Two interviews can be of the same duration and produce decisions with very different levels of quality. The difference is not in “asking good questions” in a generic way, but in deciding before the conversation what needs to be observed, what evidence counts and how different answers will be scored.
A structured interview does just that: it uses previously defined questions, applied consistently, and common criteria to evaluate responses. The U.S. Office of Personnel Management (OPM) and the CIPD describe the use of the same questions, previously agreed scoring criteria, and a common rating scale as central components of this format.
This does not make the interview foolproof or eliminate all bias. However, it makes the decision more traceable: instead of comparing “who made the better impression”, the team compares evidence related to the work. A meta-analysis published in International Journal of Selection and Assessment found relevant predictive capacity of interview assessments and indicated benefits of formal scoring procedures, especially for contextual performance aspects such as collaborative behaviors.
Start with the vacancy, not the question bank
The error usually appears before the first question. The job description contains expressions such as “good communication”, “analytical profile”, “proactivity” or “ability to work under pressure”. They may serve as labels, but they still don't tell the interviewer what to look for.
The first conversion is from abstract requirement to work-relevant behavior.
OPM recommends starting with a job analysis: identifying tasks and responsibilities, determining which skills are needed to perform them and separating those that already need to be present at entry from those that can be developed later. It also recommends confirming critical skills with people who actually know the role.
Before turning a requirement into a question, four questions help to clean it up:
- What real work situation requires this skill?
- What does a competent person do in this situation?
- What behavior would differentiate a weak execution from an adequate or strong one?
- Is an interview really the best way to measure this?
The fourth question avoids trying to evaluate everything by talking. Technical knowledge, written output, data analysis, or execution of a task can best be observed in tests or work samples. The CIPD recommends combining the interview with other evidence when this makes sense for the role.
As an operational reference, the OPM guide typically addresses four to six competencies in a structured interview, although this is not a universal rule and may vary by role.
A possible transformation would be this:
| Job requirement | Skills to be assessed | Evidence that matters | Structured question |
|---|---|---|---|
| “Knowing how to work with multiple priorities” | Prioritization | Defines criteria, evaluates impact, renegotiates when necessary and monitors what remains pending | “Tell me about a recent situation in which you had to deal with important demands competing for the same deadline. How did you decide what to do first and what was the result?” |
| “Good communication with different areas” | Stakeholder management | Adapts the message, confirms understanding, handles disagreements and keeps those involved informed | “Describe a situation in which two areas had different expectations about a deliverable under your responsibility. What did you do?” |
| “Analytics profile” | Analysis and decision making | Verifies data, identifies hypotheses, compares alternatives, and recognizes limitations | “Tell me about a relevant decision you had to make with incomplete information. What data did you use, what was missing and how did you arrive at the decision?” |
The table is not intended to provide universal questions. It shows the logic: each question arises from a real requirement of the role and seeks specific evidence.
Behavioral and situational questions serve different functions
Structured interviews may use questions about past experiences or present hypothetical work-related situations. Both formats appear in the OPM guidance.
The behavioral question looks for something that has already happened:
“Tell me about a situation when...”
It works best when asking for a specific episode and allows you to identify context, individual action and result. Questions inspired by STAR logic — situation or task, action and result — help remove the answer from the field of generalities.
The situational question presents a plausible role problem:
“Imagine that you receive two critical demands, from different managers, with the same deadline. How would you proceed?”
It is useful when candidates may not have had exactly the same experience, especially in entry-level positions or career changes.
In both cases, the interviewer needs to distinguish an articulate response from strong evidence. Talking well about collaboration does not necessarily mean demonstrating collaboration. The score should depend on the content observed, not on fluency, friendliness or similarity between interviewer and candidate.
It is also advisable to standardize in-depth questions. Examples such as “what specifically was your action?”, “how did you decide?” and “what was the result?” can be predicted in the roadmap to obtain enough information without leading different candidates down completely different paths.
The grade anchor solves the problem that the scale alone does not
Putting a “grade from 1 to 5” on the form does not structure an interview. If no one has defined what differentiates 2 from 4, each interviewer brings their own reference to the room.
Behavioral anchors give meaning to the scale. The OPM guide recommends using a common range of proficiency and developing examples of behavior that help evaluators distinguish between levels. These examples serve as reference, not as answers that the candidate needs to repeat verbatim.
For the competency “prioritization under competing demands”, a scale could be drawn like this:
| Note | Evaluation anchor |
|---|---|
| 1 | Acts without clear criteria, simply responds to those who put the most pressure or leaves critical demands unaddressed. It does not identify impact or communicate risks. |
| 2 | Recognizes the conflict and chooses an order, but uses criteria that are fragile or poorly connected to the impact. Communication and follow-up are incomplete. |
| 3 | Compares urgency and impact, defines a justifiable order, communicates constraints and tracks deliveries. |
| 4 | Beyond the expected level, it anticipates dependencies, negotiates priorities with those involved and creates alternatives to reduce the impact of the conflict. |
| 5 | Demonstrates consistent performance in highly complex situations, explains trade-offs, mobilizes stakeholders, protects critical objectives and leaves a better process for future situations. |
A good anchor describes behaviors and decision quality. A bad anchor uses adjectives: “weak”, “good”, “excellent”, “senior profile”. These labels just transfer subjectivity to another column.
