Fixed-Fee AI vs Hourly Human Tutors: A Budgeting Guide for School Leaders

School leaders are no longer just asking which tutoring model works best academically. They are asking which model gives the best value for money, which model can scale without blowing up school finance, and which model can be justified to governors with clean evidence. That is especially true after the National Tutoring Programme reshaped expectations around intervention spend and made schools far more rigorous about tutoring pricing and measurable outcomes. In practice, the debate is not simply fixed-fee AI versus hourly human tutors; it is a decision about cost per pupil, staffing capacity, safeguarding, and the probability of achieving the desired impact within a finite budget.

This guide is built for leaders who need to forecast the economics of intervention, not just compare product features. We will model annual fixed-fee AI tutoring against hourly human tutoring, unpack hidden costs, and show how to estimate return on investment per pupil using realistic school scenarios. Along the way, we will connect the budgeting decision to safeguarding, implementation, and procurement discipline, because schools that treat tutoring as a strategic purchase rather than an emergency spend usually get stronger results. For a broader view of online tuition options, see our overview of the best online tutoring websites for UK schools.

1) The Core Budget Question: What Are You Really Buying?

Access to hours, or access to outcomes?

Hourly human tutoring sells time. Fixed-fee AI tutoring usually sells capacity, consistency, and unlimited or near-unlimited access over a school year. That distinction matters because two services that look similar on a quote can produce very different usage patterns and therefore very different cost-per-pupil outcomes. A school paying £25 per hour for a tutor is buying a scarce input, while a school paying a fixed annual subscription is buying a controllable intervention engine that can run repeatedly without each additional session increasing the invoice. This is why the old instinct to compare only the headline rate can lead leaders astray.

The most useful budgeting metric is not simply hourly price or annual subscription price. It is cost per completed, high-quality learning interaction, adjusted for the number of pupils reached, the proportion who attend, and the likelihood that staff actually use the programme in the intended way. In other words, the real question is whether your budget buys delivery or impact at scale. To think more clearly about procurement trade-offs, it can help to borrow from supplier SLA thinking, where the emphasis is not merely on the contract price but on service reliability and verification.

The after-NTP context changed expectations

The end of the National Tutoring Programme matters because it shifted tutoring from a centrally funded expansion exercise into a school-level budgeting decision. During NTP, many schools experimented with tutoring at scale, but post-NTP the conversation has become more selective: schools want clear subject fit, reliable delivery, and hard evidence of improvement. That means budget holders are now comparing tutoring the way they would compare a software subscription, a contractor, or a staffing solution. The winners are often the models that provide the best mix of predictable cost and predictable implementation. This is one reason schools now scrutinise online tutoring far more carefully than they did in the early pandemic years.

As a result, leaders should build budgets around the intervention journey, not around one session at a time. Consider how the cost to onboard, monitor, and sustain tutoring compares with alternatives such as small-group catch-up, in-class support, or targeted teacher-led revision. For a broader strategic view of intervention choices, the logic is similar to choosing between an outsourced service and an in-house workflow, as explored in operate vs orchestrate decision-making. Schools that manage tutoring like a managed service usually get more consistency than schools treating it like ad hoc enrichment.

Why school leaders need scenario-based budgeting

Most schools do not have the luxury of unlimited intervention funds. A primary with two year groups needing catch-up, or a secondary juggling maths, English and science gaps, must forecast carefully or risk underfunding the pupils who need the most support. Scenario-based budgeting allows leaders to model best-case, expected-case, and stretch-case adoption. It also helps leaders answer governors’ favourite question: “What happens to cost per pupil if uptake is lower than planned?” That question is especially important for AI models, where the per-school fee may look attractive but underuse can still produce poor value for money if implementation is weak.

Pro tip: Do not budget tutoring from the top down as a lump sum. Budget from the pupil level up: intervention need × expected sessions × realistic attendance × cost per session. Then compare that with fixed-fee annual coverage.

