Two shapes of provider, four criteria that separate them, a method for reading the DfE outcome data, and the questions to ask before you sign.
Data Analyst Apprenticeship Providers UK: How to Choose
By James Cotton · Last updated · 18 min read
By James Cotton, Founder, iO-Sphere
You have probably been handed a shortlist: a few provider names, pulled from a search result or a colleague's recommendation, with an instruction that comes down to rank these and pick one. The shortlist is the wrong unit to work from. A list of names invites the question which name is best, and that is not what decides whether your apprentice finishes able to do the job.
Providers delivering the same Level 4 Data Analyst apprenticeship differ far more in how they run it than in the name over the door, and those differences do not resolve into a single ranking. So work the problem as a method: decide which shape of provider fits your situation, read the outcome evidence that is already public, and ask the few questions a sales page will never answer. Do that, and the shortlist ranks itself.
What every data analyst apprenticeship shares: the ST0118 standard
Before the differences, the part that is identical everywhere. Every Level 4 Data Analyst apprenticeship in England runs on one approved standard, the Data Analyst standard, ST0118 (currently version 1.1, approved for delivery), set by Skills England, the government body that approves apprenticeship standards. It builds someone who can collect, clean, analyse and present data to answer business questions: SQL and a programming language such as Python, dashboards and visualisations, and findings a decision-maker can act on. Because there is a single standard, the curriculum on paper is broadly common from one provider to the next.
That matters in a specific way. The standard is common ground, so two providers can both promise the ST0118 outcome and turn out very different analysts. When a comparison leans on curriculum content or tool lists to tell providers apart, it is comparing the part that barely varies.
A few numbers anchor the rest of the page. The standard carries a minimum of 370 hours of off-the-job training, and by law those hours sit inside the apprentice's normal, paid working hours as protected learning time, so ask every provider how they schedule them.
Skills England sets the funding band, the cap on the training cost, at up to £15,000 for ST0118, and revises the bands periodically. The standard's typical duration is 24 months; our own Advanced Data & AI programme delivers ST0118 in 15 months of training plus a 3-month end-point assessment, a useful prompt to check whether a duration figure in any comparison describes the standard or a provider's actual delivery.
Across the occupation the ONS classifies as data analysts (Standard Occupational Classification code 3544), median gross annual pay was £38,107 (ONS ASHE Table 14, 2025 provisional). For a buyer that is a useful anchor on what the capability is worth in the market: it is the measured median across everyone in the role, part-time included, not a starting salary, and it moves with sector, region and experience.
Is there a written exam?
No, not by default. ST0118 is assessed at the end through an end-point assessment (EPA) built on the apprentice's own work: a portfolio and project review plus a professional discussion, judged by an independent assessment organisation. A knowledge test appears only where the standard's assessment plan calls for one. One change to track: under the 2025-26 reforms the government is moving from a single end-point assessment towards assessment spread through the apprenticeship, so confirm which model applies to your start date.
The first cut: which shape of provider fits your situation
The names on a data apprenticeship shortlist sort into two shapes well before they sort into any ranking, and the shape is the first thing to settle, because it decides which providers can serve your situation at all.
Scale-generalists run apprenticeships across many disciplines at once, leadership, project management, digital, data, with the enterprise account machinery and high-volume onboarding to match. Multiverse, QA, BPP and Kaplan sit here. If you are rolling several hundred apprentices across multiple standards and departments through one supplier, that machinery is exactly what you want, and the data programme is one lane inside a broad scheme.
Data and AI specialists run a narrow portfolio deeply. Cambridge Spark, Corndel, Decoded, NowSkills and iO-Sphere sit here. When the capability you need is deep data and AI practice, a specialist's whole operation is pointed at that one outcome. We are a specialist, and the trade-off comes with saying so: if what you need is a broad, multi-discipline rollout run through enterprise procurement, a scale-generalist will probably fit you better than we do.
