Hivo — Solving Cold Start in a Two-Sided Marketplace

Context

Hivo is a two-sided coworking marketplace — remote professionals ("guests") booking underused commercial space from "hosts." The core product was built for North America. As Product Owner I owned the roadmap to take it into the Gulf: first Saudi Arabia, then the UAE. Counting the North-American home market, that meant one codebase serving three regulatory regimes at once, without forking into three products.

The problem

The booking engine worked; the market didn't. International card rails hit high domestic decline rates against local payment networks. The app-download-first flow lost users who search the web before they install anything. And the product was legally non-compliant with Saudi ZATCA e-invoicing while blind to the cultural trust signals local users expect. Translating the interface into Arabic would have fixed none of it.

What I did

Split discovery into three workstreams — fintech, cultural UX, and regulatory compliance — and used a weighted RICE matrix to set MVP scope, so prioritization was settled by score rather than by seniority.
Fintech: prioritized Mada (the national debit network) plus STC Pay and Apple Pay so domestic, one-tap payments actually cleared.
Cultural UX: re-architected the layout for bidirectional RTL and added region-specific amenity filters (prayer-room proximity, spatial zoning) and Ramadan availability logic — localization as architecture, not translation.
Regulatory: automated ZATCA Phase-2 e-invoicing inside the booking pipeline and built a Commercial-Registration host-verification flow so commercial hosts could onboard without legal exposure.
Led the Dubai launch: AED multi-currency, local gateways, and UAE Free Zone compliance on the same core.

The decision I’d defend

I moved the Gulf launch to web-first discovery instead of the app-download-first model the global core team's architecture assumed. In a market where users search before they install, forcing an install before showing any value was bleeding early-funnel demand.

Rejected: app-download-first — the global default the core team preferred.

It cost real engineering effort and diverged from the core team's architecture, which I had to defend. But acquisition cost is set at the top of the funnel, and no downstream polish recovers a user who bounced before they saw a listing.

Outcome

The Dubai launch grew host inventory 45% in the first month and lifted host activation 30% through a redesigned onboarding flow, and the focused MVP directly supported a $250K seed round. The web-first change was modeled to cut early-funnel drop-off by roughly 35–40% — a projection from market testing, not yet a measured result, and I'd treat it as one.

What I'd do differently

I'd have shipped the web-first bet as a live A/B from day one instead of leaning on modeled funnel numbers. A measured lift would have made the case unarguable to the core team — and I wouldn't have to caveat the projection now.