Insights
Use your performance review criteria in interviews, word for word
Performance review criteria can be used as interview criteria without rewording. But the questions companies commonly ask leave some criteria without evidence. Using published criteria from the Japanese civil service, we simulated interviews with 30 personas of final-year university students who differ in experience and speaking style, to see which criteria ran short of evidence and what changed when we added questions.
Why review criteria
When hiring and reviews use the same words, you can check hires against later results
Many companies build their hiring criteria from scratch, separately from their performance reviews. But when hiring and reviews use different words, you cannot tell which criteria led to success after joining, even with interview scores and later review results side by side.
If you use your performance review criteria unchanged as hiring criteria, nothing needs rewording, and interview scores can be checked against later reviews criterion by criterion. For how to check hires against later results, see “Was your interview evaluation right? You find out after the hire”.
What reviews assume
Performance review criteria assume someone is watching day-to-day work
Performance review criteria are written on the assumption that the rater watches how the person works over the review period. For the Japanese civil service, the official personnel evaluation guide gives this as one reason why competency, which is rated separately from job performance, is rated once a year: judging whether the expected behaviour is shown consistently calls for watching over a certain period.
An interview is a conversation of an hour at most. If nothing in the experiences a candidate talks about matches a criterion, there is nothing to rate it on. A 2001 meta-analysis by Huffcutt and colleagues classified 338 ratings from 47 studies and found that personality and social skills were the qualities interviews rated most often. Some things show readily in an interview conversation, and some do not.
Sources: National Personnel Authority and Cabinet Bureau of Personnel Affairs. Personnel evaluation guide: The system as a whole (April 2026 edition) [in Japanese]. https://www.jinji.go.jp/content/000013132.pdf / Huffcutt, A. I., Conway, J. M., Roth, P. L., & Stone, N. J. (2001). Identification and meta-analytic assessment of psychological constructs measured in employment interviews. Journal of Applied Psychology, 86(5), 897–913. https://doi.org/10.1037/0021-9010.86.5.897
Our test
With only the questions companies commonly ask, four criteria often lacked evidence
We used simulated interviews with personas to see which criteria run short of evidence when performance review criteria are used directly as interview criteria.
Simulation setup
- CriteriaThe 12 competency indicators (the points raters look for) used to rate junior staff in central government ministries, used as interview criteria without rewording.Sources: National Personnel Authority and Cabinet Bureau of Personnel Affairs, Personnel evaluation guide: Procedures for evaluatees, April 2026 edition (names); Personnel evaluation manual: Reference materials, September 2021 edition (descriptions). Both in Japanese.
- QuestionsTen questions, covering every topic listed in two surveys of company HR staff about what they ask in interviews.Sources: Gakujo, Survey of HR staff on interviews, February 2023 (70 HR staff; main new-graduate recruitment round); Rikunabi job-hunting guide, Common internship interview questions and tips for answering, survey December 2024 to January 2025 (internship selection). Both in Japanese.
- Personas30 final-year university students, combining 10 kinds of experience (sports club, research, part-time work, study abroad, long internships and so on) with 3 speaking styles (concrete, says little, tends to stay abstract).
- ScoringThe interview records were scored with the same scoring instructions as Proomi. The interviews and the scoring were run separately.
When nothing a candidate says supports a criterion, Proomi can record it as “no evidence”. We scored this simulation that way. The figure below shows, for each criterion, how many personas ended up with no evidence.
Personas with no evidence for each criterion, using the ten common questions (out of 30)
Criteria where evidence came up
3 or fewer out of 30 for all six
Criteria where evidence often failed to come up
12 to 20 out of 30 had no evidence on four criteria
Ten of the competency indicators for junior central government staff, leaving out the two ethics indicators. The interviews are simulations with personas.
None of the 30 personas lacked evidence for persistence, organising information or initiative. Asking about what they put the most effort into as students, or about part-time work and clubs, naturally brings out stories of getting through difficulties and of taking the lead.
