A remote hiring case study with 32 tutors in 3 weeks
In this remote hiring case study, a Series A EdTech platform in Boston moved tutor interviews to The Cognitive's live AI video interview before back-to-school. Customer-reported results: 32 tutors hired across 4 time zones in 3 weeks, against 6 a month before, with time to first interview down from 4 days to 45 minutes.
What changed for the EdTech platform's tutor hiring?
The first 3 rows are the platform's reported figures. The last 2 are our arithmetic on them.
| Metric | Before | After | Change |
|---|---|---|---|
| Tutors hired (customer-reported) | 6 a month, target 30 | 32 in 3 weeks | Target beaten by 2 |
| Time to first interview (customer-reported) | 4 days | 45 minutes | About 95 hours sooner |
| Candidate quality score (customer-reported) | 5.2 out of 10 | 8.4 out of 10 | Up 3.2 points |
| Months to hire 32 tutors (our arithmetic) | About 5.3 at 6 a month | Under 1 | Cohort ready before term |
| Quality score change (our arithmetic) | 5.2 | 8.4 | About 62% higher |
Figures reported by the platform in our case study record dated 31 March 2026.
Why could the platform only hire 6 tutors a month?
The customer is a Series A EdTech platform in Boston, named in our record as EdTech Platform. It needed to grow its tutor network before back-to-school season, hiring remote tutors in the US, Canada, the UK and Australia. The target was 30 tutors a month. The team was managing 6.
Two things held it back. Scheduling across 4 time zones meant 4 days to arrange a first interview, and candidates took offers from faster competitors in the meantime. Quality was uneven too, because each hiring manager judged communication skills in their own way.
An illustration of how 4 days builds up: a tutor in Perth applies on a Monday. Boston sends 3 possible times, the tutor replies overnight, one slot clashes with a student session, and a new round of emails starts. By Thursday the tutor has taken other work. Each lost candidate sends the team back to the applicant list, which is how a team lands at 6 hires a month.
How does 3 weeks compare with typical hiring times?
Published benchmarks cover all roles, not tutors alone, but they show the pace most teams accept. Sources read 5 October 2026.
| Measure | Typical | This platform | What the gap means |
|---|---|---|---|
| Time to fill | Median 39 days for a nonexecutive role (SHRM 2026) | 32 tutors in 3 weeks (21 days) | A whole cohort in about half the median time for one role |
| Application to offer | About 38 days for business roles (Ashby 2026) | Every hire inside 21 days | At least 17 days faster than the typical business hire |
| Interview hours per hire | 8.9 hours for customer support, 12.2 for business roles (Ashby 2026) | First rounds run by the AI interviewer | Manager hours moved from talking to reading reports |
| Applicants past the first screen | 8% of applicants advance past initial screening (Gem 2026) | Tutors judged in a 20 minute interview | Teaching ability judged on what tutors say, not resume keywords |
Benchmarks read 5 October 2026. Platform figures as reported in our March 2026 record.
We teach students that speed matters. Our old hiring process was the opposite. Head of Talent, EdTech Platform, Boston
Why was back-to-school a deadline, not a target?
Tutors hired after term starts are worth a fraction of tutors hired before it. Students sign up in a rush, and every week without enough tutors is a week of demand the platform cannot serve. At 6 hires a month, the cohort would have arrived long after it was needed.
The team first pushed harder on the process it had: more calendar chasing, faster resume review, and hiring managers squeezing tutor screens between student-facing work. That only made the bottleneck more visible. Interviews still took 4 days to arrange, and each manager's idea of good communication stayed slightly different.
How does a tutor in Sydney interview with a team in Boston?
A live, two-way AI video interview: camera and microphone check, then the AI asks and follows up.
Nobody in Boston has to be awake. The tutor opens the invitation, sees start times in their own local time and books one, with no account to create. The AI runs the interview at that hour and the report is ready for the team the next time it logs in. Geography stops being a scheduling problem.
What questions show whether someone can actually teach?
