Case study
PlayerHunter
A two-sided marketplace that opened football transfers to players outside the traditional scouting network.
The challenge
The global football transfer market moves over $400 billion a year and runs on 1950s infrastructure. Players depend on luck, local connections and expensive agents. Clubs depend on scouting networks that cover a fraction of the world. Agents run their books out of spreadsheets and WhatsApp.
The cost of that is concentrated on both ends of the market. Talented players outside a scout's travel radius are effectively invisible. Clubs pay for scouting coverage and still miss obvious matches. One scout put the problem plainly: he was certain the right left-back for his club existed in Brazil, and equally certain he would never find him, because the system only showed him about one percent of available talent.
What I built
I led the technical development from concept to a platform serving 20,000+ players and 1,800+ clubs across 50 countries.
Matching, not searching. The core of the product is a matching engine rather than a search box. It scores player attributes (position, age, experience, statistics) against club requirements including league level, budget, playing style, geography and work-permit eligibility. A club describes what it needs in under a minute and gets a ranked shortlist instead of a directory.
Two products, one market. Clubs and agents work at a desk, players work from a phone. So clubs and agents got a React web application with a Kanban-style candidate pipeline, direct secure messaging, posting analytics and bulk operations for agencies managing large books. Players got native iOS and Android apps built around a digital football CV: career statistics, match history, references, medical and fitness data, one-click applications, and cloud video upload for skill footage. Push notifications closed the loop, since opportunities in this market expire in days.
Infrastructure for a global two-sided market. A Spring Boot API over PostgreSQL for structured player and club data, Elasticsearch for the search and filtering layer, a CDN for video and asset delivery worldwide, and a REST API for third-party integrations. A collaborative-filtering recommendation engine came later, once there was enough behavioural data to make it useful.
The sequencing mattered more than any single component. We researched existing scouting workflows and studied how other two-sided marketplaces (LinkedIn, Uber, Tinder) solved cold-start and trust, before designing the matching requirements. Then core platform, then mobile, then club and agent tooling, then scale optimisation. Building the sophisticated matching layer first would have been a year spent optimising for a marketplace that did not yet have two sides.
Results
- 20,000+ registered players, including 4,000+ professionals
- 1,800+ club requests and 600+ registered agencies
- 50+ countries represented on the platform
- Players from lower leagues securing trials, and in several cases first professional contracts, at clubs that would never have found them
- Clubs signing players from outside their established scouting geography
- Agents running international books without the travel cost
The platform was covered in sports media as "LinkedIn for football", endorsed by professional players including Ajax captain Dušan Tadić, selected for the Virgin StartUp Crowdboost accelerator, and built partnerships with clubs across Europe, the USA, China and the Middle East.
What I took from it
Marketplaces fail on the side you neglect. Equal engineering investment in the player experience and the club experience was not a fairness principle, it was the thing that made liquidity possible. Mobile-first was not a design preference either. Players are, by definition, not at a desk. And the transparency features that felt like overhead early on, verification and ratings, turned out to be what made strangers willing to transact.
A $400 billion market ran on opacity for seventy years. It was not protected by anything except the absence of an alternative.