Most bad software estimates fail for the same reason: the number was produced before anyone agreed on what was being built. A fixed price against a vague brief is not a commitment, it is a guess with a due date attached.
We scope before we price
Before a number goes in writing, we produce a scope document: the screens, the data model, the integrations, and explicitly, what is out of scope. If a client cannot point to a document and say "this is what we agreed to," the estimate is not real yet.
Integrations are where estimates die
The parts of a project that blow budgets are almost never the screens. They are the third-party integrations: a payment gateway with undocumented edge cases, a legacy API with no sandbox, a compliance requirement nobody mentioned in the kickoff call. We ask about every external system in week one, not week six.
What happens when something slips
Scope creep is normal, and pretending otherwise helps no one. When a client asks for something outside the written scope, we quote it as a change separately rather than quietly absorbing it or silently blowing the deadline. You hear about a slip in the same week it happens, not at the deadline.