A Candfit assessment, start to finish
This is the report your team gets after every interview, shown for two roles: a customer support agent and a senior backend engineer. The candidates and companies are fictional. The sections, the ratings and the level of detail are exactly what you'd receive.
Summary
AI recommendation
Dana is a strong fit for a customer support role. They stayed calm with an angry caller, explained a billing problem in plain words, and showed they'd correct their own mistake on the spot. The main gap is ticketing: they said they've used Zendesk every day for two years but couldn't describe how they'd tag or escalate a ticket.
How this was scored
Eight questions were asked, and all eight test the role's core areas. Each answer was scored against what a strong answer for this role covers, and the eight scores average 81. All six core areas in the job description were assessed, so nothing is deducted. A strong fit at this level is a Pass.
Timeline
- AppliedOct 20, 2026 (Tue) at 8:05 AM
- Resume/CV screened, 76 out of 100Oct 20, 2026 (Tue) at 8:06 AM
- Invited to interviewOct 20, 2026 (Tue) at 9:32 AM
- Interview bookedOct 20, 2026 (Tue) at 6:47 PM
- Interview completed, 24 minOct 22, 2026 (Thu) at 7:26 AM
- Scored by AIOct 22, 2026 (Thu) at 7:28 AM
- Recruiter decisionWaiting
Recruiter decision
Candfit recommends. A person on your team decides.
In Candfit, one click records the decision and who made it.
Question by question
Eight questions were written from the job description, and all eight were asked. Each shows what the candidate said and why it got its rating.
-
1
Tell me about a time you turned an unhappy customer around.
Strong84 out of 100- Answer
- At a cable company's call center, a customer called for the third time about a missed installation. Dana booked the earliest slot themselves, added a note so the technician would call ahead, and phoned the customer the next day to confirm it went well.
- Assessment
- A specific example with clear actions and a follow-up the customer didn't ask for. Strong. Could have said what they changed afterward so it wouldn't happen again.
-
2
A customer says they were charged twice and wants the money back today. How do you handle the call?
Exceptional90 out of 100- Answer
- "I'd apologize first, then check both charges on their account so I know what actually happened, and tell them exactly what happens next." Asked what they'd say if the refund can't happen today: give the honest timeline, a reference number, and the day they'll call back to confirm it arrived.
- Assessment
- Checked the facts before promising anything, set clear expectations, and handled the follow-up without overpromising. Exactly what this role needs.
-
3
A caller has been on hold for 40 minutes and starts shouting as soon as you answer. What do you say first?
Strong86 out of 100- Answer
- Would let the caller finish, apologize for the wait without making excuses, and say what they'll do right now. "I don't argue about the wait. It was too long." Would keep their voice low and steady and get to the problem within the first minute.
- Assessment
- Calm, practical de-escalation that acknowledges the customer before trying to solve anything. Strong.
-
4
How would you explain overage charges on a data plan to a customer who isn't technical?
Strong80 out of 100- Answer
- Compared the plan to a monthly allowance: it includes a set amount of data, and each gigabyte over it costs extra. Would then show the customer where to see their usage in the app and offer to set up an alert.
- Assessment
- Clear and free of jargon, with a practical next step. Strong. Didn't check whether a different plan would suit the customer better.
-
5
Halfway through a call, you realize you gave the customer wrong information earlier. What do you do?
Strong88 out of 100- Answer
- Correct it on the spot: "I need to correct something I told you a minute ago." Then give the right answer, note it on the account, and tell a team lead in case other customers heard the same thing.
- Assessment
- Owned the mistake immediately and thought about other customers too. Strong, close to exceptional.
-
6
Walk me through how you'd log and escalate this call in a ticketing system like Zendesk.
Below expectations62 out of 100- Answer
- Said they've used Zendesk every day for about two years. Asked how they'd tag the ticket and decide who it goes to: "I'd just write a note, and my supervisor usually assigns it."
- Assessment
- The claimed experience didn't hold up under a basic follow-up. The job description lists ticketing as a requirement, so this is a real gap, and it's flagged as a red flag.
