Alternatives to QA for Data and AI Apprenticeships: Do You Need a Big Generalist to Be Safe?

Part of our topic guides on Data & AI Apprenticeships and Data Analyst Careers.

By James Cotton, Founder, iO-Sphere

QA in its own words

Start with QA's own pages. On its about page QA calls itself "Europe's biggest and North America's fastest growing technology and digital training partner with 40 years experience and ~7000 organisations served." Its apprenticeship hub leads with "40k+" careers accelerated and "3,500+" partners supported, with no basis or period given for either.

QA sells its apprenticeships under its own programme names. The data hub lists five, from Data Essentials Level 3, "Aligned to the Data Technician Level 3 standard", to Databricks Engineer Level 5. The AI hub lists four more apprenticeships, two of them Level 6 AI Engineer programmes, and an "AI Leader Apprenticeship Units" card at Level 5.

Each programme page has a "Programme Details" box giving the standard behind it, its length, its live instructor days, and the split between modules and the end-point assessment (EPA), the assessment taken once training is complete.

QA programme (standard, level)	Programme duration; live instructor sessions	Modules and assessment, as displayed
Data Essentials Level 3 (Data Technician, 3)	17 months; 16 days	Modules 14 months; EPA 3 months
Data Analyst Level 4 (Data Analyst, 4)	18 months; 24 days	Modules 15 months; EPA 3 months
Business Analyst Level 4 (Business Analyst, 4)	18 months; 14 days	Modules 15 months; EPA 3 months
Data Engineer Level 5 (Data Engineer, 5)	21 months; 23 days	Modules 17 months; EPA 4 months
Databricks Engineer Level 5 (Data Engineer, 5)	21 months; 23 days	Modules 17 months; EPA 3 months
Digital and AI Support Level 3 (Digital Support Technician, 3)	16 months; 17 days	Modules 13 months; EPA 3 months
AI & Automation Specialist Level 4 (AI & Automation Practitioner, 4)	13 months; 16 days	Modules 12 months; "Work-based project" plus "One method from: Presentation, Simulated Task, Professional Discussion, or Portfolio"
AI Engineer Level 6 (Machine Learning Engineer, 6)	23 months; 28 days	Modules 19 months; EPA 4 months
NVIDIA AI Engineer Level 6 (Machine Learning Engineer, 6)	23 months; 25 days	Modules 19 months; EPA 4 months

On delivery, the Data Analyst Level 4 page says the programme "integrates live and online workshops with self-paced learning, employing a guided discovery approach for individual learner contexts." Its method sits under the heading "Discover, Practise, Apply", and each learner is assigned a Digital Learning Consultant.

On time, the data hub says apprentices "spend around 80% of their time in their day-to-day role", and QA's employer page gives "~20% of time learning (off-the-job training)", off-the-job training being learning in paid working hours that is not the person's normal day-to-day work. The same page says programmes typically last 12 to 24 months, depending on level.

QA's own pages do not agree on every figure. The data hub names training centres in "London, Birmingham, and Scotland", the AI hub in "London, Manchester and Scotland". The AI & Automation Specialist card on the AI hub says 12 months where its own page says 13, and Databricks Engineer shows a 3-month EPA where Data Engineer, on the same standard, shows 4.

How big QA is by the DfE's count

The "biggest" in QA's description refers to technology and digital training overall. The Department for Education's achievement rate tables count leavers: the people who finished an apprenticeship or left it during the year. The DfE's provider data for 2024/25 holds 1,203 provider rows; measured that way, the ten with the most leavers across all standards are these.

Lifetime Training Group 14,960; Multiverse Group 6,340; BPP Professional Education 5,810; Corndel 5,520; British Army 5,510; Paragon Education & Skills 5,200; Kaplan Financial 5,120; The Opportunity Provider 4,280; QA 4,260; Realise Learning and Employment 3,750.

On the data standards the order changes. On Data Analyst (ST0118), Multiverse had 2,770 of 5,620 leavers nationally, about 49%, QA 770, about 13.7%, and Cambridge Spark 530. On Data Technician (ST0795), Multiverse had 2,600 of 5,030, about 51.7%, Corndel 640 and QA 350. On Business Analyst (ST0117), Multiverse had 760 of 1,470, QA 200 and Assist Knowledge Development 70.

A third-party list, Turing College's "Top Level 4 Data Analyst Apprenticeships", updated 4 June 2026, says "QA is one of the UK's largest apprenticeship providers". On the DfE's measure, that is 4,260 leavers across all standards and 770 on Data Analyst in 2024/25.

A leaver count says how many people went through a provider in a year. The standard, the government's maximum for it and the assessment at the end were the same whichever provider they chose.

What every provider shares

Every apprenticeship trains to a standard, the approved description of an occupation's duties, and each standard has one assessment plan that every provider works to. Find apprenticeship training (FAT), the DfE's search service, lists 130 training providers for Data analyst (level 4). QA's own entry there reads "QA LIMITED is 1 of 130 training providers for Data analyst (level 4)".

