Trade-off defence, NFRs, modernisation sequencing, and the stakeholder and cost conversations architects are actually assessed on.
Say yes if it is true, give a stage and a rough timeline without naming companies, and use it to set a decision date rather than to imply a bidding war. Inventing a process you do not have is the one answer that can end badly.
Usually yes for the label and no for the concept. A fossil name is still the word that lets you and the business point at the same thing, so keep it as the shared term, but model what the queue is for today rather than the channel it was named after.
Treat both as evidence, not verdicts: a document records an intention on a date, an expert reports today's practice inside their own scope. Most contradictions resolve as staleness, scope or the exception path, and practice departing from a control is a finding to escalate rather than a requirement to build.
Give the reasons in one pass so the pattern looks like a sequence of decisions rather than a habit, then name what you have changed about how you choose a role. The answer that works is a filter you can state, not a promise to stay.
Keep one small global directory holding only the keys needed to route a request, never the payload. Uniqueness becomes a claim on a pseudonymous token, search becomes a fan-out that returns pointers each region resolves locally, and every field promoted to the global tier is a transfer decision.
Answer no, and give the non-money reason you are leaving as the evidence. If a real counteroffer arrives later, judge it against what your employer did before you threatened to leave, not against the number on the paper.