The short version
- The best coaches translate. They turn AI and data into the learner's own world, in the learner's own words. That beats technical depth every time.
- Coaching is a quiet job. The learner should be talking eighty percent of the time. The coach's skill is the question that unlocks something.
- Average coaches assume. That the learner is too busy, only wants half an hour, is probably fine. Great ones look at each person and ask how to get the most out of them.
- Caseloads decide everything. A coach with too many learners cannot see anyone closely. That is a provider decision, and you should ask about it before you sign.
- It all spins from one connection. A happy learner makes a happy employer. Everything else, including Ofsted, follows from that.
I have interviewed a lot of coaches this year. We are growing the AI and data side of the business and it needs people. Somewhere in the middle of one of those conversations I said something I have been trying to say properly for about a decade, and I want to write it down before I forget how I put it.
The sentence I would stake the business on
The best skills coaches are rarely the cleverest people in the room. They may not have a level fifty in AI. What they are good at is translating AI and data into the learner's world.
That is the whole job, and most of the sector has it backwards. When providers recruit coaches for a technical standard they go looking for technical depth: certifications, years in the field, the person who can talk about model architectures. That is useful. It is not what makes the difference.
What makes the difference is whether the coach can sit across from a finance assistant who has never touched an automation tool, understand what her month actually looks like, and show her how the thing she is learning takes two days out of it. That is a translation skill. It has more in common with teaching than with engineering, and it is rarer than either.
It is a quiet role
People who have not done it assume coaching is about what you say. It is mostly about what you do not.
Eighty percent of the time you want the learner talking, not you. You come in with the little gems, the nuggets where they go "oh, that's good, I'll have that." The skill is knowing which question unlocks thinking they weren't already doing.
I have sat in interviews where someone talked at me for twenty-five minutes without a pause, and however impressive the content, I cannot put that person in front of a learner. A coach who fills the space leaves no room for the apprentice to work anything out. The best ones are almost uncomfortable to watch at first, because they let silences run and let the learner reach the answer.
The magic hat
There is one habit I look for above everything else, and it is hard to test for in an interview because it only shows up when someone is generous by nature.
The best coaches have a magic hat. "Have you tried this? I learned something the other day, you'll love it, can I share my screen?" That is cash to a learner. You are giving it away for free, but it is worth real money to them, because they take it back to work and use it.
On an AI programme this matters more than on almost any other, because the tools change every month and the learner's employer usually has nobody else to ask. A coach who turns up with something new and practical each time is not just teaching the standard. They are the learner's only line to what is actually possible right now, and that is worth more to the employer than any module.
Where average coaches go wrong
I do not think many coaches are bad. I think a lot of them are trained, by caseloads and by habit, into a pattern that produces average results. It comes down to one word.
The average coach assumes
- The learner is too busy to see me
- They only want half an hour
- They are probably getting on fine
- The formal review is the job; the informal chat is optional
- Once a month is what everyone gets
The great coach asks
- What does this person actually need to get through?
- Who in their workplace is on their side?
- What did they build last week, and did it work?
- What is the one thing I can show them today?
- How often does this learner need me?
That last line is the one I would underline. If someone scores themselves one out of ten on AI at the start of a programme, you cannot see them once a month and say "good luck, see you in four weeks". You need to be in there two or three times, closely, until the number moves. A coach who treats every learner as a monthly slot in a diary will lose exactly the people who needed them most.
Your nervous apprentice and your confident one will get the same coaching diary at most providers. Ask whether that is true at the one you are considering, because it is the difference between a completion and a withdrawal.
Why we cap caseloads, and why it costs us money
None of the above is possible if a coach has too many learners. That is not a criticism of coaches. It is arithmetic. So we made a decision a long time ago that costs us margin and I would make again tomorrow.
We would rather bring in more coaches with smaller caseloads than a few with enormous ones. We know what large caseloads do. Coaches burn out, learners get a check-in instead of a relationship, and the whole thing turns into a factory. We do not want that style.
We also do not set a flat number. Some learners are more complex than others, and a coach supporting someone with additional needs, or someone starting from nothing, is carrying more than the headcount suggests. So the load flexes with who is actually on it.
The way I think about it is a football team. You give your best players space on the ball and put them in the right positions. You do not play your striker in defence because the spreadsheet says everyone gets the same minutes. Coaching is the same. Space is what lets a good coach be good.
Curriculum comes from the people doing it
One more thing I feel strongly about, because I have seen the alternative.
Curriculum cannot be some bloke in an ivory tower writing a bank of tasks. That is not connected to reality. The coaches are the ones living and breathing it, so they shape it.
On our AI & Automation Practitioner Level 4 this happens constantly. A coach will tell us that a module has three separate tasks that could be one, or that a piece of evidence is hard to mark consistently, or that learners keep stumbling at the same point. We change it. That is only possible when coaches have the time to notice and the standing to say so, which brings us back to caseloads.
