Finding 01The architecture was not collapsing. It was distorting judgment.
The team could still ship. But too many decisions were being made around “what seems safest to touch” rather than what mattered most to the product. That meant the architecture problem was already commercial, not just aesthetic.
Finding 02One concentrated product area was carrying disproportionate risk.
Most of the system was messy but survivable. One zone — where product logic, state transitions, and ownership boundaries intersected — was creating outsized roadmap fear and second-order instability. That was the area worth stabilizing first.
Finding 03A full rebuild would have been emotionally satisfying and strategically wrong.
The product still had too much value, too much learnable structure, and too much momentum to justify a dramatic restart. The better answer was narrower: preserve what still served the product, stop defending the concentrated risk zone, and tighten the decision surface around the next milestone.