Two further providers, Estio and Babington, also deliver the standard and appear in the tables below; we are not sorting them into either shape here. One shape is not better than the other; they are built for different jobs. A generalist serves the enterprise buyer well and the two-person team poorly, and a specialist does the reverse. So settle the shape before you weigh the names, and be wary of any list that folds both into a single ranking.
If what you want next is the provider-by-provider run-through, each named provider's live offer with a note on who it suits, that comparison lives in our alternatives to Multiverse guide.
The four criteria that separate providers
Once you have settled the shape, four questions do the real separating. A brochure answers none of them, and the same four work on any provider, ours included. We give our own answers to all four in the section below, stated once, on the same terms we are asking you to hold everyone else to.
1. Who will coach your apprentice, and what did they do before they taught? This is the question that separates providers fastest. Some deliver through career trainers who teach the subject; others coach through people who have done the analyst's job. Ask for the actual coaches' backgrounds, not a marketing line about "expert tutors". A career trainer can suit a learner who needs structured conceptual scaffolding, but if the gap to close is practitioner craft, cleaning messy production data, handling stakeholder requests under time pressure, a coach who has never done that work cannot close it.
2. What will your apprentice practise on during those 370 hours: their own work, or a content library? Those 370 off-the-job hours have to be filled with something. Fill them with the apprentice's own live analytical problems and the learning attaches to the job; fill them with generic tutorial datasets and video modules and the skill drifts away from the role.
This is where learn-by-doing and classroom-first delivery genuinely diverge, and it is the biggest single driver of whether an apprentice comes out capable. Data skill is procedural: you learn it by doing the work under a coach, not by watching it explained first and applying it later. If a provider cannot tell you concretely what fills the hours, assume a library.
3. Can the provider run a cohort your actual size? This is the criterion buyers who chose on brand recall discover in month two. A provider built for enterprise volume often has no lane for a two-person cohort: the onboarding, the account rhythm and the escalation path are all calibrated for scale, and a small cohort inside that machine does not get lighter service, it gets no defined service at all. Ask the question in the form that exposes it: what happens to a cohort of two?
4. What can you check on the public record? Two things. First, is the provider actually approved to deliver the Data Analyst standard? The government's Find apprenticeship training service confirms that in a minute. Second, what do its published outcomes say? The DfE publishes provider-level achievement, retention and pass rates for this exact standard every year, so you never have to take a provider's word for its results: you can read them, and the next section shows you how to read them well.
How to read the published DfE outcomes: three checks no provider will publish
A league table can mislead you even when every number in it is true. The DfE data is genuinely comparable and free to open, but the headline figure hides more than it shows. Here are the two tables, then three checks to run on any provider's row.
First, the most recent year in full, with the national row in the middle so every provider rate can be read against something rather than against zero. Achievement is the headline; retention and pass are the two numbers it is built from, and leavers, the people who finished or exited the standard that year, is the base each rate is measured on: the larger it is, the more the rate means.
| Provider | Achievement | Leavers | Retention | Pass |
|---|---|---|---|---|
| Babington | 89.7% | 40 | 94.9% | 94.6% |
| QA | 70.6% | 770 | 71.9% | 98.2% |
| Cambridge Spark | 69.1% | 530 | 69.1% | 100% |
| Corndel | 67.1% | 350 | 67.1% | 100% |
| BPP | 66.7% | 110 | 66.7% | 100% |
| Estio | 63.7% | 100 | 68.6% | 92.9% |
| NowSkills | 63.6% | 20 | 63.6% | 100% |
| All providers, England | 58.2% | 5,620 | 59.2% | 98.3% |
| Decoded | 53.9% | 210 | 56.8% | 94.9% |
| Multiverse | 50.1% | 2,770 | 50.9% | 98.5% |
| Kaplan | 46.4% | 30 | 57.1% | 81.3% |
Level 4 Data Analyst (ST0118), 2024-25, all ages. Provider figures from the DfE achievement rates by provider dataset; the national row from the achievement rates by standard table. Both were published in March 2026 and both are free to open. The tables list every provider named on this page plus any other provider with 200 or more leavers on this standard in 2024-25, so read them as a sample of the wider field, not the whole of it: a high rate on a very small cohort is not evidence that a provider leads the standard.