-
7
Fifteen chats are waiting in your queue, and your handle time is over target. How do you decide what to do?
Meets expectations74 out of 100- Answer
- Would work through the chats in order, keep answers short, and use saved replies for common questions.
- Assessment
- Sensible about speed, but didn't think about handling urgent problems first, like a customer who can't make calls at all. Meets expectations.
-
8
This role works rotating shifts, including weekends. What schedule works for you, and why do you want this job?
Strong84 out of 100- Answer
- Available for rotating shifts and every other weekend. Wants to grow into a team lead role within two years, and likes "fixing a problem in one call so nobody has to call back."
- Assessment
- Availability matches the role, and the motivation fits the job. Strong.
Strengths and concerns
Strengths
- Calm under pressure: lets an angry caller finish, apologizes for the wait without excuses, and gets to a fix within a minute.
- Sets honest expectations: checks both charges before promising anything, then gives a timeline and a reference number.
- Owns mistakes: corrects wrong information on the spot, and thinks about other customers who may have heard it.
Concerns
- Red flag
Said they've used Zendesk every day for two years, but couldn't describe how they'd tag, prioritize or escalate a ticket. - Works a busy queue strictly in order, without handling urgent problems first.
- Didn't ask whether a different plan would suit the customer with overage charges.
Transcript and recording
Dana WhitfieldTalk time
Candidate 64% · Interviewer 36%Files to download or share
- Assessment reportPDF
- TranscriptPDF and plain text
- RecordingVideo and audio
- EverythingOne .zip file
English fluency
This role talks with customers all day in English, so the candidate's speech was also scored for how easy it is to follow. Roles that don't need it skip this step.
These scores measure how clearly the candidate speaks: pace, pauses, pronunciation and rhythm. They never measure accent or where someone is from, and they never decide anything on their own.
Resume/CV screening
Summary
Strengths: three years in customer-facing roles, two of them taking calls and chats at a cable company, with a customer satisfaction award and open weekend availability.
Weaknesses and risks: three jobs in four years, and Zendesk is listed as a skill with no detail about how it was used.
Must-have coverage
| Requirement | Experience | Coverage | Evidence |
|---|---|---|---|
| A year or more in customer support | 3 years | 100% | Two years at a cable company's call center, one in retail service |
| Phone and chat support | 2 years | 90% | Inbound calls and live chat for billing and service questions |
| Billing and account questions | 2 years | 80% | Handled billing disputes and refund requests |
| Ticketing system (Zendesk or similar) | 2 years, listed | 50% | Named in the skills list, with no detail |
| Rotating shifts, including weekends | Stated | 100% | "Available evenings and weekends" |
Risks
- Tenure Medium
Three jobs in four years, each one under 18 months.
Mitigation: ask what made them leave, and what would make them stay.
- Skills gap Medium
Zendesk is listed as two years of experience, but no role describes how it was used.
Mitigation: probe ticketing in the interview.
Summary
AI recommendation
Jordan is a strong fit for a senior backend role. They showed deep, practical knowledge of payment reliability, data design and service communication, explained trade-offs clearly, and handled every follow-up well. The main gap is deployment: they said they've used Kubernetes for four years but couldn't explain how a rolling update works.
How this was scored
Nine questions were asked, and all nine test the role's core areas. Each answer was scored against what a strong senior answer covers, and the nine scores average 84. Six of the seven core areas in the job description were assessed. Growth and learning wasn't, and with only one area missed, nothing is deducted. A strong fit at this level is a Pass.
Timeline
- AppliedOct 19, 2026 (Mon) at 10:12 AM
- Resume/CV screened, 81 out of 100Oct 19, 2026 (Mon) at 10:13 AM
- Invited to interviewOct 19, 2026 (Mon) at 11:40 AM
- Interview bookedOct 20, 2026 (Tue) at 8:05 AM
- Interview completed, 42 minOct 21, 2026 (Wed) at 7:43 PM
- Scored by AIOct 21, 2026 (Wed) at 7:45 PM
- Recruiter decisionWaiting
Recruiter decision
Candfit recommends. A person on your team decides.
