Time to Market
How long it takes to get a capability in front of customers — often the constraint that dominates every other architectural quality.
For a product still establishing whether anyone wants it, being six months late is usually a larger risk than being architecturally imperfect, and an architecture that optimises for a scale the product may never reach has spent the one resource that cannot be recovered.
The mature version of this is not "cut corners". It is choosing deliberately where to be quick and where not to: prefer managed services over building; keep the structure simple and the boundaries clean, since boundaries are the expensive thing to change later; and take on debt knowingly in areas that are cheap to revisit while refusing it in areas that are not — data models, public contracts and security.
Stated as a trade-off in a design review it becomes a decision rather than an accusation: "this design ships in six weeks and will need rework above 50,000 users; the alternative ships in five months and does not."