Best Practices for Conducting Technical Interviews Remotely

Remote technical interviews fail in three predictable ways: identity isn't confirmed, the network drops mid-session, or scoring drifts to "vibes" because nobody wrote a rubric. All three are mechanical problems with mechanical fixes. Done correctly, remote technical interviews are at least as defensible as in-person ones, and far more scaleable.

Confirm identity at the start

The single most common concern with remote interviews is "is the person on camera actually the candidate?" The concern is real but the fix is fast. A photo ID check on camera at session start ends most cheating concerns. The candidate holds up government-issued ID next to their face. The interviewer or platform records the moment.

AssessExpert handles this automatically with facial recognition matched to the registered candidate's profile photo. The platform refuses to start the session if the face does not match. For interviewer-led sessions without a platform, a 30-second ID check at the start covers most identity risk.

The objection — "this feels invasive" — is usually softer than feared. Candidates understand identity verification is necessary for the data to mean anything. The intrusion concern applies to surveillance during the session, not identity verification at the start.

Run a pre-flight check

About 20% of remote interview sessions encounter a technical issue if the candidate has not been pre-flighted. Camera not working. Microphone routed to the wrong device. Bandwidth insufficient for video. The session then fails at minute 2, wastes everyone's time, and produces no data.

The fix is a five-minute pre-flight check 24-48 hours before the session. The platform tests camera, microphone, screen-share, and bandwidth, and walks the candidate through fixes for common problems. By the time the session starts, the technical risk is minimised.

For platform-led sessions, the pre-flight is automated. For interviewer-led sessions, send the candidate a one-page setup checklist with a screenshot for each step. Both work; both prevent the same failures.

Record the session

Recording the session is not surveillance theatre. It is operational risk management. A junior interviewer who is unsure about a candidate can hand the recording to a senior for second opinion. A disputed hire decision can be reviewed against the actual session. Without a recording, every decision is one person's memory.

Consent first. Tell the candidate the session is recorded, what the recording is used for, who can access it, and how long it is retained. Most candidates accept this; some have strong opinions. Either way, the consent moment is non-negotiable for legal reasons in most jurisdictions.

Storage matters too. Recordings should be retained for the legal minimum required by your jurisdiction (often 90 days) and deleted on schedule. Indefinite retention creates legal risk without operational benefit.

Score against the rubric in real time

The single most common scoring failure in remote interviews is post-session scoring from memory. Interviewers remember the last 10 minutes. They remember the candidate's energy more than their answers. They confuse charisma with competence. By the time they sit down to write the score, the data is degraded.

The fix is real-time scoring against a written rubric. Open the rubric in a second window. As the candidate hits a rubric item, tick it. As they miss one, note it. At the end, the score is the sum of ticks and notes, not a memory exercise. Recency and confirmation bias drop sharply.

For panel interviews, every interviewer scores independently before discussion. Pooling scores before discussion eliminates anchoring on whoever speaks first. Disagreements then trigger productive conversation rather than groupthink.

Close with the candidate

Remote candidates feel ghosted faster than in-person ones. They do not have the visual cue of leaving an office. The interview ends, the screen goes blank, and the silence afterwards feels longer than it would in person.

State the next step and the timeline at the close of every session. "You will hear from us within 5 business days with the outcome. If you don't hear by then, please ping me." Specific, time-bound, with a contact channel for follow-up. The candidate leaves the session knowing what to expect, which improves their experience and reduces follow-up emails from anxious candidates.

For declined candidates, send the decline within the promised window with one line of reason. "Strong communication, technical depth did not match the seniority of this role. We'd be happy to consider you for a more junior role if interested." Candidates respect this; they resent silence.

What to do when the network drops mid-session

It will happen occasionally even with good pre-flight. The plan should be agreed in advance.

For short drops (under 60 seconds), resume where you left off. Note the drop on the score sheet so it can be considered if the candidate later disputes.

For long drops (over 5 minutes), reschedule. Continuing after a long drop produces noisy data — the candidate is rattled, the interviewer is annoyed, the session is no longer comparable to other candidates.

For repeated drops in the same session, reschedule and offer the candidate a quieter time slot. Some candidates have weak home networks; the fair response is accommodation, not penalty.

How to keep the candidate calm

Remote candidates run hotter than in-person ones. The asymmetry is real — they cannot read your body language, they cannot tell if they are doing well, the silence between questions feels longer. Three mechanical fixes help.

First, open with two minutes of low-stakes introduction. "I'm X, I work on Y, here is how this session will run." The candidate settles into the rhythm before the technical questions start.

Second, use audible acknowledgement. "Mm-hm," "yes," "interesting." Without these cues, the candidate cannot tell if you are listening. The silence reads as judgement.

Third, telegraph the section transitions. "OK, that was the system design section. Next we'll do a short coding exercise — about 20 minutes." The candidate knows where they are in the session, which reduces anxiety.

What to test in a remote technical interview

The same things you would test in person, with one adjustment. Hands-on tasks should be designed for the remote constraints. Whiteboard problems become collaborative document problems. Pair programming becomes shared editor sessions. The point is to keep the work product visible to the interviewer so the scoring is grounded in observation, not self-report.

For roles where the work is heavily tool-dependent — CAD designers, BIM coordinators — consider a managed-session format where the platform delivers the task in a sandbox the interviewer can observe. AssessExpert handles this for engineering roles with screen recording and proctor observation.

Anti-cheat in remote sessions

Active proctoring during the session — face detection, screen monitoring, audio anomaly detection — catches most cheating attempts. The candidate paste-tabbing to ChatGPT generates distinctive signal. Their face leaving the camera generates a flag. Multiple voices on audio generate a flag.

None of these auto-disqualify. They go into the integrity note for the proctor or interviewer to review. Most flags are innocent (the candidate looked away to check their notes); some are not. The human review step decides which is which.

How AssessExpert handles remote sessions

For managed remote technical assessments — where AssessExpert proctors run the session and produce the scored report — see Technical Interview Assessment. For self-service remote assessments where your team runs the session in your workspace, see Online Assessment Platform. Both include identity verification, pre-flight checks, recording, and proctor sign-off.

FAQ

How do you stop the candidate from getting outside help during the session?

Active proctoring — face detection, screen monitoring, audio anomaly detection — with human review. No single layer is sufficient; together they are robust.

What if the candidate refuses to enable camera?

For a proctored assessment, the camera is non-negotiable. Reschedule and explain why the camera is required. If they continue to refuse, the candidate experience signal is informative — most willing candidates accept this.

Should we use video for the whole interview or only for the technical sections?

Whole interview. Switching off and on creates friction and inconsistent identity signal.

How long should a remote technical interview be?

60-90 minutes. Longer adds noise from fatigue; shorter is too thin to be predictive.

What about candidates in different time zones?

Schedule for the candidate's working hours, not yours. A candidate at 11pm their time will underperform regardless of skill.

Do recordings create legal risk?

Less than not having them. Recordings provide evidence in disputed decisions. Consent at session start plus a retention policy plus access controls minimise the risk.

Next steps

If you are running remote technical interviews and want to standardise the mechanics, book a demo. We will walk through your current process and identify the highest-leverage improvements.