2) The Economics of Fixed-Fee AI Tutoring

How fixed-fee AI changes the cost curve

Fixed-fee AI tutoring is appealing because it decouples usage from marginal cost. If a school pays one annual price, additional sessions typically do not increase spend, which means the cost per pupil falls as utilisation rises. That is a major advantage for schools with large cohorts, recurring gaps, or subjects that need frequent practice. Third Space Learning’s AI maths tutor, Skye, is a useful example of this model: it offers unlimited one-to-one maths tutoring for primary and secondary schools at a fixed annual price starting from £3,500. For a school with many pupils needing regular maths support, that can radically change the economics compared with hourly tuition.

Fixed-fee models also support planning certainty. In school finance, predictable costs can be more valuable than apparently cheaper but volatile hourly arrangements, especially if budgets are set months ahead. This is similar to how leaders compare high fixed costs with variable costs in other sectors: the school pays for capacity upfront, then tries to maximize utilisation. If you are deciding when a fixed annual model makes sense, a budgeting mindset similar to scalable template thinking is useful, because the best fixed-fee tools are the ones you can deploy repeatedly without reinventing the process for every cohort.

Where AI tends to generate the strongest ROI

AI tutoring often performs best when the need is frequent, narrow, and repetitive. Maths fluency, retrieval practice, worked example rehearsal, and exam technique are all areas where a structured AI system can provide consistent delivery at scale. That does not mean AI replaces teachers. Rather, it frees staff from the impossible task of giving every pupil a live tutor-like experience all year. Schools using AI well usually pair the technology with teacher oversight, progress checks, and a clear timetable for intervention rather than leaving pupils to work in isolation.

The ROI case becomes especially strong when the school can support steady usage across several year groups. A fixed fee spread across 80, 150, or 300 pupils can produce a dramatically lower cost per pupil than a half-used hourly package. The same logic applies in procurement and distribution: the more consistently a service is adopted, the better the unit economics. That principle is echoed in diffusion and cluster logic, where concentrated demand makes expansion more efficient.

Hidden costs schools should still budget for

AI is not cost-free just because it is fixed-fee. Schools should account for implementation time, staff onboarding, device access, timetable integration, and monitoring. If the programme requires a coordinator to chase logins, analyse reports, or troubleshoot access issues, those hours should be treated as real costs. There may also be indirect costs if the AI programme needs a specific digital infrastructure, privacy review, or parent communications. A good contract may reduce these frictions, but it will not eliminate them.

There is also an opportunity cost. If a fixed-fee AI package replaces a stronger but more targeted human intervention, then the cheapest model on paper may not deliver the best outcome. School leaders should compare AI not only with tutors, but with other improvement strategies competing for the same budget. A useful parallel is integration risk management: every added tool carries adoption and governance overhead, so the cheapest-looking option can become expensive if it creates admin burden or fragmented usage.

3) The Economics of Hourly Human Tutors

Why hourly tutoring still matters

Hourly human tutors remain the strongest option in several scenarios: exam-critical subjects, pupils with complex misconceptions, older students needing advanced content, and cases where relationship-building or adaptive explanation matters more than repetition. If a pupil is stuck on a specific topic, a skilled human can diagnose the barrier, pivot instantly, and provide the kind of encouragement that software can struggle to replicate. For many schools, this is why hourly tutoring still commands a place in the intervention mix, especially for GCSE and A level support. Providers such as MyTutor, Fleet Tutors, Tutorful, Spires, Tutor House, and First Tutors all serve different needs across the market, from one-off matching to fully managed school partnerships.

But hourly tutoring has a scaling problem. The school pays not just for teaching, but for the time spent waiting, matching, rescheduling, and sometimes dealing with cancellations. If the tutor rate is £26 to £37 per hour and the school wants two sessions a week across several months, the invoice rises quickly. That may be manageable for a small cohort, but the model becomes challenging when intervention needs widen across a year group or across multiple subjects. The most effective schools therefore use hourly tutoring selectively, not universally.

