September 7, 2026 · 7 min read
TL;DR: Options that appoint a named owner, state who answers for an outcome, or take a decision to the person accountable for it are keyed 8 times against 1 as a distractor in this bank. Nine appearances is a small sample and worth saying so plainly. But the reasoning behind those eight generalises well past the count: they all key in scenarios where something important has no owner, which means nobody was ever going to act on it.
Most of what we publish here is about which answers to distrust. This one runs the other way, and it is worth the space because positive patterns are rarer and harder to find than traps.
Across the bank, options whose text names who is accountable, who owns a decision, or who answers for an outcome come up nine times. Eight are keyed. One is a distractor. That is the most lopsided positive lean we have measured, more extreme even than the diagnose-first pattern at 25 keyed to 5.
The honest caveat comes first: nine is not a big number. You should not walk into the exam planning to pick any option containing the word "accountable." What makes this worth learning is that the eight keyed cases share a recognisable setup, and once you can see the setup you do not need the count.
Something matters, and nobody is responsible for it. Not "nobody is doing it" — nobody is responsible for it, so nothing will trigger anyone to look.
An insurer put a machine learning model in front of its claims process to sort claims into fast-track and full-assessment queues. It was accurate at go-live. Two months later, as the claim mix shifted underneath it, it is fast-tracking a third fewer claims and one that plainly belonged there sat for nine weeks. Nobody changed anything. The service desk, the data team and the claims manager each say the routing is not theirs.
The options offer retraining the model on the claims it routed wrongly, adding a manual override so the claims manager can move individual claims, appointing a named owner who answers for the routing and has authority to change it, and reporting the nine-week claim to the project board as a defect.
The named owner is keyed, and the stem tells you why in one sentence: nobody changed anything. Nothing broke. The model is doing exactly what it was built to do on a claim mix that has moved. There is no fault for anyone to raise and no moment at which someone was supposed to look. Three teams can each be honestly right that the routing is not theirs when nobody was ever made accountable for it. Retraining fixes today's drift and leaves the next one equally unowned.
| What the stem shows | Keyed move | The tempting alternative |
|---|---|---|
| Three teams each say it is not theirs | Appoint an owner with authority to change it | Fix the current instance |
| A commitment reported faithfully and ignored | Have the board name who answers for it | Present the figures more prominently |
| A risk reduced as far as it can go | Have the accountable director accept the residue | Log it in the risk register and proceed |
| A funding gap mid-project | Take it to the sponsor, accountable for funding | Rewrite the budget to fit the money left |
| A problem whose cause sits outside the project | Escalate to whoever answers for the affected service | Manage it as a project risk |
The person whose name is on the outcome.
An awarding organization is preparing a machine learning marker for short written answers in a national qualification. Everything reducible has been reduced: the question types it handled badly are excluded, appeals go to a person, and moderation of a large sample puts its accuracy level with the examiner panel. One exposure remains. A candidate who challenges a mark expects to be told why it was awarded, the organization has always been able to say why, and the supplier's engineers cannot produce that reason for any script. Grades are issued in the organization's name and defended publicly by its director of awarding.
Three options handle it as project work: ask the supplier to produce reasoning for a sample and hold go-live until they can, record the residual exposure in the risk register and go live, or estimate the residual mis-marking rate from the moderated sample and accept it within the project's own tolerance.
The keyed answer has the director who answers for the organization's grades accept what is left of the exposure, and records that acceptance. The moderated sample answers how often marks are wrong. It cannot size this exposure at all, because any mark may be challenged and the organization has no account of how it was reached. What it stands to lose turns on who challenges, not on a rate. That is not a risk a project tolerance can absorb, and the project manager does not own the thing exposed.
Common trap: treating the risk register as the place accountability goes to be resolved. Across this bank, options that add something to the risk register and proceed are keyed once against 14 as distractors. Recording an exposure is bookkeeping. It creates no owner, no threshold and no moment at which anybody has to act, which is exactly the failure the eight keyed options above are each fixing. The register is where an owned risk is tracked, not where an unowned one becomes managed.
When you name the wrong one.
The single distractor in the set escalates to the product owner "who owns this backlog," in a scenario where the backlog was not what was stuck. The phrasing is identical to the keyed options; the fit is not. That is the discriminator, and it also explains the keyed escalation cases. A pension service's generative AI assistant produces materially different explanations for the same member's case from one run to the next, on a platform procured centrally and configured identically for five services. Neither cause nor remedy sits inside the project. The keyed answer escalates to the head of member services, who answers for the advice members are given, and stops managing it as a project risk once that function has taken it on.
Compare that with the ordinary escalation trap, where escalation is keyed 27 times against 227 in this bank. The difference is not the act of escalating. It is whether the option names a person whose accountability actually covers the stuck thing, and whether the project manager hands the problem over or merely reports it upward.
Is naming an accountable owner usually the right answer on the PMP exam? In this bank it is keyed 8 times against 1 as a distractor. That is only 9 appearances, so treat it as a strong lean rather than a rule. What makes it worth carrying is the reasoning: these options key when the scenario describes something with no owner and therefore no mechanism for anyone to act.
When does the accountable-owner option fail? When the named person is the wrong one for the decision at hand. The single distractor escalates to the product owner because they own the backlog, in a scenario where the backlog was not the problem. The option still has to name someone whose accountability covers the thing that is stuck.
Why does the exam favour appointing an owner over fixing the problem directly? Because several of these scenarios describe a problem that will recur. The claims-routing model drifts as the claim mix shifts, so retraining it once fixes today's drift and leaves the next one equally unowned.
Who accepts a risk that cannot be reduced any further? The person accountable for the outcome that is exposed, not the project manager and not the risk register. In the exam-marking question, the keyed answer has the director who answers for the organization's grades accept what remains and records that acceptance.
PMP Practice's 2,141 questions are re-certified against PMBOK 8 and the July 2026 exam content outline, with every wrong answer explained rather than just marked wrong, which is how a nine-question pattern turns into something you can actually recognise under time pressure. Start the free 20-question sample — no card, no signup required to try it.
Is naming an accountable owner usually the right answer on the PMP exam?
In this bank it is keyed 8 times against 1 as a distractor. That is only 9 appearances, so treat it as a strong lean rather than a rule. What makes it worth carrying is the reasoning: these options key when the scenario describes something that has no owner and therefore no mechanism for anyone to act, which is a situation you can recognise from the stem without counting anything.
When does the accountable-owner option fail?
When the named person is the wrong one for the decision at hand. The single distractor in this bank escalates to the product owner on the grounds that they own the backlog, in a scenario where the backlog was not the problem. Naming an owner is not a magic phrase; the option still has to name someone whose accountability actually covers the thing that is stuck.
Why does the exam favour appointing an owner over fixing the problem directly?
Because several of these scenarios describe a problem that will recur. An insurance claims-routing model drifts as the claim mix shifts, with three teams each saying the routing is not theirs. Retraining it once fixes today's drift and leaves the next one equally unowned, which is why the keyed answer appoints someone who answers for the routing continuously.
Who accepts a risk that cannot be reduced any further?
The person accountable for the outcome that is exposed, not the project manager and not the risk register. In an exam-marking question where every available reduction has been made, the keyed answer has the director who answers for the organization's grades accept what remains and records that acceptance.