The same service lists 113 providers for Data technician (level 3), 79 for Business analyst (level 4), 18 for Data engineer (level 5), 10 for Data protection and information governance practitioner (level 4) and 146 for AI and automation practitioner (level 4).

The funding band is the government's maximum for a standard, not a provider's price. On FAT the band is £13,000 on Data Technician (ST0795), £15,000 on Data Analyst (ST0118), £18,000 on Business Analyst (ST0117) and on AI and Automation Practitioner (ST1512), £19,000 on Data Engineer (ST1386) and £10,000 on Data Protection and Information Governance Practitioner (ST0967).

QA's programme pages show the same figures as "Total cost", each adding that "the maximum funding for this programme is" the band. For its two Level 6 programmes on the Machine Learning Engineer standard, QA's pages give £22,000.

Who pays what share of a band depends on the apprentice's age and on whether the employer pays the apprenticeship levy (it does if its annual pay bill is over £3 million). The funding rules set that share, so it is the same at every provider. The split for a small business and the split for existing employees both follow from those two facts.

Results are counted the same way for everyone, too: the DfE's achievement rate tables give each provider a row per standard, beside a national row of the same shape.

The shape test

Inside those fixed terms, each provider chooses how the programme runs. Four questions separate one from another. Each is answered below for QA from its own pages where they answer it, and for five other Data Analyst providers in their own words.

1. Who coaches your people, and what did they do before they taught?

QA's Data Analyst Level 4 page says "Learners are assigned a Digital Learning Consultant (DLC) for personalised coaching and support. These specialists ensure their successful progress, well being, and readiness for assessments." What a consultant did before coaching is the half of the question to put to QA.

Multiverse says "Learners receive 1-to-1 coaching from industry experts, regular group coaching and community collaboration." BPP says "Fully-qualified Performance Coaches and industry-leading tutors bring their firsthand expertise to your studies". Corndel names "Our signature one-to-one coaching".

Cambridge Spark says "We provide each learner with a dedicated Data Mentor and Learner Success Coach." Baltic Apprenticeships says "each learner has a designated coach that will support them throughout the entirety of their apprenticeship."

Every one names a coach. Ask each what that person did before coaching, how many learners they carry, and whether they stay for the whole programme.

2. What do your people practise on?

QA's three steps answer this in its own words. Discover: "Leveraging QA's learning platform, learners follow a development path focused on their job role." Practise: learners come together for instructor-led sessions, "practising their skills through hands-on labs and sandboxes in a safe environment." Apply: "These learnings are applied through work-based activities at key and sequenced stages, supported by Digital Learning Consultants."

Multiverse's weekly model includes "Structured and unstructured application of learning to apprentices' roles". Cambridge Spark says its EDUKATE.AI platform offers "real practice on real projects with immediate, personalised feedback." Baltic pairs a "Self-directed learning phase using our Baltic360 platform" with a two-day stage of "hands-on workshops and real-world projects".

Labs, sandboxes, real projects: the words overlap. Ask whose data and whose problem the assessed project is built on, and whether the work submitted is work your organisation needed done.

3. How many instructor days does the programme hold?

QA gives a count: 24 live instructor days on Data Analyst Level 4, 14 on Business Analyst. Where they happen is reported three ways. QA's hubs name centres in different cities; Ofsted's June 2025 report said "All training is online"; and QA's FAT listing for Data analyst shows "At learner's workplace" plus block and day release at International House, St Katherine's Way, London.

The others count in other units. Multiverse's own weekly model is about 3 to 4 hours of structured learning (40%), including live workshops and group coaching, and about 4 to 5 hours applying it in role (60%), "Approx 7-8 hours per week" in all. BPP's page offers "online live or face-to-face learning, either weekly, monthly or intensive blocks".

Corndel lists "17 Masterclasses delivered across a 12-month period by academics and industry experts from Imperial Executive Education", and says each session "is available live and on-demand". Baltic's page describes each of its four core modules as a one-day instructor-led session, a self-directed phase and a two-day instructor-led stage. Cambridge Spark says "We deliver all our programmes online".

Hours a week and days a programme are different units; convert before comparing. Ask for the days of live teaching on your standard, how many are in a room, and what happens when one is missed.

4. What happens to a cohort of two?

QA's data hub says "Our data programmes are designed for organisations who want to build data skills at scale across a variety of teams and functions." Its Data Analyst page says timelines are not set in stone and adds "Talk to our experts to find out more." How two people start is the question to put to QA.

BPP gives start dates as "Various depending on delivery option", and Cambridge Spark says "We tailor our delivery to your workforce needs". Ask each provider whether two people start on a fixed date beside learners from other employers, wait for a group to fill, or are taught alone, and who shares their live sessions.

How to read any provider's record

The DfE's achievement rate tables, in a release published 16 July 2026, show each provider's results by standard. The release classes them as official statistics, without independent review by the Office for Statistics Regulation. The 2024/25 figures are as first published in March 2026; earlier years stand as originally published.

