Two senior stakeholders want incompatible things and both outrank you. How do you handle it?

Your leverage is framing, not rank

You cannot out-rank either of them, and you should stop looking for a way to. What you control is the shape of the problem when it reaches whoever can settle it: whether it arrives as two positions and an implicit request to pick a favourite, or as a business trade-off with options, consequences, a recommendation and a date by which silence becomes expensive. That reframing is the whole of the job here, and it is what the question is scoring.

Assume you have already done the diagnostic work of tracing each stated requirement back to the objective behind it, because most apparent conflicts dissolve at that level. Do that first, always. This question is about the case where it did not dissolve: the objectives are genuinely in tension, both are legitimate, and someone has to give something up.

Find the decision owner before arguing about the decision

The first practical move is to establish who is entitled to decide. Ask who owns the outcome this feature serves, who holds the budget, and how a decision like this was made the last time. Often nobody has ever articulated it, which is itself the finding: the two stakeholders are not being obstructive, they are both acting inside authority that overlaps, and the organisation has never resolved the overlap.

Three outcomes are common. Sometimes one of the two clearly owns it once the question is asked plainly, and the other's role is to be consulted — which the second person will usually accept if told honestly rather than discovering it later. Sometimes a shared owner exists one level up, a programme director or steering group, and the path is a short paper and a slot on an agenda. And sometimes there is no single owner, in which case name that gap explicitly, because "we have no decision rights for cross-departmental trade-offs in this area" is a governance defect worth surfacing regardless of this feature.

Do not become the channel through which the disagreement is conducted. Shuttling between two offices makes you the owner of a problem you have no authority over, and it lets both parties believe their position was misrepresented rather than declined.

Put the trade-off on one page

A senior person can decide quickly if the decision is prepared, and slowly if it is not. So write it down, keep it to one page, and lead with what you need rather than with history.

DECISION REQUIRED - Credit-check step in corporate onboarding
Owner: Director, Commercial Banking      Needed by: Fri 7 Aug (sprint 24 planning)

The tension
  Risk (Head of Credit) requires the credit decision before an account can transact.
  Sales (Head of Corporate Sales) requires same-day activation for the Q3 campaign.
  Current provider response time is 4 hours to 2 working days; both cannot hold.

Options
  A  Hard block until credit clears.
     Risk appetite met in full. Est. 35% of applications activate next day or later.
     Sales campaign commitment not met.
  B  Activate immediately with a GBP 5,000 transaction cap until credit clears.
     Exposure capped at GBP 5,000 per customer for up to 2 days. Campaign met.
     Needs a new cap rule in payments (est. 8 dev days) and a Credit sign-off on
     the cap value.
  C  Activate immediately, no cap, retrospective review.
     Campaign met. Uncapped exposure. Not consistent with the stated risk policy.

Cost of not deciding
  Sprint 24 planning is Fri. Without a decision the team builds neither path and
  the campaign date is at risk regardless of which option is eventually chosen.

Recommendation
  B. It meets the commercial deadline with a bounded, measurable exposure, and the
  cap is a configuration value Credit can tighten without a release.

Decided: ......................  By: ......................  Date: ..............

Three features of that page do the work. The options are real and mutually exclusive, so nobody can approve the pleasant parts of all three. Each consequence is expressed in the currency the reader cares about — exposure in pounds, applications in per cent, effort in days — and where a number is an estimate it says so rather than pretending to precision. And the cost of delay is stated, which converts silence from a neutral default into a visible choice.

Running the conversation with both in the room

Get them together once, with the paper circulated beforehand, and open by owning the framing rather than the conflict: both requirements are legitimate, they cannot both hold as written, here are three ways forward and here is what each costs. Then stop talking. Your job in the room is to keep the discussion on consequences rather than positions, to correct factual claims with evidence you brought, and to write down what is agreed.

Bring data where it exists, because it changes rooms in a way argument does not. How many applications last quarter would actually have been affected, what the observed provider response times were, what the historical loss rate on uncapped activations looks like. A disagreement about facts is solvable; a disagreement about values needs the decision owner.

If the meeting will clearly turn into a status contest, pre-brief each of them separately first — not to lobby, but so neither is surprised in front of a peer. Being surprised in front of a peer is what hardens a position that would otherwise have moved.

When nobody will decide and the sprint starts Monday

Set a decide-by date the moment the trade-off is documented, and state what happens if it passes. Then split the work by reversibility. Build what is common to all three options, keep the divergent behaviour behind configuration or a feature toggle where that is cheap, and be honest with the team about which parts are provisional so nobody hard-codes an assumption you know is contested.

If the date passes, escalate on schedule and without drama. The escalation is the paper, the record of who was asked and when, and the delivery consequence — never a characterisation of either stakeholder. And do not fill the vacuum by choosing yourself. Building the more senior person's version and calling it pragmatism produces a requirement with no recorded rationale, which fails in user-acceptance testing with you as its only owner.

Escalation without a recommendation is abdication

Here is where most candidates lose the question. Some say they would escalate, full stop, which hands a director two positions and no analysis and takes a week. Others say they would find a compromise, which usually means inventing a middle option neither party asked for that serves neither objective, and which you will then own. A few say they would defer to the more senior stakeholder, which is honest about the politics and abandons the analysis entirely.

The distinguishing answer does three things at once: it names the decision owner, it arrives with a recommendation and its reasoning, and it treats the losing stakeholder as somebody you will need again next quarter. That last part is undervalued. Tell them the outcome yourself before they read it in a status report, tell them accurately why it went the other way, and record the residual risk they raised in the risk log under their name. A stakeholder whose objection is on the record loses gracefully. One whose objection quietly disappeared stops giving you information, and losing your source inside Risk costs you far more than this feature was worth.