Solutions
Own the criteria for each role, and use them beyond the offer
What you look for changes with the role, but your hiring managers’ diaries are already full. The AI Interviewer runs first interviews 24 hours a day, 365 days a year, many at once. You define the criteria for each role yourself, so selection and later performance reviews use the same yardstick.
The problem
The criteria change with the role, and so do the interviewers
In mid-career (experienced) hiring, what matters differs from role to role, and the interviewer is usually a line manager for whom hiring is not the day job. So the criteria stay unwritten, in each interviewer’s head.
- Managers’ diaries fill up, and it takes one to two weeks to reach a first interview
- In the meantime, the candidate moves further along another company’s process
- Interviewers change from role to role, and even within one role the bar drifts
- Selection scores are never compared with later performance reviews, so whether a hire worked out stays a matter of opinion
Building the criteria
Write the criteria for each role as behaviour
What do you look at, role by role? Japan’s Ministry of Health, Labour and Welfare publishes the Vocational Ability Evaluation Standards as a model. Work is broken down into occupation, job, ability unit and ability item, and behaviour that leads to results is set out as the standards for job performance. They are published for 56 industries; the clerical occupations cover nine occupations and nineteen jobs.
Those standards distinguish levels by the scope of responsibility and the difficulty expected. Apply the same idea, written so that interview answers can be scored directly, and you get something like this, using sales as the example on five levels.
Facing the customer
Framing the problem
Showed the customer a problem they had not yet named, and set the agenda
Checked what lay behind the request and redefined the problem to be solved
Made the request concrete and confirmed, in the customer’s words, why it mattered
Captured the request accurately and turned it into a proposal as stated
Can recount how the deal went, but not what they treated as the problem
Moving the organisation
Getting it done
Where the usual approval route would not work, reworked the terms with the departments involved until they were approved
Brought other departments in early and had agreement before the proposal went out
Asked the right departments and had documents and quotes ready on time
Followed a manager’s instruction and worked through the necessary steps
On internal work, describes only the part they handled
Passing it on
Repeatability
Put the reason it worked into words that can be tested, so others now use it
Organised what worked into a hypothesis and tried it on the next deal
Can explain, in their own words, why the result came about
Can point to results, but attributes them to the client or the circumstances
Reports the result without explaining why it happened
Sources
How ability is broken down — Ministry of Health, Labour and Welfare, “Vocational Ability Evaluation Standards.” Work is divided into occupation, job, ability unit and ability item, with behaviour that leads to results set out as the standards for job performance. Published for 56 industries; the clerical occupations cover nine occupations and nineteen jobs.
How the levels work — the same standards set four levels (1–4) by the scope of responsibility and the difficulty expected. The five levels above are a restatement to match a five-point score and do not map onto the official four.
The L1–L5 wording — written by us, following that format. It is not officially defined.
AI interview services differ in whether this format can be written differently for each role. Where the criteria for sales and for engineering can each be defined as free text, the definitions you already use in performance reviews go straight in.
What to settle
What to decide before you start mid-career hiring
01
Whether the criteria differ by role
One set across every role, or a separate set for each. How many sets you can hold, and how freely you can rewrite them, decides which type of service fits.
02
Who owns the first interview
Whether the line manager joins from the first interview, or someone screens candidates first. If the AI handles it, the manager starts at the second interview, with the scores and the quotes behind them already in hand.
03
Whether it shares a yardstick with performance reviews
Write the selection criteria in the same words as your performance definitions and the two can be compared. You can then judge which hires worked out from data, not impressions.
Running it
The cycle for one opening
Mid-career hiring runs all year round. A cycle does not end when the opening closes: it continues through the post-hire performance review to the criteria you set for the next opening.
01
Opening the role
Set the criteria for that role. If your performance reviews already define them, load those.
02
First interview
Send the URL the day the application arrives. No scheduling needed, so the first interview is done before other companies have moved.
03
Handover to the manager
Points worth probing are passed on. The hiring manager reads the remarks behind each score before meeting anyone.
04
Joining and ramp-up
Use the selection scores to shape onboarding. You already know which criteria the new hire scored low on, so early support can be targeted.
05
Revising the criteria
Compare the selection scores with later performance reviews to see which criteria made the difference. Revise them role by role.
To the next opening
An example: one opening in five stages. Where the boundaries fall depends on your own selection process.
Stages 4 and 5 assume the criteria are yours. If you own them, you can check how new hires settle in and feed that back into the criteria. A vendor’s standard framework can run the first interview, but if you want to add an axis when a new role appears, or judge candidates by the same standard as your post-hire reviews, criteria you define yourself are the better fit.
Related pages
More detail, by topic
Related topics are covered on these pages. Start with whichever is closest to your question.
Hand the first interview to AI so hiring managers start at the second. Scheduling and write-ups no longer take up their time.
When scores differ by interviewer: one set of criteria for everyone, with the reasons on record.
How far you can narrow the field before the interview, when the paperwork is not enough.
How to choose an AI interview tool →
What to put in your comparison table, and why who owns the criteria decides which service fits.
AI interview services compared →
The leading services, sorted by what they optimise for and compared on questions, criteria and verification, using public information.
Rewrite your criteria, then test them on past interviews →
Re-score past interviews against rewritten criteria. What the full transcript can test, and what it cannot.
Frequently asked questions
Can the criteria differ by role?
They can. Criteria are set per opening, so sales and engineering can have separate definitions. The name of each criterion and the wording of each level are free text, so adding a role does not affect the criteria for existing roles.
Will hiring managers accept it?
A score on its own does not convince anyone. Each score is kept with the remark it is based on and the recording, so the manager can check it themselves. The AI’s score and the interviewer’s score are recorded separately, and the decision stays with people.
Do we have to use the Vocational Ability Evaluation Standards?
No. This page shows an example built on the format of a published standard. If you already have role-by-role competency definitions, set those as the criteria. If the criteria use the same wording as your performance reviews, selection and later performance reviews share one yardstick.
Can interviews run in English?
Yes. Interviews run in Japanese and English. An interview held in English can be read in Japanese — both the summary and the reasoning behind each score — so first interviews keep moving even when few of your staff can interview in English. For other languages, talk to us.
How much faster is it from application to first interview?
No scheduling is needed, so the URL can go out the day the application arrives. Candidates choose when to take it, so those who are still employed are not held up.
Is there a minimum volume?
Flex is the easiest way to start: you begin with a minimum of 100 interviews and can top up in blocks of 10. Around 100 interviews is also when results usually start to show. For steady use there is Starter (240 interviews a year), and for higher volumes Growth (500 a year). See pricing
What if we have not written criteria for each role yet?
You do not need to have them written. We can build them with you, from your performance-review definitions and from what your interviewers actually look at. What matters is that the finished criteria are yours, and that you can rewrite them afterwards.
Tell us which roles you are hiring for, and how you judge them
We will go through your criteria for each role and your current process, and show you which steps the AI can take over. Free, about 30 minutes.