the top 100 trending apps before launch.

The call

On the surface it worked: strong watch times, high engagement, growing MAUs. But the instrumentation we'd insisted on told the real story — users were watching videos and not downloading apps. We had built an entertainment experience inside a utility product. Users came to an app store with intent; a lean-back feed fought that intent instead of serving it. I could have reframed the KPIs around engagement and asked for more headcount. Instead I went back to first principles — what is an app store fundamentally for? — and killed my own feature.

The insight worth keeping wasn't the feed, it was the format: video sells apps better than screenshots. So I repositioned the same underlying tech as contextual app trailers — short-form video embedded where the install decision actually happens, on the app detail page, instead of a destination feed.

Impact

• Install conversion: 12.9% → 15.5% — a 20% relative lift
• Installs per video watched more than doubled, and backend costs dropped — we were no longer serving an entertainment feed that didn't convert
• The surrounding review + short-video ecosystem drove +20% 30-day retention and 150k+ organic ratings

The lesson I carry: the sunk cost was the feed, not the capability. Repositioning beat both persisting and abandoning.

Skills: product sense · metric-led decisions · killing features · consumer video/UGC