Insights

Competency rubric for customer success hiring: anchors, questions and scoring rules

FlourishAIHR · Published November 13, 2026 · Method note

A photograph taken during an interview
A photograph taken during an interview. Photo: CommonMan1212. Public domain, via Wikimedia Commons.

This page is a working rubric for hiring customer success staff: five competencies, the behaviours that mark three levels of each one, the structured questions that draw those behaviours out, and the scoring rules that keep two interviewers' marks comparable. The method follows the US Office of Personnel Management's Structured Interviews: A Practical Guide, and applies the rubric method set out in this site's own guide to structured interview evaluation.

The reason for the structure is consistency. OPM's guide contrasts the two formats directly: in a structured interview all candidates are asked the same questions in the same order, all are evaluated on a common rating scale, and interviewers agree in advance on what acceptable answers look like. Unstructured interviews, the guide notes, typically demonstrate low reliability and low to moderate validity, and their lack of standardisation makes them susceptible to legal challenges; structured interviews have demonstrated a high degree of reliability, validity and legal defensibility. This site's method article makes the same point from practice: same questions, anchored scoring and independent ratings remove the room for first impressions and similarity bias.

The scale: five proficiency levels

OPM's first step in building a rating scale is to specify one proficiency range that applies to every competency. Its guide advises creating at least three proficiency levels, aiming for five to seven, and labelling at least three of them. The five-level scale OPM itself uses, and the one this rubric uses, is:

LevelLabelWhat it means in general
1AwarenessApplies the competency in the simplest situations; requires close and extensive guidance
2BasicApplies the competency in somewhat difficult situations; requires frequent guidance
3IntermediateApplies the competency in difficult situations; requires occasional guidance
4AdvancedApplies the competency in considerably difficult situations; generally requires little or no guidance
5ExpertApplies the competency in exceptionally difficult situations; serves as a key resource and advises others

The rubric table below writes out anchors for levels 1, 3 and 5 only, following this site's method article, which has interviewers score against three written anchors per criterion: what a 1, a 3 and a 5 sound like. Levels 2 and 4 are scored as between-anchor judgements. OPM's instruction to interviewers applies throughout: the behavioural examples are a general guide, not an absolute, because candidates' answers will differ with their experience.

Choosing the competencies

Start from the job, not from a generic list. In OPM's development sequence, a job analysis first identifies the tasks of the role and the competencies they require, and the interview then takes the competencies it can actually observe. OPM notes that a structured interview typically assesses between four and six competencies, and that some, such as Oral Communication and Interpersonal Skills, are particularly well suited to assessment through an interview. This site's method article adds a ceiling from the interviewer's side: five criteria is about the limit of what an interviewer can score reliably in one conversation.

The five below fit a customer success role: the work exercises them daily, and each can be evidenced in an answer. Write a one-line definition for each before writing anchors, the way OPM's guide defines Interpersonal Skills as showing understanding, courtesy, tact and empathy, developing effective relationships, and dealing with individuals who are difficult, hostile or distressed.

A job interview
A job interview. Photo: bpsusf, CC BY 2.0 (https://creativecommons.org/licenses/by/2.0/), via Wikimedia Commons.

The rubric: behavioural anchors for five competencies

Draft anchors from real answers wherever possible: this site's method article recommends taking strong and weak answers from past interviews for the same role, removing the names, and writing the anchor that separates them. Three drafting rules from the same article keep anchors scorable: describe behaviour, not traits; put one behaviour in each anchor; and never write comparisons, because anchors measure the answer against the role's bar, which stays fixed across the candidate pool. OPM adds who should do the drafting: subject matter experts work out how employees at each proficiency level would actually respond, discuss their examples, and keep the ones they agree best reflect each level.

Competency1: below the bar3: at the bar5: clearly above the bar
Customer orientation: reading what the customer is trying to doTalks about the request only in terms of policy or process; the example ends with the ticket closed, not with the customer's problem solvedAsks what the customer is trying to achieve before proposing a fix; checks the customer has understood and agreed the next stepReframes the request around the customer's underlying goal, offers a better route than the one asked for, and follows up unprompted to confirm the outcome landed
Problem diagnosis: finding the cause, not the symptomTreats the first symptom as the cause and jumps to a standard fix; cannot say what evidence would change the diagnosisSeparates symptom from cause with questions or data before acting; names the check that confirmed the causeBuilds a quick test that distinguishes between rival causes, finds the fault is affecting other accounts too, and feeds it back so it stops recurring
Communication: making the answer usableGives long, jargon-heavy answers that bury the action the customer must take; the listener has to ask againStates the answer first, then the reason, then the next step with a time attached; adjusts the level of detail when the customer signals confusionTurns a complex limitation into a plain explanation plus a short written summary the customer can forward internally; answers the follow-up question before it is asked
Difficult conversations: holding a hard call togetherBecomes defensive or hides behind policy wording when challenged; escalates at the first sign of anger, or goes quietLets the customer finish, names the impact of the problem in the customer's terms, states plainly what can and cannot be done, and offers the available alternativeDe-escalates a hostile conversation without conceding what cannot be conceded, agrees a concrete recovery step and a follow-up, and logs the case so the next owner sees the full history
Ownership and follow-through: a case is closed when the customer says soHands cases off without notes; makes follow-up promises loosely and does not track themOwns the case until the customer confirms it is resolved; leaves notes the next person can act on without re-asking the customerSpots the pattern across several cases, raises it with product or engineering with the evidence attached, and closes the loop with every affected customer

