The useful result
A one-page table of likely questions, each tied to a quoted line in the job post and to one true example you can tell.
- Build a question map that links each job-post requirement to likely behavioural and situational questions.
- Match each question to one of your real examples and find the gaps to prepare.
Many interviews are built from the job description. Employers often use structured interviews, where every candidate gets the same job-related questions. A large 2022 review of hiring research by Sackett and colleagues, summarised by SIOP, found structured interviews had the highest average validity of the methods it looked at (about r = .42), while also noting wide spread between studies. That is a good reason to prepare from the post itself.
This lesson turns the post into a question map. The AI drafts likely questions; you supply the real examples. The questions are a guess, not a leak. The real interview may include others, so the goal is to have true stories ready for each main requirement.
Before you start
- The full text of one job post.
- A short list of your real work, study or volunteer examples.
- Any free AI chat assistant.
Review evidence
Documentation-based review · 2026-10-28
Sources opened and checked on 28 October 2026. Survey figures are dated and mostly US-based; they describe what people reported, not rules. Prompts and examples are illustrative; no model output or job-search outcome is claimed.
Map every requirement to two questions
Paste the whole job post. Job posts are public, but leave out anything you received privately. Ask the AI to quote each duty or requirement exactly and write two questions for it: a behavioural one ("Tell me about a time...") and a situational one ("What would you do if..."). Asking for exact quotes stops it from inventing requirements that are not in the post.
Here is a job post. Step 1: list each duty or requirement, quoted word for word. Do not add any that are not written. Step 2: for each one, write one behavioural question ("Tell me about a time...") and one situational question ("What would you do if...") an interviewer could ask. Step 3: mark each requirement HIGH if it is repeated or near the top, otherwise NORMAL, and say this is only a guess. Answer as a table: Requirement (quote) | Behavioural | Situational | Priority.
JOB POST:
[PASTE JOB POST]Example / expected result
Illustrative example; not a recorded model result.
"Resolve customer issues by phone and email" | Tell me about a time you calmed an upset customer. | What would you do if a customer asked for a refund you cannot give? | HIGH (appears twice)
Check your result
You can find every quoted requirement in the original post. Delete any row whose quote is not there.
Match each question to a real example
Under the table, write one line per requirement naming a real example from your own life: a job, a course, volunteering, caring work, a side project. Write "none yet" where you have nothing. Then ask the AI to sort the table and ask you questions where your example is thin. It must not invent examples.
Here are my real examples, numbered to match your table:
[1. YOUR EXAMPLE OR "none yet"]
[2. ...]
Add a column "My example" using only what I wrote. Add a column "Gap" saying STRONG, THIN or MISSING. For each THIN or MISSING row, ask me one question that could help me find a true example. Do not invent examples, results or numbers. If something is unknown, write [?].Example / expected result
Illustrative example; not a recorded model result.
Row 1: My example: night shift at a pharmacy, calmed a customer about a late prescription | Gap: THIN | Question: What did you say or do, and what happened next?
Check your result
The "My example" column contains only examples you wrote yourself. Nothing new has appeared that you did not do.
Rehearse the top five with follow-ups
Pick the five HIGH rows. For each, prepare a short story with a simple shape: the situation, your task, what you did, and the result. Then ask the AI to interview you one question at a time and push with one follow-up, the way a careful interviewer might. Answer out loud or type, then look at its feedback.
Interview me for this job using these five questions, one at a time:
[PASTE FIVE QUESTIONS]
After each of my answers, ask one follow-up such as "What exactly did you do?" or "How did you know it worked?". Then give two short notes: one thing that was clear, one thing that was vague. Do not add facts to my stories. Wait for my answer before the next question.Example / expected result
Illustrative example; not a recorded model result.
Follow-up: You said the customer left happy. How did you know? Note: Clear - you named what you said. Vague - the result; say what actually happened, for example they came back the next week, if that is true.
Check your result
You can answer each follow-up without notes, and every result you mention is something that really happened.
When it goes wrong
The AI cannot know what a specific employer will ask. Its priority marks are guesses based on the wording of the post. It may also write questions that are too technical or too generic for the role. Do not memorise scripts word for word; prepare true stories you can tell in different ways. Some employers use AI-led or recorded interviews; prepare the same way, and it is fine to ask how answers are reviewed.
Make it your own
Take one MISSING row and spend ten minutes finding a true example from any part of your life. Write it in four short lines and test it with one follow-up question.
Useful next resources
Sources and review
- SIOP: Is cognitive ability the best predictor of job performance? (Sackett et al. 2022)
- University of Sheffield Careers: Using AI for applications
Last checked Oct 28, 2026. Features and account limits can change. Verify important results against your original material.
Did this help you complete your task?
Tell us what needs improving. No account or private output required.
Report incorrect information or a broken linkGenius Forges Editorial
Practical job-search exercises based on checked sources. Examples are illustrations, not recorded model results.