What gets missed in simple hourly comparisons

When leaders compare hourly tutoring quotes, they often underestimate the true cost per successful hour. If a tutor cancels, arrives late, spends the first session on baseline assessment, or requires repeated admin follow-up, the cost of each effective learning hour rises. There is also the cost of coordination: timetabling sessions, confirming safeguarding, chasing attendance, and aligning the tutor to the school’s curriculum sequence. For a school office already under pressure, these “soft costs” can be substantial even when they never appear on a purchase order.

Human tutoring also tends to be more variable in quality than a well-designed fixed-fee system. The best tutors are excellent; the weakest can be merely compliant. If a provider’s vetting is light or supervision is weak, the school may get patchy impact despite paying a premium. For practical reference, the tutor marketplace articles in our library such as online tutoring websites for UK schools underline that vetting, DBS checks, and reporting standards should be part of the price comparison, not an afterthought.

When hourly tutoring produces the best value

Hourly tutors are often most cost-effective when the intervention is narrow, high-stakes, and time-bound. Think of a Year 11 pupil needing eight sessions to close a specific gap before exams, or a small cohort needing a short burst of A level support. In those cases, the school can buy exactly the number of hours required and avoid paying for unused annual capacity. That creates a strong short-term ROI if the sessions are well targeted and attendance is strong.

Schools should also use hourly tutoring when individualisation matters more than volume. Pupils with atypical learning profiles, advanced subject needs, or highly specific misconceptions may benefit from a human specialist who can work outside a scripted path. This is where the decision starts to resemble choosing a specialist agency: the best fit is not always the cheapest line item, but the one most likely to solve the right problem quickly and well.

4) Side-by-Side Budget Comparison: Fixed-Fee AI vs Hourly Human Tutors

Illustrative annual scenarios

The table below uses simple assumptions to show how cost per pupil can vary depending on usage. The numbers are illustrative, not universal, because actual pricing will differ by provider, subject, and service level. What matters is the shape of the economics. A fixed-fee AI package can become dramatically cheaper per pupil as utilisation rises, while hourly tuition remains tightly linked to the number of sessions purchased. That makes the best option depend on scale, subject concentration, and the frequency of support needed.

At first glance, the fixed-fee AI model appears unbeatable on cost per pupil once you can deploy it at scale. But schools need to remember that the numerator and denominator both matter. If the AI package reaches 100 pupils but only 30 use it meaningfully, the effective cost per engaged pupil rises sharply. By contrast, if a human tutor is booked for just the pupils who truly need them, the school may pay more per hour but less for waste. That is why leaders should model both capacity cost and active use cost.

ROI calculations school leaders can actually use

A simple ROI formula for tutoring can be stated as: ROI = expected impact value / total intervention cost. Schools rarely assign a cash value to attainment gains, so a more practical version is to compare cost against expected outcome improvements, such as pass-rate gains, reduced re-sit risk, or narrowing of attainment gaps. If a £3,500 AI package helps 20 borderline pupils improve by one grade equivalent, the value may far exceed the spend. If the same package is underused, the apparent bargain disappears. For leaders wanting more on evidence and measurement, a disciplined approach similar to reading beyond the headline helps avoid overclaiming results from limited data.

For hourly tutors, ROI should include the probability of hit rate. A tutor charging £30 per hour who delivers 20 high-quality hours to a pupil on the cusp of a grade boundary may produce better ROI than a cheaper tutor who needs extra sessions because the fit is weaker. Schools should estimate the uplift they reasonably expect, then divide by the total cost of sessions, coordination, and any provider management fees. The result is not perfect, but it is far better than comparing price alone. That discipline is similar to how leaders compare low-cost versus premium offerings: what looks expensive may still deliver better value if it solves the problem faster.

