A behaviorally anchored rating scale (BARS) is a rating scale in which each score is tied to a written example of what someone at that level actually does, so a top score means the candidate did something specific, not just that the rater was impressed. Interviewers use these anchors to rate answers consistently, and so do some AI interviewers: as of October 2026, HireVue says its AI Interviewer's scores are tied to behavioral anchors (BARS). If you know what a top anchor usually describes, you can make sure your answer contains it.
This guide explains the method, shows a full example scale for one competency, covers how human panels and AI interviewers apply anchored rubrics and ends with a practice routine. Take the assessment yourself and follow the employer's rules — use CoPilot Interview to practice beforehand.
What a behaviorally anchored rating scale is
A BARS has two parts: a number scale, often 1 to 5, and an anchor for some or all of the points that describes observable behavior at that level. The method goes back to a 1963 paper in the Journal of Applied Psychology by Patricia Cain Smith and L. M. Kendall, which set out how to build "unambiguous anchors for rating scales" (Smith and Kendall, 1963).
The original approach started from real examples of effective and ineffective job behavior. A second group of judges matched each example back to the skill it showed and rated how good it was, and the examples the judges agreed on became the anchors. Modern interview rubrics are usually simpler to build, but they keep the core rule: describe behavior you could observe, not adjectives like "strong" or "good."
The U.S. Office of Personnel Management describes the same idea for federal structured interviews, where answers are "scored according to benchmarks of proficiency" (OPM). Its published example rates interpersonal skills from 1 (low) through 3 (average) to 5 (outstanding), with a benchmark definition and a behavioral example at each level (OPM example). For how structured interviews differ from free-form ones, see structured vs unstructured interviews.
BARS vs other rating scales
| Scale type | What a score means | Weakness | Where you'll meet it |
|---|---|---|---|
| Generic 1–5 (poor to excellent) | The rater's overall impression | Two raters read "good" differently | Informal or unstructured interviews |
| Yes/no (meets or doesn't meet) | Whether a requirement was shown | No credit for depth | Screening and qualification questions |
| Behaviorally anchored 1–5 | Which written description your answer matches | Takes work to write good anchors | Structured interviews, and AI interviewers that use anchored rubrics |
| Percentage or 0–100 | A finer-grained version of either | The precision can be false | Some AI interview tools |
Vendors often support several of these. As of October 2026, Ribbon's documentation says custom scores can be out of 5, a percentage or yes/no, and HireVue says skill questions in its AI Interviewer can use a custom 1–5 rubric.
Example: a 1–5 behaviorally anchored rating scale for problem solving
Here's a full scale for one competency. We wrote it as an illustration; real employers write their own anchors for each role, and you usually won't see them.
| Score | Anchor: what the candidate did | What it sounds like (Example) |
|---|---|---|
| 5 Outstanding | Defined the problem precisely, found the root cause with evidence, compared options on stated criteria, acted, measured the result and changed something so it wouldn't recur | "I compared three fixes on speed and risk, shipped the caching change, and added a load test so it couldn't come back." |
| 4 Strong | Found the cause, weighed at least two options, acted and measured the result | "I tested two fixes and chose the one that cut load time in half." |
| 3 Acceptable | Solved the problem with a reasonable fix, but said little about options or results | "I found the slow query and rewrote it, and it was faster." |
| 2 Weak | Described the problem and what the team did, with an unclear personal role or no result | "We looked into it and the team sorted it out." |
| 1 Poor | No relevant example, or passed the problem on without acting | "That wasn't really my area, so I handed it off." |
Look at what the scale rewards: a defined problem, evidence, options, a reason for the choice, a measured result and prevention. Every one of those is something you say. A rater, human or AI, can only give a 5 if those words are in your answer.
How AI interviewers and human panels use anchored rubrics
Human panels
In a structured interview, each interviewer listens, takes notes and matches your answer to the anchor it fits best. OPM describes panels where each member rates responses individually and then the group discusses large scoring differences. Anchors are what make those ratings comparable. The hiring side of this is covered in our guides to writing an interview scorecard and calibrating interviewers.
AI interviewers
AI scoring usually works from the transcript of what you said. As of October 2026, Ribbon's scoring documentation says its AI reviews the full transcript before scoring, and that the employer's description of each criterion holds the guidelines "the AI uses this to evaluate." HireVue's AI Interviewer page says its scores are tied to behavioral anchors and that recruiters always make the final call.
