What breaks when every squad has its own QA
One tester per squad feels like coverage. It is usually five single points of failure.
A dedicated QA per squad is the most intuitive model there is. Everyone knows who to ask, context never leaves the room, and the tester learns the product deeply.
It also quietly builds five single points of failure.
The symptoms come before the diagnosis
Someone takes two weeks off and a vertical stops homologating. A high-criticality demand waits because the only person who knows that flow is finishing something trivial in another squad. Nobody can answer how much quality capacity the company actually has, because capacity is not a number anywhere — it is five private calendars.
Centralising capacity is not centralising people
Moving to managed capacity does not mean pulling testers out of the squads. It means the decision about what gets tested next stops being made inside each squad in isolation, and starts being made against criticality, volume and priority across all of them.
That decision needs a board everyone can see and criteria everyone agreed to before the first conflict, not during it.
The part nobody likes
Shared capacity makes scarcity visible. When a vertical waits, it is now obvious that it is waiting, and obvious why. Teams that used to feel well served start feeling the queue.
That discomfort is the model working. The queue existed before; it was just invisible, and invisible queues get resolved by whoever shouts loudest.
Cross-coverage is the exit
The only real fix for the single point of failure is more than one person able to cover each area. It costs ramp-up time that produces nothing in the first month, and it is the difference between a team and a set of individuals who sit together.
