A job posting is a question list in disguise

A job posting tells you what they will ask. Not the wording. The territory.

Interviewers are not inventing questions from a wishlist. They start from a job analysis of the role, pull out the competencies it says matter, and write a question for each. A job posting is a description of that same role. Read the two side by side and the posting is a question list with the questions removed.

Here are four mappings you can check yourself, the part of the rating scale almost nobody explains, and an honest limit I could not find a source for.

Why the posting and the question are the same object

The US Office of Personnel Management publishes its own method for building a structured interview, step by step. Step one is to conduct a job analysis, and the guide defines the purpose plainly: to identify the requirements of the job and the competencies necessary to perform them. The guide then lists where the analysis gets its material. Position descriptions. Classification standards. Task statements. Interviews with subject matter experts.

A posting is a position description. So the posting is an input to the analysis, and the questions are an output of it. The same job, described twice.

Everything downstream follows from that. The guide says interview questions should be reflective of competencies derived from a job analysis and realistic of the responsibilities of the job, that each question should measure at least one specified competency, and that a structured interview typically assesses between four and six of them. SHRM organizes its published question bank the same way: choose the most important attributes required to be effective in the job, then gather questions from each selected competency.

This is also why structure beats conversation. Schmidt and Hunter report a validity correlation of r = .51 for structured interviews against .38 for unstructured ones. Other frameworks report uncorrected values near .25 for the same construct. Those are not contradictory measurements of one quantity; they are different corrections. The directional claim survives both.

Four mappings, printed so you can check them

Each of the first three blockquotes is a real line from a published O*NET occupation profile, not a paraphrase. The fourth is the employer publishing its own mapping. The question is the one the line implies.

Software developer

O*NET lists this as a core task for Software Developers, at importance 69:

Modify existing software to correct errors, adapt it to new hardware, or upgrade interfaces and improve performance.

That implies: Tell me about the most difficult bug you ever tracked down. What were the symptoms, and how did you narrow it down?

What is being scored is the debugging process, how you eliminated possibilities, and the judgement call that finally separated the cause from the noise. O*NET files this task under "Making Decisions and Solving Problems", which it rates at importance 84 for the occupation. Nobody is asking you to recite a list of tools. They want to see whether you can act on incomplete information.

Registered nurse

O*NET lists this among the core tasks for Registered Nurses:

Administer medications to patients and monitor patients for reactions or side effects.

That implies: Describe a medication error or a near miss. What happened, who did you tell, and what changed afterwards?

Patient safety, honesty about your own part in it, whether you escalated rather than absorbed it, and whether the corrective action held. The near-miss half is not the softer half: reporting the thing that nearly went wrong is the behaviour being looked for, and a candidate with only error stories has half an answer.

Financial and investment analyst

O*NET lists this among the core tasks for Financial and Investment Analysts:

Assess financial status of clients.

The same profile rates two work styles at the top: Attention to Detail at impact 100, and Integrity at impact 93.

That implies: Tell me about a time you found an error in financial data before it reached anyone else. How did you find it, and what did you do?

The scoring target is the control that let the error through, not the error. A story that ends with you catching it is worth less than one that ends with the process changing.

One more analyst detail, the same mechanism at smaller scale. O*NET defines its In Demand skills as those frequently included in employer job postings, and for this role they are SQL, Microsoft Excel, Power BI, Tableau and SAP software. A posting naming one of those is signalling a competency, and a competency gets a question. Expect to be asked what you built with it and what decision it changed.

Client-facing roles, mapped by the employer in public

This one is not from a job ad. It is the US Office of Personnel Management doing the mapping itself and publishing both halves. The competency is Interpersonal Skills, defined as:

Shows understanding, friendliness, courtesy, tact, empathy, concern, and politeness to others; develops and maintains effective relationships with others; may include effectively dealing with individuals who are difficult, hostile, or distressed.