Then the three-year line, because one year cannot tell you whether a rate is a track record or a good twelve months.
| Provider | 2022-23 | 2023-24 | 2024-25 |
|---|---|---|---|
| Babington | 65.2% | 52.0% | 89.7% |
| QA | 61.4% | 62.9% | 70.6% |
| Cambridge Spark | 48.1% | 65.8% | 69.1% |
| Corndel | 47.3% | 67.1% | 67.1% |
| BPP | 43.5% | 68.6% | 66.7% |
| Estio | 66.0% | 66.1% | 63.7% |
| NowSkills | 70.0% | 57.1% | 63.6% |
| All providers, England | 47.7% | 59.6% | 58.2% |
| Decoded | 40.9% | 51.5% | 53.9% |
| Multiverse | 43.8% | 56.5% | 50.1% |
| Kaplan | suppressed | 44.4% | 46.4% |
Achievement rate on the same standard and source, all three years published March 2026. "Suppressed" means the DfE withholds a rate where the cohort is too small to publish safely; Kaplan's 2022-23 cohort was 10 leavers. Leaver counts move year to year and the small rows move with them: Babington's three rates rest on 20, 30 and 40 leavers, NowSkills's on 10, 10 and 20, Kaplan's on 10, 20 and 30, and BPP's on 70, 50 and 110. A rate built on tens of people carries very little signal; a rate built on hundreds carries real signal.
Check one: read every rate against the national row, never against zero. The national rate moved from 47.7% to 59.6% to 58.2% across the three years, so most provider rates rose in the middle year simply because the whole field rose, by 11.9 points between 2022-23 and 2023-24. A provider that gained 12 points that year barely moved relative to the field. The national row is the tide; what you want to know is how each provider swam against it.
Check two: weight by volume before you trust the average. Multiverse's 2,770 leavers in 2024-25 are close to half of the national 5,620. When one provider carries that much of the total, the national average largely is that provider, and by simple arithmetic almost everyone else has to sit above it. So "above the national average" is a weaker claim on this standard than it sounds. Compare rows against each other at comparable volume, rather than against an average one provider dominates.
Check three: split retention from pass, because the headline cannot tell you which one broke. Achievement is, near enough, retention multiplied by pass. So a single achievement figure cannot tell you whether a provider is losing people before the assessment or failing them at it.
Take the largest row. Multiverse's pass rate of 98.5% is a shade above the national 98.3%, while its retention of 50.9% sits 8.3 points below the national 59.2%. The people who reach the end-point assessment pass it at the field's rate; the loss opens earlier, among the roughly half who leave before reaching assessment at all. If you are about to put people onto a programme, retention is the number to ask every provider for by name.
Where iO-Sphere sits, and what to weigh instead of a rate
We are not in the tables above: iO-Sphere is new, so our first cohorts have not reached assessment yet, and there is no rate to publish. When they do, our rate will appear in that same DfE table, on the same basis as every provider above. The line is young by design, because we built it after AI changed how analysts work, so its curriculum is native to the work as it is done now.
So here is what you can check today. iO-Sphere is listed on the Apprenticeship Providers and Assessment Register (APAR) and is eligible for Ofsted inspection, which puts us on the same register as any provider on your list. On the criterion that separates providers fastest, our coaches are practitioners who did the analyst's job in industry before they taught it.
On the other two criteria: apprentices work on their own live analytical problems from the start, coached through them, which is also what leaves them with genuine portfolio evidence for assessment rather than a folder assembled near the end. And where the standard calls for a skill their own workplace cannot practically supply, they practise it on realistic case studies and simulations, our Prism method, so no part of the standard goes uncovered.