What a break-even point looks like

The break-even point is the number of tutoring hours or pupils at which a fixed-fee AI package becomes cheaper than hourly tutoring. For example, if the AI model costs £3,500 and the hourly tutor costs £25, then the AI model equals 140 hours of tutoring at the simple price level alone. If you factor in admin time, cancellations, and coordination overhead, the break-even may arrive sooner. Schools with high-volume maths need will often cross this threshold quickly, especially if they can use the tool across year groups or embed it into the curriculum. Schools with low-volume, high-stakes needs may never cross it, and that is perfectly rational.

Budgeting is therefore not about finding the universal winner. It is about finding the right fit for your volume, subject mix, and implementation capacity. The same logic appears in scaling product lines: a model that works beautifully at one scale can become inefficient at another. Leaders who think in terms of break-even points and utilisation rates are far less likely to overspend.

5) Hidden Costs That Can Distort the Budget

Implementation and staff time

The most common hidden cost in tutoring is staff time. Even a great programme requires onboarding, scheduling, communication, reporting, and review. If a deputy head, SENDCo, or intervention lead spends several hours a month keeping tutoring on track, those hours represent real budget pressure. In some schools, this internal labour is the difference between a programme that scales and one that stalls. The cheapest external quote can become the most expensive total solution if it demands too much internal management.

Another overlooked cost is the time spent aligning tutoring with classroom teaching. If a tutor works from a generic sequence that does not match the school’s scheme of work, teachers may need to re-teach, clarify, or adjust. That creates friction and weakens ROI. In effect, the school pays twice: once for the intervention and once for the correction. A model that integrates better with curriculum planning is often better value even if the quote is higher.

Safeguarding, compliance, and data privacy

Schools cannot budget as though safeguarding is optional. Any tutor model must be checked for DBS status, online safety controls, session recording or note-taking, data processing arrangements, and school DSL escalation routes where relevant. With AI tutoring, data privacy and content moderation matter just as much as tutor vetting. With human tutors, identity verification, supervision, and professional conduct are the key concerns. Either way, compliance should be written into the procurement brief before the price conversation starts.

It can help to think like a risk manager. Schools should ask what happens if a tutor is unavailable, if a pupil disengages, if a platform records insufficient evidence, or if a safeguarding concern emerges mid-programme. That is why articles like automating supplier verification and risk from third-party integrations are relevant even outside their original sectors: the principle is the same. A low-friction service is useful only if it remains trustworthy under real operational conditions.

Underuse is a hidden cost too

Fixed-fee AI models can suffer from the opposite problem: the school pays for capacity it never fully uses. If only one year group adopts the tool, if timetable barriers limit access, or if staff do not prioritise the intervention, then the cost per used session rises quickly. This is why leaders should never judge a fixed-fee model solely on theoretical capacity. The best question is how many pupils will actually complete enough sessions to shift outcomes. Underuse is not just a technical issue; it is a budget leak.

Schools can reduce underuse with a very specific implementation plan: nominate an owner, set weekly usage targets, build prompts into the timetable, and review engagement every half-term. If the platform supports reporting, use it. If it does not, ask whether the absence of reporting weakens your ability to prove value for money. Reliable performance data is the tutoring equivalent of a strong operational dashboard, much like the logic behind reliability-first decision-making in tight markets.

6) Templates for Forecasting Impact Per Pupil

Template 1: simple cost-per-pupil calculator

Start with three variables: total annual tutoring budget, expected number of pupils served, and expected sessions per pupil. Divide the budget by the number of pupils to get cost per pupil, then divide again by expected sessions to estimate cost per session per pupil. For example, a £3,500 fixed-fee AI package serving 100 pupils works out to £35 per pupil. If each pupil completes 10 sessions, the implied cost per session is £3.50 before internal staffing costs. That is a powerful way to communicate affordability to senior leaders and governors.