Examples in the spirit of the record's rubric: subject knowledge, explanation quality, patience and engagement. They are not the platform's own questions.
Explain why dividing by a fraction is the same as multiplying by its reciprocal, as you would to a middle school student.
Starts from something the student already knows, uses a concrete example, and checks understanding before moving on.
The student still looks lost. What do you try next?
A student keeps getting the same kind of problem wrong in a session. Walk me through what you do.
Diagnoses the misconception instead of repeating the explanation louder, and keeps the student's confidence intact.
How would you know the misconception is actually fixed?
Your student has their camera off and gives one-word answers. How do you run the next 20 minutes?
Asks questions that need more than yes or no, gives the student control of something, and adjusts pace.
What would you tell the parent afterwards?
A Year 11 student asks why the quadratic formula works, not just how to use it. Talk me through your answer.
Knows the derivation from completing the square and can pitch it at the student's level without skipping steps.
Which step do students usually get stuck on, and why?
Your video freezes halfway through a worked example. How do you keep the session on track?
Has a fallback such as a shared whiteboard or the chat, recaps where the student was, and does not lose the thread.
How do you check the student followed the part they missed?
How did the platform hire its cohort, step by step?
Today's workflow in The Cognitive. The platform's setup dates from March 2026, and the product has changed since.
- Agree one tutor rubric The record lists subject knowledge, patience, explanation quality and engagement. Write each as a weighted criterion that describes what good teaching sounds like. Our suggested weights: explanation quality 35, subject knowledge 30, patience 20, engagement 15. Explanation leads because it is the skill students pay for.
- Set the interview to 20 minutes Long enough for the AI to ask for an explanation and then follow up on it, which is where knowing a subject and teaching it come apart. 10 minutes fits a quick availability and communication check. 20 minutes fits an explanation, a follow-up and a struggling student scenario.
- Move every applicant in Import from your ATS through one of 60+ ATS integrations, or upload a CSV of applicants. Then bulk invite them all in one go. Imported candidates get a note with score, summary and report link written back to the ATS. Stages never move on their own.
- Let tutors book their own slot Open booking dates wide so tutors in every time zone can find an hour that suits them. The record says candidates took to self-booking faster than the team expected. Slots show in the tutor's own local time, so Sydney and Toronto see the same open dates without anyone converting time zones.
- Calibrate on the first batch Hiring managers at first challenged scores where polished resumes did poorly. The team compared examples against the rubric before relying on the reports. A practical method: 3 managers read the same 5 reports and recordings, agree what a 3 and a 5 sound like for each criterion, then split the queue.
- Read reports and decide A 1 to 5 score per criterion, overall written feedback, a weighted score out of 100, transcript and recording. Hiring managers chose who joined; nothing was rejected automatically. A simple review rule: read every report with a 4 or 5 on explanation quality first, and watch the recording when two criteria disagree.
- Top up a thin subject If applicants dry up for one subject, search ~900M public profiles from a plain-English brief, reveal verified emails and phone numbers, and invite the people you shortlist. A brief such as: tutor with a physics degree and online teaching experience, based in the UK or Australia.
Can you add your own tutor questions?
Adding questions to a role: type them in or generate them with AI, each with an expected answer.
Yes. Add the teaching scenarios that matter to your platform. The AI asks them in its own order, adds questions of its own and follows up when an explanation is thin. The rubric stays fixed for every tutor, wherever they live, so a candidate in Sydney and one in Toronto are scored on the same terms.
The arithmetic behind the cohort
Inputs are the platform's reported figures plus one SHRM benchmark. Everything after an equals sign is our arithmetic.
