Choosing an AI apprenticeship provider is one of the highest-leverage L&D decisions a UK employer will make in 2026, and one of the least well-shopped.
The honest answer is that very few employers picked their AI apprenticeship provider the way they'd pick any other critical supplier. They picked the brand they'd heard of, or the one their account manager nudged them towards, or the one their levy-management platform integrated with. The actual diligence, Ofsted grade, achievement rate, coaching ratios, EPA outcomes, mostly didn't happen.
The UK apprenticeship market has had a difficult month. Several large providers have appeared in national press for the wrong reasons, downgraded inspection grades, completion rates well below the national benchmark, and learners stuck in cohorts that aren't delivering. Quietly, employers across the country are starting to ask the same question: are we with the right provider?
Below are the nine questions we'd ask any apprenticeship provider before signing in 2026, the answers that should make you confident, and the answers that should make you walk. We'll happily answer all nine on our own behalf at the end.
Buying an AI apprenticeship is a bit like buying breakfast cereal. Two boxes on the shelf can look almost identical, flip them over, and one has whole oats; the other is held together by artificial sweeteners and three E-numbers you can't pronounce. The nine questions are how you read the back of the box. , Lisa O'Reilly, Director, TESS Group
Copy them, take them into your conversation with any provider, write the answers down. Compare. The provider who answers cleanly on all nine isn't selling you on marketing, they're selling you on delivery.
What's your current Ofsted grade, and when was your most recent inspection?
The single most important question. Ofsted's Education Inspection Framework grades apprenticeship providers Outstanding, Good, Requires Improvement or Inadequate. The grade reflects everything from leadership and teaching quality through to outcomes for learners.
A provider rated Requires Improvement cannot accept new apprenticeship starts. A provider rated Inadequate is placed on a 3-month re-inspection clock and at serious risk of having its apprenticeship licence withdrawn entirely.
If your apprenticeship provider has been downgraded since you signed, your existing apprentices are in a difficult position, and your right to switch them to another provider is clearer than most employers realise. (See question 9.)
If the grade isn't Good or Outstanding, every other answer is less important than fixing that.
What's your achievement rate (QAR) compared to the national benchmark?
QAR, the Qualification Achievement Rate, is the percentage of starters who actually finish and pass. The national benchmark just moved: as of the 2025/26 data published in 2026, the national average is now 65.5%, up from 54% the year before. That's the biggest single-year jump in a decade, and it means the bar for a 'good' provider is meaningfully higher than it was when most procurement decisions were last made. A high-quality provider should still sit measurably above the new benchmark, not just clear last year's bar.
Ask for two numbers: the overall company QAR and the QAR specifically for the standard your apprentice would be on. Some providers have a strong company average dragged up by one or two standards and dragged down by everything else, you want the standard-specific number.
A provider that won't share their QAR is telling you something. A provider whose QAR sits below the national benchmark is telling you something louder.
What's your coach-to-apprentice ratio, for the cohort my apprentice will actually join?
Some providers operate at industrial scale: skills coaches carrying caseloads of 80, 100, even 120 apprentices. At that ratio, the model has to become effectively self-service with periodic check-ins. There's no real 1:1 coaching available because the maths doesn't allow it.
A healthy ratio sits in the 20–35 range. Substantive 1:1 sessions happen regularly, feedback turnaround is days not weeks, and the coach actually knows what your apprentice is working on.
Always ask about your cohort, not the company average. The two numbers can be very different.
Coaching ratio is the single biggest hidden quality variable. A great curriculum delivered by a coach with 100 other learners isn't a great experience.
Are your skills coaches employed by you, or freelance contractors?
This is a model question that almost no employer asks. Some providers employ their skills coaches full-time; others rely heavily on freelance contractors paid per learner per month. The contractor model scales faster, but it has downsides: less consistency, weaker continuity, and (legally awkward) coaches who often work simultaneously for two or three competing providers.
Ask: Will the coach assigned to my apprentice be employed by you? Will they be the same coach for the full duration of the programme? What's your coach turnover rate?
The same name on the front of the brochure doesn't mean the same delivery model behind it.
What's your End-Point Assessment pass-first-time rate?
EPA is the independent assessment that determines whether your apprentice has met the standard. Pass-first-time rate is a strong indicator of provider quality: a provider whose apprentices regularly need re-sits is either teaching to the wrong target or not preparing learners thoroughly.
Re-sits are stressful for the apprentice, costly in time for the employer, and a sign that something upstream isn't working. Ask for the standard-specific number, not the company average.