In Candfit, one click records the decision and who made it.
Question by question
Ten questions were written from the job description. Nine were asked. Each shows what the candidate said and why it got its rating.
-
1
Walk me through a backend system you owned end to end. What would you change if you built it again?
Strong88 out of 100- Answer
- Described the ledger service they built in Go at a payments startup: double-entry records in PostgreSQL, Kafka for settlement events, about two million transactions a day. Would split reporting reads from the main database earlier: "We added a read replica a year too late."
- Assessment
- Clear ownership from design to on-call, with real numbers and a specific lesson. Strong for a senior role. Could have said more about how the team weighed the original trade-offs.
-
2
When would you choose a message queue over a synchronous API call between two services?
Exceptional90 out of 100- Answer
- A queue when the caller doesn't need the result right away, when the receiver can be slow or down, or to absorb bursts. A synchronous call when a person is waiting on the answer. Gave the example of moving receipt emails onto a queue after a mail-provider outage blocked checkouts.
- Assessment
- Compared the options by failure mode, latency and ownership, and backed it with a real incident. Exactly the judgment this role needs.
-
3
How do you change a database schema on a live system without downtime?
Strong86 out of 100- Answer
- Expand, then contract: add the new column, write to both, backfill in batches, switch reads over, and drop the old column in a later release. Mentioned being careful with locks on large tables.
- Assessment
- Correct, well-ordered approach with attention to locking. Didn't say how they'd verify the backfill before switching reads.
-
4
How would you make a payment webhook handler safe to retry?
Exceptional92 out of 100- Answer
- "I'd store the event ID with a unique constraint and make the handler idempotent, so a retry returns the saved result instead of charging the customer twice." Asked what happens when two retries arrive at the same moment: the first insert wins, and the second hits the constraint and returns the stored response.
- Assessment
- Precise and complete, including the race condition the follow-up probed. This is the most important skill for the role, and the answer was senior-level.
-
5
A query on the orders table got slow as the table grew. How would you find and fix the cause?
Meets expectations74 out of 100- Answer
- Would add a composite index on customer ID and created date, and check whether the query selects more columns than it needs.
- Assessment
- A reasonable fix, but went straight to an index without reading the query plan to confirm the cause. Meets expectations for a senior role, no more.
-
6
Your service's p99 latency doubles right after a deploy. Walk me through what you do.
Strong86 out of 100- Answer
- Confirm the timing lines up with the deploy, roll back first if customers are affected, then compare traces from before and after. Described a real incident where a new ORM version ran one query per item in a loop, and how they wrote the postmortem and added a latency check to the release pipeline.
- Assessment
- A calm, ordered response that puts customers first, plus a real example with a lasting fix. Strong.
-
7
Walk me through how you've run services on Kubernetes.
Below expectations56 out of 100- Answer
- Said they've used Kubernetes for about four years, mostly on EKS. Asked how a rolling update works and how to roll one back: "Honestly, the platform team handled deployments. I'm not sure of the details."
- Assessment
- The claimed experience didn't hold up under a basic follow-up. The role lists Kubernetes as a requirement, so this is a real gap, and it's flagged as a red flag.
-
8
How would you set up autoscaling for a service with spiky traffic?
Not askedThe candidate had just shown they don't know Kubernetes deployments well, so the interviewer moved on to a different topic instead of drilling deeper into the same one.
-
9
How do you test code that depends on a third-party payment provider?
Exceptional92 out of 100- Answer
- Unit tests against a small fake of the provider's client, contract tests against the provider's sandbox in CI, and replaying recorded webhook payloads, including malformed ones and duplicates.
- Assessment
- A layered, practical strategy that covers the failure cases that matter in payments. Exceptional.
-
10
Tell me about a time you disagreed with a product decision. What did you do?