By contrast, 20 of the 30 had no evidence for understanding instructions and for reporting to a supervisor, 15 for accuracy, and 12 for speed. The scoring notes repeatedly say things like “nothing is said about receiving and understanding instructions or guidance from a supervisor or others.” All of these criteria assume working day to day under someone more senior, and questions about student life do not bring out that kind of situation.
Speaking style does not explain it. Personas who spoke concretely and personas who said little ended up with about the same number of gaps. What was missing was not the candidates’ stories but questions about the situations those criteria describe.
Sources: National Personnel Authority and Cabinet Bureau of Personnel Affairs. Personnel evaluation guide: Procedures for evaluatees (April 2026 edition) [in Japanese]. https://www.jinji.go.jp/content/900006282.pdf / National Personnel Authority and Cabinet Bureau of Personnel Affairs. Personnel evaluation manual: Reference materials (September 2021 edition) [in Japanese]. https://www.jinji.go.jp/content/900006260.pdf / Gakujo Co., Ltd. (2023, February). Survey of HR staff on interviews [in Japanese]. https://service.gakujo.ne.jp/wp-content/uploads/2023/10/230222-comenq.pdf / Rikunabi job-hunting guide. Common internship interview questions and tips for answering [in Japanese]. https://job.rikunabi.com/contents/internmensetsu/3286/
Adding questions
Adding three questions matched to the criteria closed almost all the gaps
With the same 30 personas, we added three questions matched to the criteria that often lacked evidence and simulated 13-question interviews. For understanding instructions, for example, we added this question.
An example of an added question (for understanding instructions)
Please tell me about a time you worked on something after receiving instructions or guidance from someone above you, such as a more senior colleague or a supervisor.
The same persona tells different stories depending on the question. Below are records from a persona with a part-time cafe job. With the ten questions, asking about part-time work led to a story about teaching new staff, and no evidence came up for understanding instructions. The added question brought out a story about acting on the manager’s instructions.
The same persona before and after adding questions (criterion: understanding instructions)
With the ten common questions
AIWhat do you do in your part-time job at the cafe? If there is an experience that stayed with you, please tell me about it.
CandidateI work the till and make drinks. Since my second year I have also been asked to train new staff for their first week, and I have trained about six people so far.
AIIs there anything you pay attention to when teaching?
CandidateI try not to explain everything on the first day. When I started, I was told everything at once and could not remember it all, so I split it up by day, for example only the till on day one.
No evidenceScoring note: “nothing is said that shows correctly understanding instructions or guidance from a supervisor or others.” What came up was how they teach new staff.
With the added questions (13 in all)
AIPlease tell me about a time you worked on something after receiving instructions or guidance from someone above you, such as a more senior colleague or a supervisor.
CandidateAt my cafe job, the manager asked me to cut food waste. I was told to adjust how much food we prepare by looking at sales on each day of the week.
(part omitted)
AIWhile working on it, did guidance from the manager change how you thought about it?
CandidateAt first I only thought about cutting back, but the manager told me that disappointing customers by selling out is a loss too. So I started adjusting in ways that were not just cuts, like preparing a little more on rainy days or when there was an event nearby.
Evidence foundScoring note (summary): understood the manager’s instruction and adjusted preparation, then took on board the point that selling out is also a loss and changed how they adjusted.
Excerpts from simulation records with a persona, translated from Japanese. Some acknowledgements and lead-ins are omitted.
Personas with no evidence, before and after adding questions (out of 30)
Before: the ten common questions
1 to 6 criteria with no evidence per persona
After: 13 questions, with 3 added for these criteria
Down to 0 to 2 per persona
The same 30 personas, with only the questions changed.
For understanding instructions, personas with no evidence went from 20 of 30 to none. Accuracy and speed also went to none. The number of criteria without evidence per persona fell from 1–6 to 0–2.
Only reporting to a supervisor still had four personas with no evidence. In their records, the stories about mistakes and problems end with fixing things themselves or contacting the other party, such as a customer or an inn, and never with reporting to someone above them. Adding a question is not enough on its own; the way it is asked also has to match the behaviour you want to check.
Which questions you ask also affects what evidence comes up. Acquiring knowledge had only 3 personas with no evidence even with the ten questions, because questions about their university seminar group or research topic and about what they worked on in high school naturally bring out stories of learning.