For hourly tutoring, multiply the hourly rate by the number of hours each pupil will receive. A 10-hour package at £30 per hour costs £300 per pupil before admin costs. If the school adds coordination time, cancellation risk, and reporting, the true figure can be higher. This calculation makes the contrast clear: hourly tutoring is more linear and more expensive per pupil, while fixed-fee AI becomes more efficient as adoption rises. Schools that use both models can budget separately by need profile rather than forcing one solution onto every pupil.

Template 2: expected impact grid

Create a grid with four columns: intervention type, target cohort, expected usage, and expected outcome. For example, “AI maths tutoring,” “Year 5 and Year 8 catch-up,” “weekly use for 12 weeks,” and “improve arithmetic fluency and confidence.” Then add a fifth column for evidence strength, such as strong, moderate, or emerging. This helps leaders compare interventions not just on cost but on probability of impact. In practice, a school can rank options by cost, scale, and confidence level rather than relying on instinct.

This is also where schools should think about the implementation shape of the programme. A highly scalable AI model may be ideal for universal catch-up, while an hourly tutor may be ideal for top-end GCSE intervention. The key is not to choose one forever; it is to align the right model to the right objective. If you need a practical lens on readiness and adoption, the logic is similar to a legal and ethical boundary check: good planning protects both the institution and the end users.

Template 3: governor-ready business case

When presenting tutoring spend to governors, use a one-page structure: problem, chosen intervention, cost, expected reach, expected impact, risks, and review date. Include a simple comparison line showing what the same money would buy in each model. For example, “£3,500 buys unlimited AI maths support for a year versus roughly 116 hours at £30/hour.” Then note the likely pupil reach and whether the use case is whole-cohort, targeted, or exam-specific. Governors respond well to clarity and caution, especially where outcomes are uncertain.

A strong business case also sets triggers for review. If usage falls below a threshold, the school should either change implementation or stop the programme. That discipline protects budgets and reduces sunk-cost bias. For schools balancing multiple choices, a structured decision process borrowed from change management is often more effective than a purely intuitive one.

7) Choosing the Right Model by School Scenario

Primary school with broad maths gaps

A primary school with widespread maths catch-up needs usually benefits most from fixed-fee AI, especially if the school can spread usage across several classes. The reason is simple: many pupils need frequent, structured repetition, and a fixed annual model can provide that without the cost ballooning every time a new pupil joins. If the school is already using online interventions effectively, this model can offer strong value for money and a lower cost per pupil than hourly support. The key is to ensure teachers integrate the tool into weekly routines rather than treating it as optional extra practice.

In this context, a human tutor may still be valuable for small groups with severe misconceptions, but the core budget often belongs with the scalable model. A school leader looking for a broad comparison of options can revisit our coverage of online tutoring platforms to see how pricing and support differ across providers. That helps avoid overbuying live time when the intervention need is repetitive and wide.

Secondary school with GCSE pinch points

Secondary schools often need a mixed strategy. AI can be excellent for lower- and mid-attaining pupils who need sustained practice, while hourly tutors may be better for pupils on the grade boundary or those taking exam subjects with highly specific gaps. A blended budget can protect value for money while preserving flexibility where it matters most. Schools should not feel forced to choose one model for all pupils if the need profile is mixed.

In GCSE years, weekly attendance, homework completion, and progress checks become crucial. If a human tutor is used, the school should make sure the sessions are tightly aligned to exam specifications and that the provider offers strong reporting. That is where premium hourly providers may justify their rate, especially if the school prioritises specialist subject knowledge and close progress tracking. For a view on choosing and comparing tutoring services, see how to build a scorecard-style selection process and adapt the same discipline to tutoring procurement.

Small school with limited admin capacity

Small schools often underestimate the admin burden of hourly tutoring. Even if the sessions are targeted, the coordination work can fall on a headteacher, office manager, or intervention lead who already has too many responsibilities. In those settings, a fixed-fee AI model can offer a cleaner operational shape because it is easier to standardise and less dependent on endless tutor matching. The trade-off is that staff still need to monitor usage and quality, or underuse will erode the benefit.