Questions keyed to each row of the table

OPM's guide works with two question formats, and this bank uses both for every competency. Behavioural questions ask about past behaviour, on the premise that the best predictor of future behaviour is past behaviour under similar circumstances. Situational questions present a realistic job scenario and ask what the candidate would do, on the premise that intentions are closely tied to behaviour. Whichever format you use, OPM's rules for the questions themselves are: open-ended, clear and concise, free of jargon, reflective of the competencies, and written to draw out the situation, the actions taken, and the impact. Probes are decided in advance: interviewers use very similar probes for all candidates, within a probing range agreed before the interview starts.

Customer orientation. Behavioural: "Tell me about a time when what a customer asked for was not actually what they needed. How did you work that out?" Probe: "What did you ask, or check, before you decided?" Situational: "A long-standing customer asks for a discount your policy does not allow and says a competitor has offered it. Walk me through what you would do."

Problem diagnosis. Behavioural: "Describe the hardest fault you have had to diagnose for a customer. What did you test first, and what ruled it out?" Probe: "What was the check that finally confirmed the cause?" Situational: "Three customers report the same error on the same morning, and your status page shows every system as normal. What are your first three steps?"

Communication. Behavioural: "Tell me about a time you had to explain a limitation of your product to a customer who was depending on it. What did you actually say?" Probe: "How did you know they had understood?" Situational: "You have to tell a customer in writing that a fix will take two weeks longer than promised. Talk me through the message you would send."

Difficult conversations. Behavioural, using OPM's own example question for Interpersonal Skills: "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?" Probe: "What did you say first?" Situational: "A customer joins a call already angry: a promised callback never happened, and their own deadline is today. How do you open the conversation?"

Ownership and follow-through. Behavioural: "Tell me about a case you owned that took weeks to close. How did you keep it moving?" Probe: "What did you write down so someone else could have picked it up?" Situational: "You are about to take a week's leave with two cases mid-flight. What exactly do you hand over, and to whom?"

Scoring rules that keep marks comparable

Each interviewer scores each competency separately, and scores before any discussion. OPM's scoring guidance is that each interviewer records the score he or she gave the candidate for each competency, on the proficiency scale, so a competency receives a 1 (Awareness) to 5 (Expert) from every interviewer. This site's method article is blunt about the ordering: debriefs happen after independent scoring, never before, because a score given after the room has talked is the room's score, not the interviewer's.

Weight the competencies equally unless you have a reason you can write down. OPM recommends equal weights, calling them generally the most effective and defensible course of action in the absence of a clear and documented rationale for doing otherwise, and instructs teams that do use different weights to document the justification. Set the advancing bar before the first interview, not after you have met the candidates: for example, advance candidates with no competency scored at 1 and at most one at 2. When you score, tie the mark to the specific behaviour in the answer that matches, or fails to match, the anchor: OPM instructs interviewers to use the behavioural examples as their frame and to keep ratings tied to specific behavioural examples in the response.

An evaluation agent can prepare this paperwork without touching the decision. As this site's method article describes it, the agent reads the interviewer's notes and drafts a scored evaluation with the sentence it relied on beside each score; the interviewer confirms or changes each score and adds anything the notes missed. The agent does not compare candidates, rank them or recommend advancing or rejecting anyone: those calls stay with the interviewer and the hiring manager, logged under their names.

The calibration drill

Before the first real interview, calibrate. This site's method article sets the drill: every interviewer scores the same two recorded or written mock interviews independently, then the panel compares marks. Where two scores for the same answer differ by two or more points, the anchor is rewritten until the panel converges; the drill repeats quarterly because drift is normal. This mirrors OPM's development sequence, which pilot-tests the questions and the process before use and has subject matter experts build the behavioural examples the panel scores against. After go-live, the method article's four numbers tell you whether the rubric is holding: how often two interviewers' confirmed scores land within one point; how often interviewers keep an agent's proposed score (below about 70 percent, the rubric or anchors need work); time per evaluation, measured rather than estimated; and selection rates by protected group at each stage.

The record each interview leaves

Keep four things for every interview: the question set and rubric version used; each interviewer's independent scores, recorded before the debrief; any score changed after discussion, with the reason; and the final decision with the name of the person who made it. The reason, as this site's method article sets out, is that the federal Uniform Guidelines expect employers to document their selection procedures and their impact, including the documentation of impact and validity evidence covered in 29 CFR 1607.15, and the EEOC's best practices include keeping selection procedures updated so they remain predictive of success in the job. The record is what lets you run the adverse-impact check in the calibration numbers above, and answer a candidate, an auditor or a court that asks how a decision was reached. OPM's guide makes the underlying point: interviews used to make employment decisions are subject to the same legal and psychometric requirements as any other assessment method.

Sources

Put it to work

Run this on your own hiring data

A six-week recruiting pilot measures exactly these things — parser accuracy, reviewer agreement, time per application and compliance records — on your role family, with your reviewers, before you decide anything.