When “data driven” isn’t enough

Why recruitment models diverge between men’s and women’s football

Liam Henshaw’s three-bucket framework of data driven, data informed, and sporadic use describes how clubs like to think they work. In the men’s game, it is a useful shorthand. In the women’s game, the buckets often sit on the wrong foundation. The core issue is not only how clubs use data, but whether the data and models actually describe the women’s game.

There is good evidence the games differ in meaningful, model-relevant ways. Peer-reviewed work shows systematic technical and tactical differences between men’s and women’s football: distinct passing and possession dynamics, different shot locations and conversion rates, and role and context effects that change how features should be specified. If you port a men’s feature set and weightings into women’s recruitment, you risk filtering out high-value players before a human ever sees them.

On the ground, analytics capacity in women’s clubs remains uneven. A new American Soccer Analysis survey of NWSL analytics staff, based on responses from seven of 14 clubs, highlights thin staffing, limited access to tracking data, and widespread frustration with league-provided data quality. Even the clubs perceived by their peers as leaders built analytics in from scratch with a small team, not the multi-layered departments you see at top men’s sides. The picture that emerges is very few fully data driven, most data informed, and many sporadic, for structural reasons as much as preference.

The vendor ecosystem is expanding but still catching up to the men’s side, which matters for recruitment inputs. Opta’s Opta Vision added complete player tracking and increased coverage by 35 percent to over 60 men’s and women’s competitions only last season, a sign of rapid progress but also of how recently broad tracking has become accessible. StatsBomb’s decision to provide free platform access to women’s teams across major leagues underscores the same point: the pipeline is maturing, not mature. Build “top five percent only” rules on immature or mis-specified inputs and you will miss players.

So what does Henshaw’s spectrum look like when you account for the women’s game?

Data driven exists, but it is rare end-to-end. Where it appears, it is typically in clubs that hired analytics leaders early and integrated them across football operations, not just as a reporting function. Even there, model design and feature engineering must be gender-specific to be trustworthy. A women-trained xG framework, for example, captures different shot selection and conversion realities than a men-trained one, and those differences affect how you rank strikers or value chance creation.

Data informed is the real median. Scouts and coaches start the conversation, analysts support or challenge. The catch is that if both the human priors and the data pipeline are inherited from the men’s game, the process becomes self-confirming. The remedy is not to use more data, but to use the right data and context: role-aware, league-adjusted, and trained on women’s match events so the model questions flawed priors instead of echoing them.

Sporadic use remains common by necessity. Many clubs still pull analysts into recruitment ad hoc while those same analysts cover opposition prep and post-match. The ASA survey makes that plain: respondents cite more staff and higher-quality raw data as the two biggest unlocks, and only one club reported using tracking data, which was GPS or sports-science rather than broadcast or optical tracking. This is not an anti-data culture problem, it is a resourcing and infrastructure problem.

There is also a fourth pattern that Henshaw did not name but that is visible in women’s football: network-led, data-optional. Transfers are booming from a very low base, with women’s international moves hitting a record 2,284 in 2024 and fees more than doubling year-over-year to 15.6 million USD, yet the market is still thin enough in many places that trusted agents and pre-existing relationships drive shortlists, with data validating after the fact. FIFA’s agents report shows agent involvement and fees rising in the women’s game, which supports, though does not quantify, that relational gravity. Frame this as observed practice, not a universal law, and it is accurate.

The practical takeaway is simple and uncomfortable. Before arguing about whether you are data driven or data informed, prove your inputs are built for the game you are actually playing. The academic literature shows the women’s game has different signal, the industry survey shows staffing and data access are uneven, and the vendor releases show coverage is expanding but still new. Those three facts make “model first, humans second” risky unless the model is truly women-specific and context-aware.

QQSI’s position follows from that evidence base. Same sport, different game. Build recruitment on women-trained models, role and league adjustments, and contextual multipliers that reflect how women’s football is actually played and developed. In a world where very few clubs can be rigorously data driven end-to-end, the winning edge is not faith in numbers or faith in networks. It is using the right numbers to challenge the right priors at the right moments.