How to Train Volunteer Mentors for a Coding Club
Our club recruited nine volunteers in one term. Three were still turning up by week six. The two who dropped out fastest were the strongest programmers in the room, a backend developer with eleven years of experience and a maths teacher...

Our club recruited nine volunteers in one term. Three were still turning up by week six. The two who dropped out fastest were the strongest programmers in the room, a backend developer with eleven years of experience and a maths teacher who writes C for fun, and both left for the same reason: they could not stand watching a thirteen year old type the same mistake four times without reaching for the keyboard.
That is the real problem with mentor training. It is not that volunteers lack technical knowledge. Most have far more than they need. What nobody has given them is permission to be slow, and a few concrete things to say when a student is stuck and the clock is at eleven minutes past the hour. I have run this onboarding about fifteen times now, and the version that works is shorter and more awkward than the one I started with.
The habit that ruins a good mentor
There is one behaviour I watch for in a new volunteer's first session: whose hands are on the keyboard. A mentor who takes the laptop, fixes the indentation error and hands it back has just removed the only part of the hour that was teaching anything. The student does not learn where the error came from, and worse, learns that being stuck is an emergency that an adult resolves.
It happens because the adult is uncomfortable, not because the student asked. Watching someone fail to spot a missing colon for ninety seconds feels unbearable if you can see it from across the room. So I say this out loud in training and then again in the first session: your discomfort is not evidence that the student needs help. Ninety seconds of struggle is normal, and three minutes is usually still fine.
The rule we settled on is that the mentor's hands stay off the keyboard unless the student is installing something or the machine itself has broken. Everything else gets handled with a question or a pointed finger. Our piece on helping a beginner debug without fixing the bug for them goes into the specific questions that work.
The ninety minute onboarding we actually use
I used to run a two hour session with slides about pedagogy. Attendance at the second half was poor and nothing transferred. What replaced it fits in ninety minutes and is mostly practice.
- Fifteen minutes on who is in the room. Ages, who comes with a friend, who comes because a parent made them, and the two or three students who will finish every task early. No teaching theory, just names and patterns.
- Twenty minutes of roleplay where I play a stuck beginner and the volunteer tries to help without touching the keyboard. I deliberately give wrong answers to their questions. This is the part everyone finds uncomfortable and the only part that reliably changes behaviour.
- Fifteen minutes on the three sentences to use when you do not know the answer, which I cover below. Volunteers rehearse them aloud. It feels silly and it works.
- Twenty minutes on the actual project the club is building this term, done at a beginner's pace on a beginner's machine, so the mentor has hit the same setup problems the students will hit.
- Ten minutes on logistics: safeguarding contact, what to do if a student discloses something, where the spare chargers live, and who locks up.
Everything else, assessment, differentiation, how to explain recursion, waits until a volunteer has done two sessions and has real questions. Front loading it just fills a notebook nobody opens again.
Why screening mentors for technical depth is the wrong filter
The advice I kept reading when I started was to recruit mentors who know the language well, and to pair strong programmers with the students who are ahead. We tried it for two terms. It produced the worst retention we have had, on both sides.
The volunteers who lasted were a nurse who had done a twelve week Python course, a retired bus driver who had never coded at all but was extraordinarily patient, and a sixth former who had finished the same curriculum eighteen months earlier. The sixth former was the best mentor in the room, and her technical knowledge was roughly one step ahead of the students she helped. That gap is an advantage: she still remembered what confused her, and she could not solve anything instantly even if she wanted to.
The filter that predicts a good mentor, as far as I can tell from a sample of about thirty volunteers, is whether they can sit with silence. I now ask one question at the informal chat before anyone signs up: tell me about a time you taught someone something badly. People who answer with a real story, including the part where they went too fast, almost always work out. People who cannot think of one tend to be the keyboard grabbers.
Technical depth still matters for one role, the person who fixes the broken install on the school laptop that has not been updated since 2021. Keep one of those. You do not need six.
Three sentences that cover the moments mentors freeze
New volunteers panic most when a student asks something they cannot answer. They either guess, which creates a misconception that takes weeks to undo, or they go quiet and the student reads it as judgement. So we rehearse exact wording.
For not knowing: I am not sure, let us find out together, where would you look first? This models the actual job. I have had volunteers tell me months later that this was the single most useful thing they took away, because it removed the fear of being caught out.
For a student who is frustrated to the point of shutting down: that error message beat me for twenty minutes last week too, shall we read it one line at a time? Naming your own failure first does more than any encouragement does.
For the student who finished in nine minutes and is now bored and loud: can you write a version that breaks, and I will try to work out why? It buys fifteen minutes and teaches more than the extension task would have.
Keeping volunteers past the fifth week
Mentor churn is mostly a scheduling problem dressed up as an enthusiasm problem. Asking someone for every Tuesday until July is asking for a resignation in March. We run a rota where each volunteer commits to one session in three, which means we need six people to staff two mentors a week, and nobody feels trapped.
The second thing that moved the number was a fifteen minute debrief after each session, standing up, no agenda beyond two questions: what went badly, and who do we need to watch next week. It is the only time volunteers hear that the thing they found hard was hard for everyone. Our attendance among mentors went from about half the rota to roughly four in five once the debrief became routine, and nothing else changed that term.
If you pair mentors with each other rather than one adult to a table, read why pair programming works differently with complete beginners first, because the same dynamics apply to two adults sharing a group. More on running sessions sits in the teaching code section.
FAQ
How many mentors does a club actually need?
One mentor for every six or seven beginners works for a mixed group. Below that ratio the adults end up firefighting installs instead of teaching. For a first session, where half the hour is setup, aim for one to four if you can get the bodies.
Do mentors need a background check?
If the club runs on school premises with students under eighteen, follow whatever the school requires, and assume it takes four to eight weeks. Build that delay into recruitment, because a keen volunteer who waits two months for paperwork often stops being keen.
What if a mentor keeps giving away answers?
Give the specific feedback, not the principle. Saying you solved three bugs for Amira tonight lands; saying remember to let them struggle does not. Two rounds of that fixed it for every volunteer except one, who was happier managing the equipment and sign-in sheet.
Can teenagers mentor younger students?
Yes, and they are often the best at it. Keep them in sight of an adult, give them one table rather than roaming, and do not ask them to handle behaviour problems. The ones who did it for a year came back as organisers.
More in Teaching Code
Teaching Code
How to Teach Beginners to Read Python Documentation
I ran a small test at club last February because I was tired of a specific pattern.
Teaching Code
How to Teach the Difference Between print and return
In week five of our adult evening group, a learner called me over because her code had stopped working.
Teaching Code
Trace Tables: Teaching Beginners to Run Code on Paper
A student shows me a loop that should add the numbers from 1 to 10 and prints 45 instead of 55.