Exceptional92 out of 100- Answer
- Product wanted instant payouts on day one. Jordan showed the fraud-loss risk with data from a similar launch, proposed a limit for new accounts, and the team shipped that version first.
- Assessment
- Disagreed with evidence, offered a workable alternative, and kept the working relationship intact. Clear, specific and senior.
Strengths and concerns
Strengths
- Designs for failure: explained idempotent webhook handling, including what happens when two retries race.
- Weighs trade-offs out loud: compared queues and synchronous calls by failure mode, latency and team ownership.
- Owns incidents end to end: rolled back first, found the cause in traces, wrote the postmortem and added a release check.
Concerns
- Red flag
Said they've used Kubernetes for about four years, but couldn't explain how a rolling update works or how to roll one back. - Went straight to an index for the slow query without checking the query plan.
- Didn't bring up monitoring or alerting until asked about the latency incident.
Transcript and recording
Jordan RiveraTalk time
Candidate 61% · Interviewer 39%Files to download or share
- Assessment reportPDF
- TranscriptPDF and plain text
- RecordingVideo and audio
- EverythingOne .zip file
English fluency
This role requires spoken English, so the candidate's speech was also scored for how easy it is to follow. Roles that don't need it skip this step.
These scores measure how clearly the candidate speaks: pace, pauses, pronunciation and rhythm. They never measure accent or where someone is from, and they never decide anything on their own.
Resume/CV screening
Summary
Strengths: seven years building backend services in Go, most recently in payments, with clear ownership of a ledger at scale and hands-on PostgreSQL and Kafka work.
Weaknesses and risks: Kubernetes is listed as four years of experience, but no project describes it, and there's little on monitoring or on-call.
Must-have coverage
| Requirement | Experience | Coverage | Evidence |
|---|---|---|---|
| Go or Java services in production | 7 years | 100% | Built payment and ledger services in Go at two companies |
| PostgreSQL design and tuning | 6 years | 90% | Led a schema redesign of the orders database |
| Payment systems and webhooks | 4 years | 90% | Integrated card and bank payment providers |
| Message queues (Kafka or RabbitMQ) | 3 years | 80% | Kafka consumers for settlement events |
| Kubernetes in production | 4 years, listed | 50% | Named in the skills list, with no project detail |
Risks
- Technical gap Medium
Kubernetes is listed as four years of experience, but no project describes it.
Mitigation: probe deployment ownership in the interview.
- Experience gap Low
Most experience is at companies under 200 people, and this role supports a larger platform team.
Mitigation: ask about working across teams.
What each part tells you
Where they land against your bar
The interview score sits on the same 0 to 100 scale as the bar you set for the role, so you can see at a glance how far above or below it a candidate lands.
The evidence behind every rating
Each question shows what the candidate said and why it got its rating. Disagree with one? Play that part of the recording and decide for yourself.
Concerns named, not buried
Gaps and contradictions get their own list, so the thing you'd most want to know is never in the small print.
Scored against your role, at the right level
Every candidate for a role is scored the same way, so you can compare them fairly.
- Exceptional90 to 100Direct, specific and deep, with no gaps
- Strong80 to 89Substantive, with examples, at or above the level
- Meets expectations70 to 79Solid on the core, light on depth
- Below expectations50 to 69Knows the topic, with notable gaps
- Weak30 to 49Vague or incomplete
- Poor0 to 29Major gaps, or contradictions
The overall score
It's the average of the scores for every question that tests the role's core areas. If fewer than five were asked, the total is still divided by five, so a short interview can't score high. If two or more of the role's core areas weren't covered, 10 points come off.
Level matters
The same answer scores higher for a junior role than for a senior one. Clear communication counts for more as seniority rises. Accent and personality never count.
From score to recommendation
The recommendation is the overall judgment of fit: a strong fit is a Pass, a moderate fit is Borderline, and a weak fit is a Fail. A person on your team confirms or overrides every one.
See this report for one of your candidates.
Sign up, add one role and invite one candidate. Candfit runs the interview and sends you the full assessment, free.