It is also useful to establish in advance what should not increase the grade. A long answer, sophisticated language, or a positive result obtained by external circumstances do not replace evidence that the person performed the behaviors evaluated.
Scorecard needs to record grade and evidence
The scorecard is the link between the individual interview and the subsequent comparison. Your role is not to create an appearance of mathematical precision, but to prevent the process from ending in phrases like “I liked her better”, “he seems to have more energy” or “I think he fits better on the team”.
A lean model can use this structure:
| Competence | Weight | Grade from 1 to 5 | Recorded evidence |
|---|---|---|---|
| Critical Competency 1 | 25% | Reported behaviors, decisions, and outcomes | |
| Critical Competency 2 | 25% | Reported behaviors, decisions, and outcomes | |
| Critical Competency 3 | 25% | Reported behaviors, decisions, and outcomes | |
| Critical Competency 4 | 25% | Reported behaviors, decisions, and outcomes |
Equal weights are a simple solution when there is no clear justification for differentiating skills. If the job analysis shows that one of them is materially more important, the weight may change — but this decision must occur before meeting the candidates and be documented.
The same goes for minimum grades. If a certain skill is truly essential upon entry, it makes sense to define a minimum level for it in advance. Therefore, a high total score does not automatically compensate for a serious deficiency in a critical requirement.
With weights that add up to 100%, the consolidated score can be calculated by adding the score for each competency multiplied by the respective weight. But the final number must be accompanied by evidence. A 4 without useful annotation is difficult to audit; “Identified three competing demands, explained the impact criteria, renegotiated the deadline with an area and monitored delivery” allows us to reconstruct the judgment.
Give the grade before starting the debate on the candidate
When there is more than one interviewer, an important practice is for each evaluator to record their observations and assign scores before collective discussion.
The CIPD recommends that interviewers carry out independent assessments before debating candidates, reducing the possibility of someone else's initial opinion — including someone more senior — influencing other scores. It also recommends preparing the panel in advance and using common questions and criteria.
A possible routine is:
- Everyone receives the same script, skills and anchors before the interview.
- The candidate receives the main questions under the same conditions as the others.
- Each interviewer writes down evidence, avoiding global assessments during the conversation.
- Individual grades are recorded without consulting classmates' grades.
- Only then does the panel compare divergences.
- Relevant scoring differences are discussed based on the evidence and anchors, not the hierarchy between raters.
- The final decision and any exceptions are documented.
This sequence also helps identify a design problem. If trained interviewers repeatedly give very different ratings for the same evidence, perhaps the competency is vague or the anchors are not discriminating between levels well.
Take a test before using the interview to decide hires
A question that appears clear in the document may produce ambiguous answers in practice. OPM recommends piloting the questions to see if they are understood and can generate a useful range of responses before using them in the actual process.
The pilot can check whether:
- the question actually measures the intended competency;
- candidates can understand the scenario without additional explanation;
- the predicted deepening is sufficient;
- two trained people can apply the anchors in a similar way;
- the interview fits into the time available;
- some question produces practically the same answer as another.
If a question fails, the best time to fix it is before there is a candidate whose approval depends on it.
“Cultural fit” should not become a note without definition
Broad criteria such as “fit”, “posture”, “energy” or “maturity” are especially vulnerable to different interpretations.
If there is something legitimate behind them, translate it into behavior.
“It fits with our culture” can hide criteria that are irrelevant to the role. “Can disagree with a decision while maintaining a productive working relationship” is a skill that can be described, asked about and scored.
The same reasoning applies to seniority. Instead of evaluating whether someone “seems senior,” describe the level of autonomy, complexity of problems, scope of decisions, and need for guidance expected in that role.
Structuring also means removing questions that shouldn't be there
In Brazil, the interview criteria must remain connected to professional activity. The Ministry of Labor and Employment advises that questions focus on training, professional history, experiences, skills, availability and other aspects related to potential performance in the role. The body also warns of questions about criteria that, as a rule, are discriminatory, including age, ethnicity, place of birth, sexual orientation, political opinion, religion, personal finances and existence or intention to have children, in addition to health or disability issues unrelated to the requirements of the position.
Law No. 9,029/1995, in its current wording, prohibits discriminatory and limiting practices for access to employment relationships for reasons such as sex, origin, race, color, marital status, family situation, disability, professional rehabilitation and age, among others.
This reinforces a useful design rule even outside the legal discussion: if it is not possible to explain the relationship between a question and the expected performance in the role, it should not gain space on the scorecard without an objective justification.
The scorecard starts working before the first interview
The biggest gain from the structured interview does not appear when the form is filled out. Appears when the team is forced to define in advance what they are hiring.
If “proactivity” cannot be translated into observable behaviors, perhaps the requirement is still poorly defined. If three managers want to evaluate different skills for the same vacancy, the misalignment needs to be resolved before the interviews. If no one can explain why a skill would be given greater weight, the weight is probably based on preference, not job analysis.
For the next wave, the shortest way is to take the really critical requirements and do the full chain:
role requirement → competency → observable evidence → question → deepening → anchors from 1 to 5 → independent grade → scorecard comparison.
This preparation reduces room for improvisation without turning the interview into a mechanical conversation. The candidate continues to be able to explain context, decisions and results. What changes is that everyone gets a comparable opportunity to demonstrate what the job requires — and interviewers need to justify their scores with something more solid than impressions.
