TL;DR
An integrity score summarizes what happened during an online hiring test, but it describes the session, not the person. Learn what each band and signal means, how to review a flagged candidate step by step, and why a low score should start a conversation.
An integrity score is a summary of what an online test observed while a candidate took it, such as tab switches, pasted text and unusual timing. It is usually shown as a number with a confidence band. It is evidence for a person to review, not a verdict on honesty, and a low score should start a conversation rather than end a candidacy.
Small businesses are adding proctored assessments because AI cheating is now a real concern on remote tests. The risk is reading an integrity score like a lie detector. This guide explains what each signal means and how the three common bands work. It also covers how to review a candidate in the middle band step by step, what to say in a follow-up conversation, and how to keep the whole process fair and consistent.
What is an integrity score in hiring?
An integrity score in hiring is a summary measure, often on a 0 to 100 scale, of the conditions under which a candidate completed an online assessment. It condenses many small observations into one figure and a band, so a busy reviewer knows where to look first. The score describes the testing session, not the person.
The score answers a narrow question: did anything happen during this session that a reasonable reviewer would want to look at? It cannot say why something happened. A tab switch is recorded the same way whether the candidate opened a chatbot or closed a pop-up from their bank.
Integrity score vs. integrity test
The phrase causes confusion because integrity tests are a separate, older category of pre-employment assessment. An integrity test is a questionnaire, often personality-based, that asks about attitudes toward rules, theft or reliability, and it produces a score about the candidate. An integrity score from a proctored assessment says nothing about character. It reports on test conditions. If a vendor or colleague uses the term, confirm which one they mean before you act on it.
How a proctored assessment collects integrity signals
A proctored assessment is any test where the candidate's session is monitored in some way, by a person, by software or by both. In remote hiring, monitoring is usually automated and records signals while the candidate works. Those signals fall into three groups.
Signals recorded during every session
- Tab or window switches: the candidate left the test page, and the record shows for how long.
- Paste events: text was pasted into an answer field, which matters most on written or coding questions.
- Timing: how long each question took, compared with the time allowed and the difficulty of the question. Very fast correct answers on hard items, or long pauses followed by polished answers, stand out.
Each signal is weak on its own. Patterns across several signals carry more weight, especially when they line up with the questions a candidate answered correctly.
Optional camera and identity checks
Some employers add camera-based checks for higher-stakes roles: periodic webcam snapshots, an identity photo, a live face check, a second camera on the candidate's phone that shows the desk, or a companion desktop app. These checks add context, such as a second person in the room or a different person sitting the test. Because they collect images, candidates should be told about each check before they start and asked to consent. Use camera checks when the role justifies them, not by default.
Prevention does most of the work
Good online test integrity starts before any signal is recorded. Tests that give each candidate a different set of questions, serve them one at a time from the server, time each one and block copying make answer-sharing and quick lookups much harder. Strong prevention means you interpret evidence less often. Our guide on how to reduce AI assistance in online assessments covers those design choices in more detail.
The three integrity score bands, explained
Most systems translate the number into bands, because a band is easier to act on than a raw score. HireValid's Integrity Score, for example, runs from 0 to 100 and uses three bands: High confidence, Review suggested and Serious concerns. Each result comes with an evidence timeline that shows what was observed and when. The same reading logic applies to any tool that groups results this way.
| Band | What it usually means | Sensible next step |
|---|---|---|
| High confidence | Few or no notable signals; the session looks consistent with independent work | Evaluate the skill results as normal |
| Review suggested | Some signals worth a look, often with ordinary explanations | Read the evidence timeline, then decide whether to ask a question |
| Serious concerns | Several strong or clustered signals, or camera evidence that needs attention | Review carefully, talk with the candidate and consider a verification step |
High confidence
A High confidence band means the monitoring saw little worth flagging. It does not guarantee the work was independent, because no monitoring catches everything. Treat the band as permission to focus on the skill scores, not as a seal of approval.
Review suggested
Review suggested is the band that needs the most judgment, and many strong candidates will land there. A few tab switches, one paste into a written answer or one unusually quick section can move a session into this band. Many of those candidates did nothing wrong.