EPA is the result. Everything before it is process. If process doesn't convert into a pass-first-time outcome, the process isn't working.
What happens if my apprentice falls behind, what's your recovery process?
Every cohort has at least one apprentice who hits trouble: a workload spike, a personal issue, a confidence dip. The difference between a high-quality and low-quality provider is what happens next.
A serious provider has a documented recovery process: additional 1:1 support, an adjusted plan, regular check-ins with both apprentice and line manager, clear early-warning indicators. A weak provider waits until the apprentice withdraws.
Ask for an example. "Tell me about an apprentice who fell behind in the last six months and what you did" is a question that separates the genuine operators from the script-readers.
You're paying for outcomes, not for hopeful starts. Recovery capability is the difference.
Can I see a sample portfolio from a recent completer on this standard?
The apprentice's end-point portfolio is the most concrete evidence of what they actually learned. Ask to see a recent one (anonymised). Look at the depth of the projects, the quality of the reflection, whether the work was meaningfully theirs or whether it was templated.
If a provider can't show you a single completed portfolio for the standard you're considering, that's worth knowing. Either they haven't completed enough apprentices on it yet, or the portfolios aren't in a state they're comfortable showing.
The portfolio is the apprentice's "CV" for the rest of their career. A weak portfolio is a 10-year drag, not a 12-month one.
Who designed the curriculum, and how often is it updated for new AI tools?
AI is the fastest-moving sector in the apprenticeship landscape. A 2024 curriculum that hasn't been updated for Claude 4.5, Gemini 3.5, GPT-5, Antigravity, MCP, and agent orchestration is already obsolete.
Ask: Who owns the curriculum? When was it last updated? How fast can it absorb a new tool like Antigravity or Gemini Spark? Who reviews it, do you have practitioners actively shipping AI in production?
For context on why curriculum currency matters in 2026, see our Google I/O 2026 skills gap analysis.
You're paying for an apprentice who'll be in the workforce for 30+ years. The curriculum that prepares them needs to reflect the next 30 years, not the last three.
What does it actually take to switch a current apprentice to a new provider?
This is the question employers don't ask because they assume switching is impossible. It isn't.
The ESFA explicitly allows employers to change provider during an apprenticeship. The receiving provider takes on the learner, conducts an initial assessment, agrees a recovery plan if needed, and continues delivery from the agreed point. Funding transfers via the apprenticeship service. Most employers find the process simpler than expected, the friction is usually less than the cost of staying with a provider you've lost confidence in.
If you're currently with a provider whose grade has been downgraded, or whose answers to the questions above haven't reassured you, a credible new provider should be able to walk you through the switching process in 20 minutes.
Sunk-cost fallacy keeps a lot of cohorts stuck with the wrong provider. The cost of staying is usually higher than the cost of switching.
How TESS answers the nine questions.
Fair's fair, if we're going to publish a checklist, we should answer it. Every question above is one we welcome.
Good across the board at our most recent inspection. We publish our inspection report and we're proud of it.
Consistently above the new 65.5% national benchmark. The benchmark jumped 5.9 points this year, the bar moved with it, and we still clear it on standard-specific numbers. We publish QAR by standard on request, no fishing.
We deliberately stay small. Caseloads in the 20–35 range, never the 80+ caseload model. We turn down cohort growth when it would force ratios up.
Our skills coaches are employed, not freelance. The same coach stays with an apprentice for the full duration. Coach turnover is very low.
Strong on both ST1512 (AI & Automation Practitioner L4) and ST1398 (Machine Learning Engineer L6). Specifics on request.
Documented, owned by a named team, with early-warning indicators triggered at week-by-week milestones. We can talk you through a recent recovery case on a call.
Happy to share sample completer portfolios under NDA for the standard you're considering.
Owned by our practitioner team, updated quarterly at minimum and immediately when a major tool ships (Antigravity and Spark are already in the L4 curriculum). Reviewed by people who actually ship AI in production.
We've onboarded apprentices mid-cohort from other providers. We can walk you through the process in 20 minutes on a discovery call, it's less complicated than most employers fear.
We're a deliberately small provider in a market where the headlines go to deliberately large ones. The trade-off is that we can answer all nine of these questions in a way the big providers can't. That's not an accident, it's a strategic choice. , Rod Doyle, Director, TESS Group
What to do next.
Most of them end with the employer realising the switch is straightforward and the risk of staying is higher than they thought. Same-day slots usually available.
We'll happily put ourselves through it on a discovery call alongside any other provider you're evaluating. No deck, no pitch, no obligation.