And on size, two people from an SME join an open cohort alongside learners from other companies and industries, with no minimum on your side, while a closed cohort built solely around your organisation is where we look for 10 to 12. The open cohort is not the budget option: its cross-company mix is the point of it, giving a lone analyst what their own workplace cannot, which is other people doing the same job elsewhere to test their thinking against.
Beyond the register, some numbers you can weigh. Since 2022 we have trained more than 900 people in data and AI through our bootcamps and funded qualifications. Those learners are not apprentices, which is a newer line for us, but the same coaches, the same learn-by-doing model and the same delivery team run all of it. Across 78 Google reviews we are rated 4.8 out of 5.
The published outcome rate is the one thing we cannot show you yet. So here is the commitment that goes with it: a page that tells you to check every provider should be checkable itself, and our cohorts will appear in that same DfE table when they complete, on the same basis as every provider above. Hold us to that. Until then, weigh us on the two criteria that decide capability and that you can check now, who coaches and what apprentices actually build, and put the identical questions to everyone else on your shortlist.
Levy funding, and why it comes last
Funding comes last here on purpose. It is real money and worth getting right, but it barely varies between reputable providers, so it should confirm a decision you have already made on fit rather than make the decision for you.
A data analyst apprenticeship is paid for through the apprenticeship levy: the Growth & Skills Levy (formerly the Apprenticeship Levy), charged at 0.5% of an employer's pay bill above £3 million. The levy is collected UK-wide, but the account you spend from only covers apprentices in England; Scotland, Wales and Northern Ireland run their own schemes. Any start you plan now falls in the 2026-27 funding year (starts from 1 August 2026), so these are the rules that apply (gov.uk apprenticeship funding rules 2026-27, version 3, amended 29 July 2026):
- A non-levy employer with an apprentice aged 16 to 24 at the start: 100% government funded, £0 to the employer.
- A non-levy employer with an apprentice aged 25 or over: you pay 5% of the price, the government pays 95%.
- A levy payer whose account has run short: you co-invest 25% for an apprentice aged 25 or over; an apprentice aged 16 to 24 at the start stays 100% government funded even when the pot is empty.
An employer that does not pay the levy is not shut out. The cost is met through co-investment, or through a levy transfer of up to 50% of a larger employer's unused funds, which often covers the whole of an SME's training cost, so it is worth asking a provider whether they can introduce a levy-transfer partner. The training cost itself is capped by the standard's funding band: up to £15,000 for ST0118 (Skills England; bands are revised periodically).
So the real question is not whether it is free. A funded programme that does not change what your analyst can do is the costliest option in front of you, because it spends working time you never get back. Decide the capability outcome first, fit the funding to it, and weigh the route as though you were paying for it in cash.
Questions to ask a provider before you sign
Take the same questions to every provider on your shortlist and lay the answers side by side. The provider whose answers are specific and checkable is the one worth a scoping call.
- Who will coach my apprentice, and what did they do before they taught? Ask for real backgrounds, not a line about "expert tutors".
- What will my apprentice practise on during the 370 off-the-job hours: their own work, or a content library?
- Are you approved to deliver ST0118, and what is your published achievement rate on it? Check both against Find apprenticeship training and the DfE tables above.
- How does your delivery produce the end-point assessment portfolio: their own work throughout, or evidence assembled near the end?
- Can you run the cohort size I actually have? You have our answer above; get theirs.
- For my situation, what will I actually pay? Ask for the money, not the percentage.
FAQs: choosing a data analyst apprenticeship provider
Which is the best data analyst apprenticeship provider in the UK?
There is no single best provider; a page that names one is either selling it or has not looked closely. Every credible option delivers the same ST0118 standard, so that cannot be the tie-breaker. The best provider is the one whose shape fits your situation and whose answers on coaching, practice and cohort size hold up: a data specialist with practitioner coaches for a small cohort needing deep capability, a scale-generalist for a large multi-discipline rollout. Three years of DfE rates for every named provider sit in the tables above, and a comparison you can verify beats a ranking you have to trust.
How long is a Level 4 Data Analyst apprenticeship?