Serious concerns
Serious concerns means the evidence deserves careful attention. Examples include a long absence from the test page followed by a run of correct answers, or a snapshot showing someone else at the keyboard. Even here, the band is a reason to look closer, not a conclusion. The candidate still deserves a chance to explain.
Why a low integrity score is not proof of wrongdoing
Integrity signals record behavior, not intent. The same event can have honest and dishonest causes, and software cannot tell them apart. Common innocent explanations include:
- A notification, a password manager or a dropped Wi-Fi connection pulling focus away from the test tab.
- Pasting a draft from a notes app because the candidate always writes that way.
- A shared family laptop with browser extensions the candidate never installed.
- Fast answers from someone who genuinely knows the material, or quick guesses under time pressure.
- Assistive technology, such as screen readers or dictation software, that interacts with the browser in unusual ways.
The last point matters legally as well as ethically. A candidate who received extra time or another accommodation may show timing patterns that look odd against the standard test. Read those signals against the candidate's adjusted conditions. Our guide to pre-employment test accommodations explains how to set accommodations up.
The practical rule is simple: never reject a candidate on an integrity score, a band or an automated flag alone. A person should look at the evidence, consider the context and, where doubts remain, ask.
How to review a Review suggested candidate
A structured review keeps decisions consistent and quick. For each candidate in the middle band, work through these steps.
- Look at the skill results first. Note the overall score and the sub-skill scores. If the candidate would not advance on skills, the integrity question may not matter for this role.
- Open the evidence timeline. See which signals occurred, when and how often. One brief tab switch in a 40-minute test reads differently from ten switches during the hardest section.
- Match signals to answers. Check whether the flagged moments line up with correct answers on difficult questions. Signals on items the candidate got wrong, or on easy items, carry less weight.
- Check the context. Ask whether the candidate had extra time, took the test on a phone, or triggered a signal on the instructions screen rather than during a question. Any of these can explain a signal.
- Apply one standard to everyone. Review every candidate with similar signals the same way, regardless of who they are. Write the standard down once and reuse it.
- Choose one of three actions. Proceed as normal, ask a short clarifying question, or add a verification step such as a brief live exercise.
- Record the decision and the reason. A two-line note protects you and the candidate if the decision is ever questioned.
A worked example
Say a dental practice is hiring an office administrator and uses a short assessment with an attention to detail test and a written email task. One applicant scores well and lands in Review suggested. The timeline shows two tab switches of a few seconds each during the multiple-choice section and one paste into the email answer.
The tab switches fall on questions the candidate answered incorrectly, which weakens any concern about them. The paste is the more relevant signal, because the written task is where outside help would matter most. A fair next step is a short interview. Ask the candidate to explain how they approached the email, then give a similar five-minute writing prompt live. If they handle the prompt well, the paste was likely a habit. If they struggle sharply, you now have real evidence to weigh, gathered fairly.
What to say when you follow up
The follow-up conversation is where you learn the most, and tone decides whether it helps. Start from curiosity, not accusation. Candidates who feel accused tend to get defensive, and good candidates may simply withdraw.
Questions that tend to work:
- Can you tell me about your setup when you took the assessment?
- The report shows the test page lost focus a few times. Was anything interrupting you?
- Can you walk me through how you approached the written question?
- Would you be comfortable doing a short exercise like this one with me now?
Avoid saying the system caught them or that the score proves anything. Describe what was observed, ask for context and listen. Then weigh the answer together with the skill results and the live exercise. If AI tools during live interviews also worry you, our article on what employers can do about AI interview copilots covers realistic options.
Common mistakes when reading test integrity evidence
- Setting a hard cutoff. Rejecting everyone below a number turns weak signals into decisions and removes the judgment the score exists to support.
- Reviewing only the candidates you already doubt. Uneven scrutiny is unfair and can create legal risk.
- Ignoring the skill results. A flagged session with modest scores rarely deserves a long review for a role that needs strong scores anyway.
- Treating the band as the whole story. The timeline holds the detail; the band only tells you where to start.