If the school has only a handful of pupils needing intensive support, a short burst of hourly tutoring may still be the best value. But if the need is recurring across multiple cohorts, the administrative simplicity of the fixed model becomes part of the ROI. In budget conversations, convenience is not a luxury; it is often a driver of effective delivery.

8) How to Present the Decision to Governors and Trust Leaders

Frame the decision around uncertainty, not just cost

Good boards do not ask only “What is cheapest?” They ask “What is most likely to work, at scale, within our operational constraints?” That means the budget paper should discuss confidence levels, implementation risks, and review points. It should also explain why one model is better suited to the school’s subject mix and staffing capacity. A nuanced presentation earns more trust than a simple bargain-hunting pitch.

Where possible, use evidence from actual usage data, not just vendor promises. If you have pilot data from an AI programme or success rates from hourly tutoring, show it clearly. Where evidence is still emerging, be transparent about that. In procurement terms, credibility often comes from restraint. This is consistent with the logic in data interpretation guidance: the strongest story is the one that acknowledges uncertainty and still makes a rational recommendation.

Use a phased procurement approach

Instead of committing all intervention money at once, many schools will get better results by phasing the budget. A smaller pilot can test the AI model’s adoption rate, while a limited hourly package can support the pupils with the most urgent needs. After one term, leaders can compare attendance, usage, and early attainment signals, then scale what appears to be working. This approach reduces risk and helps avoid locking into a single model before the school knows how it behaves in practice.

A phased plan also makes financial sense when budgets are tight. You protect flexibility while learning from real pupil behaviour rather than theoretical assumptions. That is especially helpful in a climate where schools are being asked to demonstrate efficient use of funds across a crowded improvement agenda. Governance becomes easier when the school can show a deliberate test-and-learn model with clear exit criteria.

Build in a review cycle

Set a review point every half-term or term. Examine spend, usage, attendance, staff feedback, and early pupil progress. If the cost per pupil is rising because usage is dropping, intervene early. If a human tutoring model is working well for a small cohort but not for wider catch-up, narrow its remit and redirect the rest of the budget. The aim is not to defend the original decision at all costs; it is to spend intelligently.

To support that habit, schools can use a simple dashboard inspired by reliability-focused decision-making: planned spend, actual spend, planned usage, actual usage, planned outcome, actual outcome. That keeps tutoring out of the realm of vague impression and into the world of accountable school finance.

9) Practical Recommendations: A Budgeting Playbook for 2026

Use fixed-fee AI when scale is the problem

If your core challenge is reaching many pupils repeatedly and affordably, fixed-fee AI tutoring will usually offer the best economics. It is especially compelling in maths, where frequent practice can be systematised. The fixed annual price gives schools certainty, and the ability to spread use across cohorts can produce very low cost per pupil. For leaders focused on budget control, that predictability is a major advantage.

However, the school should only choose this route if it can implement well. Appoint an owner, set adoption targets, and monitor usage from week one. If you cannot do that, the model may be cheaper on paper but worse in practice. AI is not a substitute for leadership.

Use hourly human tutors when precision is the problem

If your core challenge is a small group of pupils who need highly tailored support, hourly tutors remain valuable. They can adapt, explain, and respond in ways that many digital tools still cannot match. They are especially useful when the need is urgent, short-term, and high-stakes. The cost may be higher per hour, but the value may also be higher if the tutoring is the right intervention at the right moment.

That said, be disciplined. Demand strong vetting, clear reporting, and curriculum alignment. Hourly support without those safeguards can quickly become expensive noise. This is the point at which schools should insist on the same rigor they would expect from any contractor supplying services into a regulated environment.

Blend both when your pupil needs are mixed

For many schools, the best answer is not either/or. A blended model can combine fixed-fee AI for routine, broad-based support with hourly human tutors for high-stakes exam intervention or complex cases. That approach gives leaders both control and flexibility. It also reduces the risk of using expensive human time where a scalable system would do the job more efficiently.

