Projects go wrong.
Professional practice cannot prevent every problem.
It can make many of them easier to diagnose.
Endless feedback
If feedback never ends, stop treating the problem as a design problem.
Look at the decision structure.
Is there a clear owner?
Are evaluation criteria agreed?
Are new stakeholders arriving late?
Is everyone reacting to screens without understanding the problem?
Fix the way decisions are made before producing another revision.
Design by committee
The problem is not that many people contribute.
Complex products require many forms of expertise.
The problem is unresolved authority.
Gather information widely.
Invite disagreement.
Then establish who makes which decision.
Do not average conflicting opinions into a compromised design simply because compromise feels diplomatic.
Scope creep
Scope rarely explodes in one moment.
It accumulates.
One page.
One component.
One extra user journey.
One additional audience.
One integration described as a small change.
Refer to the agreement.
Explain the impact.
Change the fee, schedule or priorities accordingly.
Scope can change.
The agreement must change with it.
Generated mistakes
AI introduces new failure modes.
Invented content.
Incorrect research summaries.
Non-existent components.
Hallucinated APIs.
Inaccessible code.
Duplicated design patterns.
Plausible-looking false data.
Do not assume generated work is finished work.
It becomes professional work only when someone competent evaluates it and accepts responsibility.
Sometimes the problem is you
Consider that earlier than feels comfortable.
Perhaps the explanation was weak.
Perhaps the research was poor.
Perhaps engineering found a genuine flaw.
Perhaps the client keeps questioning the work because the work deserves questioning.
Professional confidence without self-examination becomes arrogance.