The standard's typical duration is 24 months, though a provider's real delivery can be shorter: iO-Sphere delivers ST0118 in 15 months of training plus a 3-month end-point assessment. A funding floor sits beneath all of this. An apprenticeship must run at least 8 months (reduced from 12 for starts from 1 August 2025), or cover 187 off-the-job hours, whichever the standard requires. So always check a provider's actual delivery length against the standard's typical figure rather than assuming they match.
Do you need a degree to start a data analyst apprenticeship?
No. There is no degree requirement for the Level 4 Data Analyst apprenticeship. The English and maths requirement is Level 2 (roughly a GCSE pass, grade 9 to 4), and for anyone aged 19 or over at the start it has been optional since 11 February 2025, with the apprenticeship still funded if they lack it and the employer agrees. No degree does not mean no aptitude, though: the role needs curiosity and comfort with numbers, which a good provider checks at the start so it can close gaps early rather than turn people away.
Which UK providers deliver the Data Analyst apprenticeship?
Many, including Multiverse, QA, BPP, Kaplan, Corndel, Cambridge Spark, Decoded and NowSkills, alongside iO-Sphere and others. Every approved provider appears on the government's Find apprenticeship training service, which is where you confirm a provider is cleared to deliver ST0118. Because the standard is shared, compare providers on delivery model, coaching, cohort flexibility and published outcomes rather than on curriculum, which is broadly common. For the named provider-by-provider comparison, see our alternatives to Multiverse guide.
Can I put just one or two people through a data analyst apprenticeship?
Yes, though not every provider is built for it. A single apprentice or a pair does not fit a provider that only runs closed cohorts of dozens. At iO-Sphere they join an open cohort alongside learners from other companies, with no minimum on your side; a closed cohort built around your business is where we look for 10 to 12. For a small team the open cohort is usually the stronger option: its cross-company mix gives a lone analyst peers doing the same job elsewhere to test their thinking against. Ask any enterprise provider what happens to a cohort of two.
How much does a data analyst apprenticeship cost the employer?
For starts from 1 August 2026, an apprentice aged 16 to 24 at the start carries no employer contribution at any employer type: the government funds 100% where there is no levy pot or it has run out, and a levy payer with funds simply draws on the pot. For an apprentice aged 25 or over, a non-levy employer pays 5% of the price and a levy payer whose funds have run short co-invests 25%. The training cost itself is capped by the standard's funding band, up to £15,000 for ST0118 (Skills England; bands are revised periodically). These are the 2026-27 funding rules; the fuller picture, including levy transfers, is in the funding section above.
Is a data analyst apprenticeship the same as becoming a data scientist?
No. A data analyst answers business questions with data that already exists; a data scientist builds predictive models and works at a deeper level of statistics and programming. The Level 4 Data Analyst apprenticeship is a strong foundation and a progression route towards that work, but it is not a data science qualification. From Level 4, the main funded route towards data science is the Level 6 Data Scientist (integrated degree) apprenticeship, ST0585. iO-Sphere does not deliver a data science standard, so if data science is the destination, factor that into your provider choice now rather than 18 months in.
Should we grow our own analysts or hire them in?
For the analytical layer, growing your own usually wins. Upskilling someone who already knows your business, your customers and your systems turns domain knowledge you are already paying for into data capability, and it does so faster than an external hire who has to learn the business from scratch. Hiring imports tool skills but not context, and context is the slower half to build. How we scope that with employers, cohort sizing included, is on our employers page.
If your need is deep, practitioner-coached data capability in a small or growing cohort, our Level 4 Data Analyst apprenticeship is built for exactly that. Weighing it against a Level 5 route? Our data engineer vs data analyst comparison draws the line, and Data & AI Essentials at Level 3 is the foundation if your team is not ready for Level 4 yet. When you want to test the fit against your own roles and data, see how we work with employers and book a scoping call.
Will it fit your cohort?
Tell us your cohort size and what your people need to be doing by next year, and we will say whether we are the right shape, including when we are not.