As a practical rule, start with the cheapest model that can plausibly deliver the required outcome, then add human expertise where the marginal value is clearly higher. That is the same principle many operators use when they decide whether to automate or escalate a task. The point is not ideological purity; it is getting the best educational return for the budget available.

10) Conclusion: Value for Money Means Match, Scale, and Proof

The best model is the one that fits the need profile

There is no universal winner in the fixed-fee AI versus hourly human tutor debate. The better model depends on whether your school needs scale, precision, specialist subject knowledge, or a combination of all three. Fixed-fee AI tends to win on predictability and cost per pupil when usage is high. Hourly human tutors tend to win on flexibility and deep individualisation when the cohort is small and the stakes are high. Leaders who understand both economics can make smarter decisions than those chasing the lowest sticker price.

Budgeting should be evidence-led, not vendor-led

The strongest school finance decisions come from clear assumptions, honest usage forecasts, and a willingness to review and pivot. Use the templates in this guide, compare your likely cost per pupil, and stress-test what happens if uptake is lower than expected. If you can show governors that your choice is tied to attainment goals, safeguarding, and implementation reality, you will have a much stronger case. That is how tutoring spend becomes strategic rather than reactive.

Next steps for school leaders

Before you renew or procure tutoring, map your need profile by year group and subject, calculate cost per pupil, and decide where the extra marginal value of a human tutor is truly worth the premium. Then compare that against the operational simplicity and scalability of a fixed-fee AI model. For a wider look at current market options, revisit our guide to the best online tutoring websites for UK schools. If you want to benchmark procurement rigor across other service decisions, our articles on verification and SLAs and operating versus orchestrating services can help sharpen your decision framework.

Fixed-fee AI is usually better value when you need to serve many pupils repeatedly, especially in a subject like maths where practice can be standardised. It is also attractive when schools need budget certainty and want to avoid cost increases every time usage rises. The key is ensuring enough adoption to spread the fixed cost across enough pupils.

2) When are hourly human tutors the smarter choice?

Hourly human tutors are often smarter for small groups, exam-critical support, and cases where individual diagnosis matters more than scale. They can be especially effective for pupils with specific misconceptions, complex needs, or advanced content requirements. If the intervention is narrow and short-term, hourly tutoring can be very efficient.

3) What hidden costs should schools include in tutoring budgets?

Schools should budget for staff time, onboarding, timetabling, monitoring, reporting, safeguarding checks, and possible cancellations or underuse. With AI, implementation and adoption are the most common hidden costs. With human tutoring, coordination and variability are the biggest hidden costs.

4) How do schools calculate cost per pupil?

Divide the total annual tutoring cost by the number of pupils who will meaningfully use the intervention. Then, if useful, divide again by the expected number of sessions per pupil to estimate cost per session. This gives a more realistic picture than comparing hourly rates alone.

5) Can schools combine AI tutoring and human tutors?

Yes. In fact, a blended model is often the best option. AI can provide scalable, routine practice for larger groups, while human tutors can focus on high-stakes or highly specialised support. This usually offers the best balance of value for money and educational precision.

6) How should governors evaluate tutoring spend?

Governors should ask about expected reach, cost per pupil, implementation risk, safeguarding, and review points. They should also ask what evidence will be used to judge success and what the school will do if the programme is underperforming. A good governance paper should show both the financial logic and the educational rationale.

Related Reading

• 7 Best Online Tutoring Websites For UK Schools: 2026 - Compare current school-focused tutoring platforms and pricing models.
• Automating supplier SLAs and third-party verification with signed workflows - A useful lens for procurement, verification, and service reliability.
• Operate vs Orchestrate - Learn how to decide between running services directly or managing them strategically.
• Turn CRO learnings into scalable content templates that rank and convert - Useful for thinking about repeatable, scalable intervention design.
• Reading Beyond the Headline - A practical guide to interpreting claims and evidence carefully.