Each row gives leavers, completers and achievers. Retention is completers as a share of leavers; pass is achievers as a share of completers; achievement is achievers as a share of leavers. Counts are rounded to the nearest 10 and "low" or "x" marks a suppressed cohort, so use the published rate; never recompute it.

Achievement blends two outcomes: people who left before completing, and people who completed and did not pass. Retention measures the first and pass the second, so read them apart. The 2024/25 national rows in the DfE's subject data show where the spread lies on the data standards.

Data Analyst (ST0118): 5,620 leavers; retention 59.2%, pass 98.3%, achievement 58.2%.
Data Technician (ST0795): 5,030 leavers; retention 61.5%, pass 98.2%, achievement 60.4%.
Business Analyst (ST0117): 1,470 leavers; retention 57.9%, pass 96.3%, achievement 55.8%.
Artificial Intelligence (AI) Data Specialist (ST0763, Level 7): 370 leavers; retention 69.5%, pass 96.5%, achievement 67.0%.

On every one, pass is above 96% and retention below 70%. Set a provider's retention and pass beside the national figures for the same standard and year, with its leaver count beside both. A base under 100 leavers moves a long way from year to year, and an all-standards rate mixes every subject a provider runs.

FAT adds two more things. Each course listing prints the provider's achievement rate with its counts and the national rate beside it. Reviews give an average rating from employers and one from apprentices, each with its number of reviews, combined across every course in the last full academic year, so the rating is provider-wide.

A second block of reviews, labelled from 1 August to today, lists items chosen as a strength or for improvement, each with the number who selected it. "Working with small numbers of apprentices" is the item closest to the cohort question; "Reporting on progress of apprentices" and "Adapting to my needs" are others. Read the count before the percentage.

FAT updates the main star rating "when we have reviews from a full academic year". QA's listing for Data analyst carries every one of these fields.

The field on the same standards

The DfE file names who else trains on each data standard. Below are 2024/25 leaver counts for a selection of the providers with a row; read any of their rates with the method above. One appears under another name on the Apprenticeship Provider and Assessment Register (APAR), the list of organisations eligible for government apprenticeship funding.

Data Analyst (ST0118), 58 providers with a row, 5,620 leavers: Multiverse Group 2,770; QA 770; Cambridge Spark 530; Corndel 350; Decoded 210; BPP Professional Education 110; Firebrand Training 110; Baltic Training Services 100; Estio 100; Just IT Training (now listed on APAR as Leep Talent) 60; Apprentify 40; Babington 40; Kaplan Financial 30.
Data Technician (ST0795), 42 providers, 5,030 leavers: Multiverse Group 2,600; Corndel 640; QA 350; Royal Navy 270; Cambridge Spark 240; BPP Professional Education 190; Baltic Training Services 130; Decoded 90; Just IT Training (Leep Talent on APAR) 80; Kaplan Financial 40; Firebrand Training 30; Apprentify 10.
Business Analyst (ST0117), 28 providers, 1,470 leavers: Multiverse Group 760; QA 200; Assist Knowledge Development 70; Firebrand Training 60; Cambridge Spark 50; BPP Professional Education 40; Just IT Training (Leep Talent on APAR) 30; Apprentify 20; Kaplan Financial 20. Corndel and Baltic have no row on this standard.

Across all standards in 2024/25, Cambridge Spark and Baltic each had 1,110 leavers, Apprentify 670, Firebrand 590, Just IT Training (Leep Talent on APAR) 460 and Decoded 290, against 299,510 nationally.

QA, Multiverse, BPP, Corndel, Cambridge Spark, Kaplan, Baltic, Firebrand, Apprentify and Leep Talent each appear on APAR as a "Main provider"; QA's row gives a start date of 13 March 2017. A listing records eligibility for funding, a separate matter from quality.

Which rung you need

The provider question comes second. The standard comes first: it follows the work the person does, and it fixes the level, on the data ladder and among the AI standards alike. Where the choice is Business Analyst or Data Analyst, the person's duties settle it.

iO-Sphere is a specialist in data and AI at Levels 3 to 5, from Data & AI Essentials (Level 3) to Data Engineering (Level 5), with every programme built with AI in it and coaching run by practitioners. The project each person practises on is built on their own work.

Coaching sessions are weekly and online; delivery is 15 months of training, then a 3-month end-point assessment. One or two people from a firm join an open cohort alongside peers from other companies, and the number we look for in a closed cohort is 10 to 12.

Once the standard is fixed, the shortlist is whoever is listed on it. On Level 4 Data Analyst, put the four questions to QA and the other Data Analyst providers, then read each one's retention and pass against the national 59.2% and 98.3%. The same four questions work for short training outside an apprenticeship, bought from commercial training providers.

Frequently asked questions

Why does the register show a different company name for a provider?

A provider is filed under its UK provider reference number (UKPRN), and the name against that number can differ from one register to another. UKPRN 10034940 is "JUST IT TRAINING LIMITED" in the DfE file and "LEEP TALENT LIMITED" on APAR; Leep Trade Limited is a separate APAR row. BPP's results sit under three UKPRNs and Corndel has two APAR rows, so search by UKPRN.