What AI brings
The interview record shows which criteria lack evidence
An AI interview keeps a scoring note for each criterion. If a lack of supporting remarks is recorded as “no evidence”, cases that cluster on particular criteria tell you that the questions for those criteria are not enough.
Human interviewers tend to give a score even where there is nothing to go on. A 2016 study by Suzuki of final-round new-graduate interviews at a Japanese company also points out that interviewers tend to score on an overall impression rather than criterion by criterion. Once a score is given, the record cannot show whether the criterion was actually checked. For differences in how interviewers interpret criteria, see “Same answer, different score: when interviewers read the criteria differently”.
Source: Suzuki, T. (2016). Inter-rater reliability of recruiting interviews: Experimental study on evaluation components [in Japanese]. Japan Journal of Human Resource Management, 17(1), 69–91. https://doi.org/10.24592/jshrm.17.1_69
Limits
This was a simulation with personas, so the numbers will differ with real candidates
These results come from simulations with personas. Real candidates may say more or less than the personas did, so the number of gaps will differ. The scoring used the same instructions as Proomi. Proomi also chooses the model best suited at the time for scoring, so actual scoring may use a different model from this simulation.
We left the two ethics indicators out of the figures. They are worded for public servants, for example “as a servant of the whole nation”, and they did not come up in new-graduate interviews because the context does not fit, not because ethics is hard to observe in an interview.
When it works
With a service that lets you write your own criteria, you can interview in the words of your reviews
You can use performance review criteria word for word only with a service that lets you write criteria in your own words. If you choose from criteria the service has prepared, you have to fit your own criteria to one of those.
In Proomi, you write the criteria in your own words. You can enter your performance review criteria without rewording, and you can add or change interview questions yourself, so you can adjust the questions to the criteria that lack evidence.
What to compare
Three checks for whether you can use your performance review criteria
When comparing AI interview services, these three checks tell you whether you can use your performance review criteria unchanged.
01
Can you enter your review criteria word for word?
If they have to be reworded, they no longer measure the same thing as your reviews.
02
Is a lack of evidence recorded?
If it is, you can see which criteria need more questions.
03
Can you add or change interview questions yourself?
If you cannot change the questions, you cannot fix the criteria that lack evidence.
Q&A
Frequently asked questions
If we have many review criteria, should we use them all?
You do not need to use them all. Adding criteria does not add usable information in proportion. Narrow them to what you want to check in hiring, and keep the wording of your reviews. On the number of criteria, see “More criteria do not mean more evidence”.
Would mid-career hiring give the same result?
We tested with new-graduate personas. Mid-career candidates have work experience, so evidence for criteria such as reporting to a supervisor and accuracy should come up more easily in stories from previous jobs. Which criteria run short of evidence needs to be checked in the interview records for each type of hiring.
Does adding questions make interviews longer?
Yes. In this simulation the average number of exchanges was 25.8 with ten questions and 32.1 with thirteen. When adding questions, also consider whether you can cut questions aimed at criteria that already have plenty of evidence.
Read next
Once the evidence is there, check it against post-hire outcomes
Once the questions are in place and the evidence comes up, interview scores can be checked against later reviews criterion by criterion. When you rewrite criteria, you can also test them on past interviews. Besides bringing in performance review items, building criteria from scratch with AI support is covered in another article.
Was your interview evaluation right? You find out after the hire →
Check hiring evaluations against post-hire outcomes.
Rewrite your criteria, then test them on past interviews →
Re-score past interviews against rewritten criteria and see how the scores move.
Can AI evaluate candidates correctly? HR writes the criteria →
AI scoring removes the differences between interviewers. How to prepare criteria and refine them by comparing human and AI scores.
Related pages
How it applies, by situation
How you connect review criteria to interviews depends on the hiring situation. If you are at the stage of comparing services, the comparison of the leading services, sorted by what each one optimises for, may also help.
AI interview services compared →
The major services, sorted by what each one optimises for.
Show us your current evaluation criteria
We will look with you at how your performance review criteria could work as interview criteria. Free, about 30 minutes.