HIRES Before: 6 a month against a target of 30. After: 32 in 3 weeks. At the old rate, 32 / 6 = about 5.3 months. 32 minus 30 = 2 more than the monthly target. WEEKLY PACE Before: 6 a month / about 4.3 weeks = about 1.4 a week. After: 32 / 3 weeks = about 10.7 a week. 10.7 / 1.4 = about 7.6 times the old weekly pace. TIME TO FIRST INTERVIEW Before: 4 days x 24 = 96 hours. After: 45 minutes. 96 hours minus 45 minutes = about 95 hours sooner. Across 32 hires: 32 x 95 = about 3,040 candidate hours no longer spent waiting. AGAINST THE BENCHMARK SHRM 2026 median time to fill: 39 days for one role. This cohort: 21 days for 32 roles. 21 / 39 = about 54% of the median, for the whole cohort. CANDIDATE QUALITY SCORE (the platform's own 10-point scale) Before 5.2. After 8.4. 8.4 minus 5.2 = 3.2 points. 3.2 / 5.2 = about 62% higher. STUDENT SATISFACTION New tutors rated 22% higher than the previous cohort.
What changed, number by number?
Hires: 6 a month to 32 in 3 weeks. The Head of Talent put it simply: "We hired our entire back-to-school cohort in three weeks." At the old pace, the same cohort would have taken about 5 months and arrived well after term started.
Time to first interview: 4 days to 45 minutes, as the platform reported it. Nobody coordinated calendars across 4 time zones, because each tutor picked a slot and interviewed. Today's booking page starts slots at least 2 hours ahead, so a new setup should expect hours rather than minutes.
Quality: the platform's own candidate quality score rose from 5.2 to 8.4 out of 10, and students rated the new tutors 22% higher than the previous cohort. Quality rising while volume rose is the result worth noting.
What did a hiring manager's week look like before and after?
Built from the platform's reported numbers and the workflow it described. The before column is the 6 hires a month process.
| Part of the week | Before | After |
|---|---|---|
| First contact with a tutor | Emails proposing times across 4 time zones | One bulk invite; tutors book their own slot |
| Wait for a first interview | 4 days, or 96 hours | 45 minutes, as reported |
| First-round interviews | Squeezed between student-facing work | Run by the AI at the hour each tutor picked |
| Judging communication | Each manager's own idea of good | One rubric, scored 1 to 5 per criterion |
| Morning routine | Rescheduling missed calls and chasing replies | Reading reports, watching recordings where scores disagree |
| Output | 6 tutors a month | 32 tutors in 3 weeks, about 10.7 a week |
Weekly pace is our arithmetic on the reported 32 hires in 3 weeks.
What does this case study not show?
- One platform, one hiring season, so treat it as a planning reference for your own cohort.
- The 10-point quality score is the platform's own measure, and the 22% satisfaction lift is its own cohort comparison.
- Today's booking page starts slots at least 2 hours ahead; the 45 minutes came from the March 2026 setup.
- The setup dates from March 2026. The product has changed since; this page describes today's workflow.
Who should hire remote staff this way?
Do
- Distributed roles judged mostly on how people explain and communicate.
- Candidates spread across time zones far from the hiring team.
- Seasonal cohorts with a hard start date.
- Teams where each manager judges communication differently.
Do not
- Roles that need a live teaching demo with a real student first.
- Hiring 2 or 3 tutors a term; interview them yourself.
- Skipping background and safeguarding checks. Those stay with your team.
What still needed a person?
Calibration and the decision. Hiring managers argued with some early scores, especially where a polished resume met a weak interview, and the team spent its first pass aligning on examples. After that, managers spent no hours on first-round interviews, but they still read the reports and chose every tutor.
Everything around the hire. Background checks, safeguarding for work with minors, contracts and onboarding stayed with the platform. The Cognitive is not an ATS, so offers and tutor records stayed in the platform's own systems.
Onboarding and student matching. The interview decides who joins the network. Whatever training, matching and support a platform runs after that decides how quickly a new tutor is ready for a nervous Year 9 student on a Sunday evening, and that work sits with the platform's own team.
Try it before your next cohort
Write one tutor rubric, set the interview to 20 minutes, open booking dates wide and invite every applicant this week. Start free at https://app.thecognitive.io/signup.