Put simply, the software compares what you said with each level's description and assigns the score whose description fits best. That's why vague answers tend to land in the middle: there's nothing specific for the top anchor to match. Our explainers on AI interview scorecards and what AI interview scoring looks for go further.
How to aim your answer at the top anchor
You rarely see the employer's rubric, but top anchors for behavioral competencies tend to ask for the same ingredients. Build them in on purpose.
- Find the competency. Each question usually targets one skill from the job description, such as problem solving, ownership, collaboration or customer focus. Name it to yourself before you answer. Our guide to competency-based interviews shows how questions map to skills.
- Pick a story with real difficulty. Higher anchors often describe harder situations. In OPM's example, a 5 handles situations "involving a high degree of tension or discomfort."
- Make your own actions the longest part. Say "I," and say what you decided and why.
- Compare options out loud. One sentence on what you considered and why you chose one shows judgment, not just effort.
- End with a measured result and what changed afterward. Prevention, a new process or a lesson you applied later is often what separates a 4 from a 5.
Here's the difference for the question "Tell me about a time you solved a difficult problem at work."
Before (about a 2 on the example scale): "We had an issue where the website got really slow during sales. The team looked into it and we fixed some stuff, and it was better after that."
After (Example, aimed at the 5 anchor): "During our spring sale, checkout pages slowed to eight seconds and orders dropped. I owned the investigation. I pulled the logs and found that one product query ran for every visitor instead of being cached. I compared three fixes: adding servers, rewriting the query, or caching the results. Caching was the fastest and lowest-risk, so I shipped it that night. Load time fell to under two seconds and orders recovered the next day. Then I added a load test to our release checklist, and it caught a similar problem two months later."
Map it to the scale: a defined problem (slow checkout, lost orders), the root cause with evidence (the logs), options on stated criteria (speed and risk), the action (shipping it), a measured result (under two seconds) and prevention (the load test). Every part of the 5 anchor is in the transcript. For structure, see what the STAR method is and our guide to the STAR method for AI interviews.
Practice against a rubric
- Pick three competencies from the job description and draft one story for each with the STAR story builder.
- Score your own draft against the example scale above. If you can't point to options, a result and a follow-up, add them.
- Answer out loud and get an outside score. Our free AI interview assistant includes a Mock Interview mode that scores each spoken answer out of 10 with what to fix and a model answer. Our guide to what a score out of 10 means explains how to read the feedback.
- Re-answer your lowest score the same day, applying one fix at a time, and compare the new score.
The free plan includes 15 minutes every day, with no card and no signup, and the free browser demo has 145 real interview questions across 10 tracks, graded by AI.
Score your answers before an interviewer does
Answer realistic questions out loud and get a score out of 10 with what to fix and a model answer. Free in your browser, nothing to install.
Try a free AI mock interviewFAQ
What is a behaviorally anchored rating scale?
It's a rating scale in which each score is tied to a written description of observable behavior at that level. Instead of rating an answer 'good,' the rater picks the anchor that best matches what the candidate described doing.
Who created behaviorally anchored rating scales?
Patricia Cain Smith and L. M. Kendall described the method in a 1963 paper in the Journal of Applied Psychology, which set out how to build unambiguous anchors for rating scales from examples of real job behavior.
What is an example of a BARS?
For problem solving on a 1 to 5 scale, a 1 might be 'no relevant example' and a 5 'defined the problem, found the root cause, compared options, acted, measured the result and prevented a repeat.' The U.S. Office of Personnel Management publishes an example scale for interpersonal skills.
Do AI interviewers use behaviorally anchored rating scales?
Some say they do. As of October 2026, HireVue says its AI Interviewer scores are tied to behavioral anchors and that recruiters make the final call. Other tools score transcripts against criteria the employer writes, so the rubric depends on the employer.
How do I get the top score on a behaviorally anchored scale?
Pick one real example that fits the competency, make your own actions the longest part, say which options you weighed and why, end with a measured result, and add what you changed afterward so the problem did not come back.
What is the difference between BARS and a regular rating scale?
A regular scale uses labels such as poor to excellent, so the score depends on each rater's judgment. A BARS ties each score to a description of behavior, which makes ratings from different interviewers, or from an AI, easier to compare.