It all spins from one connection
If I had to reduce sixteen years of running a training provider to one causal chain, it would be this.
A happy learner makes a happy employer, and that makes Ofsted happy. It all spins out from the connection between the coach and the learner. Everything else follows from that.
Providers spend a great deal of energy on the things employers can see: the brochure, the sales conversation, the badge on the website. We wrote recently about how even the Ofsted badge has changed under you. But the thing that actually decides whether your levy turned into capability is a relationship you were not in the room for. The good news is that you can ask about it before you commit.
Five questions to ask any provider about their coaches
- Who will coach my apprentices, by name, and what is their background?
- What is their caseload, and how is it set? A flat number is a warning sign.
- How often would a struggling learner be seen, compared with one who is flying?
- Do your coaches shape the curriculum, or just deliver it?
- Can I speak to a current apprentice about their coach?
A provider who can answer those five quickly and specifically is one whose coaches are doing the job described above. One who cannot is telling you something the sales deck will not. Our longer guide to choosing an AI apprenticeship provider covers the rest.
We are hiring
If you read the middle of this and thought "that's how I coach", we should talk.
We are growing our AI and data coaching team. We hire for the qualities above before technical depth, though depth helps: people who translate, who ask more than they tell, and who would rather see someone three times in a month than lose them. Smaller caseloads than you are probably used to, and a curriculum you get to change. You can also see what our learners say about the coaching on our about page.
Ask us the five questions. We will tell you who would coach your people, what their caseload is, how often a nervous starter gets seen, and put you in touch with a current apprentice so you can ask them instead of us. 25 minutes, no obligation, and if another provider is a better fit we will say so.
Book a conversation →The honest summary
I am not going to pretend we get every hire right or that every learner has a perfect experience. What I will say is that we decided years ago what we were optimising for, and it was not the number of learners per coach.
Someone once described the job I do as being a bit like a pilot: not trying to go faster than everyone else, just trying to land the thing smoothly every year. Coaching is the same. The best coaches are not the ones with the most impressive instruments. They are the ones who get the person to where they were going.
Frequently asked questions.
What does an apprenticeship skills coach actually do?
A skills coach is the person who works with an apprentice one to one across the whole programme, from the initial baseline review through to preparing them for end-point assessment. Tutors deliver the taught content in workshops. The coach is the one who turns that content into something the apprentice can apply in their own job, tracks progress, spots when someone is struggling, and keeps them moving. On an AI apprenticeship, they are usually the person the apprentice goes to when a real workplace automation does not behave.
What separates a great skills coach from an average one?
Translation. The best coaches are not necessarily the most technical people in the room. What they are good at is turning AI and data into the learner's own world, in the learner's own language. They ask questions that unlock thinking rather than talking at people, they treat each learner as an individual rather than a slot in a diary, and they give away practical things the learner can use tomorrow. The coaches who get the worst results are usually the ones who assume: that the learner is too busy, only wants half an hour, or is doing fine.
Why does the skills coach matter so much to an employer?
Because they are the biggest single variable in whether your person finishes, and finishes well. The provider's brand, the curriculum and the funding are the same for every learner on a standard. The coach is not. A learner who scores themselves one out of ten on AI at the start needs a coach who will see them more often and closer, not one who checks in monthly. That relationship is what turns a levy-funded place into an actual capability in your business.
How does TESS Group manage coach caseloads?
Deliberately, and not by a fixed number. Some learners need more time than others, so we vary the load according to who a coach is actually supporting rather than applying a flat figure. The principle is to bring in more coaches with smaller caseloads rather than a few with enormous ones, because we know what large caseloads do: coaches burn out, learners get a monthly check-in instead of a relationship, and it turns into a factory.
What should I ask a training provider about their coaches?
Ask who will coach your apprentices, by name, and what their background is. Ask what the coach's caseload is and how it is set. Ask how often a struggling learner would be seen compared with one who is flying. Ask whether the coaches shape the curriculum or just deliver it. And ask to speak to a current apprentice about their coach. A provider that cannot answer those quickly is telling you something.
Is TESS Group hiring AI skills coaches?
Yes. We are growing our AI and data coaching team and we hire for the qualities described in this article: people who can translate AI into a learner's world, who ask more than they tell, and who would rather see someone three times in a month than lose them. Technical depth helps, but it is not the thing we hire for first. If that sounds like you, get in touch.
The quotations in this piece are the author's own words, drawn from conversations with coaches and candidates during September 2026 and lightly edited for print. No individual is identified and nothing here describes any specific person or interview. TESS Group is an Ofsted Good, Skills England approved provider of AI and leadership apprenticeships.