- Hiding the monitoring. Candidates should know before the test what is monitored and why.
For a closer look at how specific AI cheating patterns show up in test data, see our guide to detecting AI cheating in online hiring tests.
Keeping integrity review fair and lawful
Integrity review is part of your selection process, so the same fairness principles apply. In the United States, Title VII of the Civil Rights Act of 1964 prohibits employment discrimination based on race, color, religion, sex and national origin. The Americans with Disabilities Act (ADA) requires reasonable accommodations for qualified applicants with disabilities, including during testing. The U.S. Equal Employment Opportunity Commission (EEOC) enforces both laws and publishes guidance on employment tests and selection procedures.
Camera and identity checks raise separate questions. Some states regulate the collection of biometric data, with the Illinois Biometric Information Privacy Act the best-known example, and the rules vary from state to state. Tell candidates what you collect, get their consent and keep images only as long as you need them.
Finally, watch your outcomes. If integrity reviews lead to more candidates from one group dropping out, that pattern deserves attention. Our explainer on the four-fifths rule for small employers shows a simple way to monitor selection rates. Because laws differ by state and situation, talk with a qualified employment attorney or HR advisor about your specific process.
Frequently asked questions
Can a candidate be rejected because of a low integrity score?
A candidate should not be rejected on an integrity score alone. The score summarizes signals that have both innocent and concerning explanations. Use it to decide where to look and whether to ask a question or add a verification step. The hiring decision should rest on a person's review of the evidence, the skill results and any follow-up, applied consistently to every candidate.
Does switching tabs during an online test mean cheating?
No. A tab switch only shows that the test page lost focus. Notifications, pop-ups, connection drops and accidental clicks all cause tab switches. They matter more when they are long, frequent and line up with correct answers on hard questions. Even then, a tab switch is a reason to ask the candidate about it, not a conclusion.
What is a good integrity score?
A good integrity score is one in the top band, meaning the monitoring saw few or no notable signals. There is no universal number, because each tool weights signals differently. Read the band and the evidence behind it rather than chasing a precise figure, and remember that a strong score does not guarantee the work was independent.
Can an integrity score detect AI cheating?
An integrity score can surface behavior that often accompanies AI cheating, such as leaving the test page, pasting text or answering hard questions unusually fast. It cannot confirm that a tool was used, and some assistance leaves few traces. That is why test design, such as unique timed questions, matters as much as monitoring, and why follow-up conversations stay important.
Should I tell candidates their test is proctored?
Yes. Tell candidates before the test what is monitored, such as tab switches, pastes, timing or camera snapshots, and ask for consent to any camera or identity checks. Transparency is fairer, reduces anxiety and discourages misuse. Some states regulate biometric data, so confirm the rules where you hire with a qualified advisor.
Should I share integrity evidence with the candidate?
Describe what was observed in plain terms during a follow-up, for example that the test page lost focus several times during one section. You do not need to share the full report. The goal is to give the candidate a fair chance to explain, not to present a case against them.
Conclusion: read the evidence, then talk to the person
An integrity score is a map of where to look, not a judgment of who someone is. Bands sort sessions by how much attention they need, signals show what happened and when, and your review supplies the context software cannot. The middle band is where fair process matters most: match signals to answers, account for accommodations, ask open questions and verify skills live when doubt remains.
HireValid builds this approach into its assessments. Each candidate gets a different, timed test, signals appear in an Integrity Score with an evidence timeline, and nothing automated rejects anyone. Try HireValid free with 3 candidates, or see the plans and pricing to find the right fit for your hiring.
Does a low Integrity Score reject a candidate?+
No. Integrity signals may have innocent explanations, including connection issues or accessibility needs. A person should review the evidence and speak with the candidate before deciding.
What will candidates need?+
A browser and a reliable connection. Typing, spreadsheets and code tasks are intended for desktop. If an employer enables camera checks, candidates must receive a clear notice and a route to request an alternative.
Are the scores official qualifications?+
No. These are tools for hiring decisions. English results are CEFR-aligned level estimates, not official certificates. Cognitive scores are not clinical IQ results.