From that definition the guide writes a behavioural question: Describe a situation in which you had to deal with individuals who were difficult, hostile, or distressed. Who was involved? What specific actions did you take and what was the result? And a situational question: A very angry client walks up to your desk. She says she was told your office sent her an overdue check five days ago. She claims she has not received the check. She says she has bills to pay, and no one will help her. How would you handle this situation?

Any advert asking you to handle escalations, difficult stakeholders or customer issues end to end is asking for that competency. The interview question writes itself.

One practical note from the same guide, and the most usable thing in it. It instructs the subject matter experts writing questions to use superlative adjectives: the most, the last, the worst, the least. So when a question says "the most difficult" or "the last time", you are being asked for one bounded incident with a beginning and an end, not a summary of a competence.

What the rating scale actually rewards

Start with the part almost nobody tells you. On a four-point rating scale, a 3 is conventionally the minimum level expected for successful performance in the role. It does not mean average candidate, and it does not mean a percentile of whoever else applied (Campion, Pursell and Brown; interview loop design).

Here is the same anchoring in a scale OPM publishes in full. Its five levels are defined by what the work demands, not by how the applicant pool compares. Level 3, Intermediate: applies the competency in difficult situations, requires occasional guidance. Level 4, Advanced: applies it in considerably difficult situations and generally requires little or no guidance.

A rating is a judgement about the difficulty of the situation handled and how much hand-holding was needed. That is a fact about the work, so you can answer to it in advance. It is also why interviewers are told to recalibrate once they notice the drift where everyone is scoring every candidate 3 out of 4 regardless of performance.

What the anchors reward is a short list: observable behaviour, quality of reasoning, the degree of difficulty, the candidate's own personal contribution, plus the outcome and any critical failure. A published prioritization rubric shows the gap. A 4 identifies constraints first, applies a clear framework consistently, makes trade-offs explicit and acknowledges uncertainty. A 2 prioritizes on gut instinct with reasoning that is unclear or post-hoc, and relies on the interviewer to find the constraints. Same day, opposite ratings, and the difference is entirely in how you reasoned.

SHRM makes the reasoning point in one line: the applicant's reasoning abilities and the thought process that leads to the answer often are as important as the answer itself.

The debrief is built the same way: review evidence by competency, separating evidence from interpretation, and flag disagreement of two or more points. Practically, written scores go in before anyone talks, the least confident interviewer speaks first, and a single well-argued no outranks three softer yes votes.

What kills an answer is not structure. It is a script. Hiring managers probe rehearsed answers precisely because scripts collapse under follow-up questions, and the documented failure pattern is candidates forcing one prepared story into every question. The defensible version of the advice is short: use the structure to think, not to speak. The argument for that is here.

The limit, which is not small

No published source measures how often a posting line becomes the question actually asked. That is a gap in the literature, not a caveat I am adding for tone.

The mechanism is documented. Questions are constructed from a job analysis, and a posting is one of the documents the analysis is built from. The precision is documented nowhere I could find. The closest thing to a research base is that job posting text is systematically analysable, as the job-ad text mining and job-task-skill pattern analysis literatures show. That is what makes a requirement sentence a weak-to-moderate predictor of the competency that will be probed.

So the honest claim is a location claim, not a prediction claim. The posting tells you where to look, not what will be said. Anyone selling you the second is selling you the confidence rather than the accuracy.

Doing the mapping mechanically

Rehearsea does the clerical part of this. You paste a posting; it returns a structured job spec in which every requirement carries the verbatim line from the advert it came from, so you can check the extraction against the source rather than trust it. It then sizes a study plan to the hours you actually have and orders the phases by interview risk, and it grades mock answers on five dimensions that the server combines into a total the model does not control. It does not predict whether you will get the job, and it does not claim to.

What to do on Monday

Requirement, competency, incident. Take one requirement line from the posting. Name the competency it implies. Find one real incident where you demonstrated that competency at the difficulty the scale calls for, and write it down with the actions, the decision, the difficulty, and the result.

Do that for the four or six requirements the posting actually emphasises. Not the ones you find most interesting. The ones the analysis would have kept.