Where to go next
AI recruiting for education · Remote worker interview questions · Remote recruitment tools · High volume hiring · Pricing · All case studies
Sources and notes
Figures and quotes reported by the customer in The Cognitive's case study record for EdTech Platform, dated 31 March 2026. Setup dated March 2026. Differences, ratios and months at the old rate are our arithmetic. The example questions are illustrations. Booking behaviour describes the product as checked on 4 October 2026.
Benchmarks, all read 5 October 2026: SHRM 2026 Recruiting Benchmarking (median time to fill of 39 days for nonexecutive roles); Ashby 2026 Talent Trends, recruiter productivity (38 days application to offer for business roles, 8.9 and 12.2 interview hours per hire); Gem 2026 Recruiting Benchmarks (8% of applicants pass initial screening).
Frequently asked questions
How do you hire tutors quickly?
Remove the scheduling round trip and judge every tutor on one rubric. This EdTech platform let tutors book a live AI video interview themselves and reported hiring 32 in 3 weeks, against 6 a month before.
How do you interview across time zones without scheduling overhead?
Let candidates book their own slot in their own local time. No interviewer calendar is involved, because the AI runs the interview, so a tutor in Australia is no harder to assess than one in Boston.
What does a tutor interview actually assess?
Subject knowledge and the ability to explain it. The AI asks for explanations and follows up on them, which shows the difference between knowing a subject and teaching it. This platform also scored patience and engagement.
How long should a tutor interview be?
This platform used 20 minutes, the longer of the 2 lengths a recruiter can set. 20 minutes leaves room for an explanation, a follow-up and a scenario about a struggling student.
Can candidates interview right after applying?
Close to it. Today the booking page shows slots starting at least 2 hours after booking, at any hour within the dates the recruiter opens. This platform reported 45 minutes from application to first interview under its earlier setup.
Did tutor quality drop as volume went up?
The platform reports the opposite: its candidate quality score rose from 5.2 to 8.4 out of 10, and students rated the new tutors 22% higher than the previous cohort.
How do you hire remote staff consistently at scale?
Fix the rubric per role, let the interview come to the candidate, and read scored reports rather than call notes. Here every tutor was scored on the same criteria regardless of country.
What is the average time to hire for education and tutoring roles?
Published benchmarks are not split out for tutors. SHRM's 2026 benchmarking puts median time to fill at 39 days for nonexecutive roles, and Ashby's 2026 data shows about 38 days from application to offer for business roles (both read 5 October 2026). This platform hired 32 tutors in 3 weeks.
How do you assess teaching ability in an interview?
Ask the candidate to teach. Give a concept, ask for an explanation pitched at a named age group, then push back as a confused student would. A 20 minute live AI interview has room for that plus a scenario about a struggling student, and the rubric scores explanation, knowledge, patience and engagement separately.
What should a remote tutor rubric include?
This platform scored subject knowledge, patience, explanation quality and engagement. Our suggested weights put explanation first at 35, then knowledge 30, patience 20 and engagement 15. Keep it to 4 or 5 criteria so every score is easy to defend.
How do you hire tutors in several countries at once?
Take scheduling off the team's calendar. Tutors book a slot shown in their own local time, the AI runs the interview, and the report waits for the team. This platform hired across the US, Canada, the UK and Australia without anyone converting time zones.
Does the AI pick the tutors?
No. Each report suggests a verdict from the weighted score, and hiring managers decide. The Cognitive rejects nobody automatically.
Does it connect to our ATS?
Yes, through 60+ ATS integrations including Greenhouse, Lever, Ashby and Workable, or by CSV upload. Imported candidates get a note with score, summary and report link written back. ATS stages never move on their own.
How much does it cost?
AI Interview starts at $99/month and AI Sourcing starts at $49/month. Plans are monthly, so a seasonal team can run them for the back-to-school push and cancel anytime after. The pricing page shows what each plan includes.
Explore: AI soft skills assessment · Candidate fraud in remote hiring · Interview scheduling software · What is high volume hiring? · How the AI interviewer works