career bridge
Say no with evidence
Three costed options, conditions specific enough for someone else to verify, and your name on the recommendation.
You have an assessment with real gaps in it. Now the part the role is named after. Somebody wants to ship, the date is already public, and you have to decide what you support, what you refuse outright, and what you allow only with conditions attached to it.
What you do
Turn findings into a decision that survives pressure. Write three options rather than one, because a flat no is easy to overrule and a conditional yes is not: ship as is, ship with named conditions, or do not ship. Say what each costs and what each risks in the organisation's own terms, revenue, delay, exposure, reputation. Then recommend one and put your name against it. Make the conditions specific enough to verify: a named human approving decisions above a stated threshold, a log retained for a stated period, a rate the system must stay under with a stated way of measuring it. Vague conditions are agreements to argue later. Set the review trigger too, because a control nobody re-checks decays quietly. Then write the escalation path in one sentence: who decides if the business rejects your recommendation, and what gets recorded when they do. That last part protects you, and more importantly it is what makes the decision auditable a year from now. Take it to the owning team before the meeting rather than during it, because a recommendation they have already seen is a decision and one they meet in the room is an ambush.
Done when
- Three options are written with costs and risks stated in the organisation's own terms.
- One is recommended in writing, with your name against it.
- Every condition is specific enough that somebody else could verify it without asking you.
- A review trigger and interval exist for each condition.
- The escalation path is one written sentence, including what gets recorded on an override.
What you end up with
A signed recommendation with three costed options, verifiable conditions, review triggers, and a written escalation path.
If you get stuck
The common failure is the unconditional no, which feels principled and gets routed around inside a week. Offer the conditional yes and make the conditions real. The other failure is refusing to put your name on it, which turns a decision into an opinion and leaves nothing for anyone to audit later. Sign it. If you are then overruled, that record is the point, and producing it is what the role exists for.