A performance department is rarely short of data. It has GPS and accelerometer load from Catapult or STATSports, event data from Opta or StatsBomb, tracking where the league provides it, video tagged in Hudl Sportscode or similar, and daily wellness questionnaires. Each sits in its own tool, keyed differently, and joining them for a single question like “how did our pressing intensity hold up in the last fifteen minutes of away games after a short turnaround?” is a day’s work.
We build the join and the question layer. Data is aligned by match, player and timestamp in a store the club controls. An analyst assistant turns a staff question into a query, returns the figures with the method shown, and links each finding to the clips it came from, so the coach watches the evidence instead of trusting a number.
Computer vision helps where data is missing: tagging set pieces or pressing triggers in your own footage, or estimating positions from a fixed tactical camera at levels without official tracking. Those outputs are marked as estimates and checked by analysts.