Online Tutoring vs In-Person Tutoring for GCSE

The honest answer is that the UK evidence on format is thinner than either side of this argument pretends. The EEF Toolkit puts one to one tuition at +5 months on average — but +4 at secondary age, and +2 in maths. Ofsted saw its weakest tuition delivered remotely. And most learners in the EEF's own online tuition pilot said they would choose in person. Here is what that actually means for a Year 11.

Who is actually making this decision?

Almost nobody weighs online against in-person tutoring in the abstract. They weigh it in the autumn of Year 11, a fortnight after a set of mock results. The Sutton Trust's Private Tutoring 2026 report found that 29% of secondary school students in England and Wales have had private tutoring at some point, up from 18% when the survey series began twenty years ago and up slightly from 27% in 2019. The interesting figure is not the total, though. It is where the total clusters. Year 11 is the most likely year for pupils to have had tutoring at 25%, followed by Year 10 at 10% — two and a half times the rate one year further down (25 divided by 10 is 2.5).

That matters because it tells you what the format question really is. It is not a question about learning styles or about screens in general. It is a question about how to run a limited number of hours, in a compressed window, against a named specification, for a student who already has a school timetable, a revision schedule and a part-time job on Saturdays. Everything below should be read in that frame. If your child is in Year 9 and the plan is a slow build over two years, several of the trade-offs here relax considerably.

The two formats are easy to define and worth defining concretely, because a lot of the disagreement about them is really a disagreement about what people are picturing. In-person tutoring means a tutor in your house or at a centre, sitting beside the student, with paper past papers, a pen, and the ability to see what the student writes as they write it. Online tutoring means a video call with a shared whiteboard or a shared document, the same past paper on screen, annotated live, usually recorded, with the tutor's and the student's cursors both visible on the page. Both, done properly, are one to one against the same exam board. Neither is a video library and neither is homework supervision. If you are booking GCSE tuition, that is the comparison being made.

One more piece of framing before the evidence. The pages competing for this exact question are, with one or two exceptions, arguing from assertion and a comparison table. One of them — currently ranking on the first page of results — cites a landmark Education Endowment Foundation study of UK tutoring that does not exist, and repeats the invented finding inside its structured data so that search engines and AI assistants pick it up as fact. We are not going to name it, but it is the reason this page reads the way it does. Everything below is quoted from a document you can open and read yourself.

What does the EEF Toolkit really say about one to one tuition at GCSE age?

The Education Endowment Foundation's Teaching and Learning Toolkit is the source everybody quotes and almost nobody reads past the first line. That first line is genuinely encouraging: "Evidence indicates that one to one tuition can be effective, providing approximately five additional months' progress on average." Fine. But the Toolkit publishes a breakdown of that average immediately underneath, and the breakdown is where a parent of a Year 11 will find the figures that actually apply to them.

Two sentences do most of the work there. "Studies undertaken in primary schools tend to show greater impact (+6 months) compared with secondary schools (+4 months)" — so the figure that applies to a GCSE student is +4, not the +5 you see quoted on tuition websites, this one included until we rewrote it. And: "Effects in mathematics appear to be substantially lower (+2 months) than in literacy (+6 months)." That is a four-month spread between two subjects inside the same intervention (6 minus 2 is 4). If you are booking maths, which is what most families book, you are booking the weaker half of the average.

This is not an argument against tutoring maths. It is an argument against expecting the same return from every subject and every session, and it lines up with something most experienced tutors would tell you anyway: reading and writing improve through a great deal of guided practice that a tutor can supervise directly, whereas a maths grade often turns on a handful of specific topic gaps that have to be found before anything else can happen. Finding those gaps quickly is the whole job. Ask any tutor how they intend to do it in the first two sessions, and compare the answers you get.

The Toolkit is also unusually specific about the shape of the sessions, and its answer contradicts the way nearly all private tuition is sold: "Short, regular sessions (about 30 minutes, three to five times a week) over a set period of time (up to ten weeks) appear to result in optimum impact." Almost nobody sells that. The standard product is a single weekly hour running open-endedly from October to May. We are not going to pretend otherwise, and we are not going to pretend the evidence supports it. What we can say is that the frequency finding is one of the few places where online delivery has a structural advantage, because a half-hour slot at half past seven on a Tuesday costs nobody a car journey.

Why we do not just quote the headline number

There is a reason to be careful with +5 months even before you break it down by phase and subject, and the reason is printed on the Toolkit page itself. The EEF rates the security of the one to one tuition evidence as moderate rather than very secure, and it flags who ran the underlying studies: "A large percentage of the studies were not independently evaluated. Evaluations conducted by organisations connected with the approach – for example, commercial providers – typically have larger impacts, which may influence the overall impact of the strand."

Read that slowly, because it is a warning aimed squarely at pages like this one. A meaningful share of the research showing that tutoring works was produced by people selling tutoring, and the EEF says outright that such studies tend to report bigger effects. That does not make the finding worthless. It does mean the true average is more likely to sit below the headline than above it, and it means a tuition company quoting +5 months at you is quoting a figure that its own industry helped inflate.

Then there is the date. The strand carries the line "Review last updated July 2021". Everything inside the headline figure was assembled before the online tutoring market as it exists today had really formed. The number that gets pasted onto every online tuition homepage is, in large part, a summary of tuition delivered in rooms. Anyone using it as proof that their online service works is using a pre-online figure to make a post-pandemic claim.

We would rather say that plainly than lose the argument later. A page that volunteers the weaker version of its own best number is a page you can check, and checking is the entire point of what follows. If the only thing you take from this article is the habit of asking any tutor or agency "which document does that figure come from, and what year was it published?", it will have done its job — whether you end up booking with us, with somebody else, or with nobody at all.

Does the evidence say online works as well as in person?

This is the question the page exists to answer, and the honest answer is that the evidence points in two directions at once. It is worth seeing both before anyone tells you which way it points.

On the side of "the format barely matters", the Toolkit addresses delivery in a single sentence: "Studies involving digital technology show broadly similar effects." That is the EEF's own answer to the question this page is about, drawn from the same evidence base as the headline figure, and it is not a hedge. Alongside it sits the EEF's Remote Learning Rapid Evidence Assessment, which pulled together systematic reviews and meta-analyses on remote teaching and concluded: "Ensuring the elements of effective teaching are present – for example clear explanations, scaffolding and feedback – is more important than how or when they are provided. There was no clear difference between teaching in real time (synchronous teaching) and alternatives (asynchronous teaching)." If live versus recorded barely registers, screen versus room is unlikely to be the decisive variable either.

On the side of "something is lost", there is the EEF's National Online Tuition Pilot — the largest UK trial of online tutoring specifically, independently evaluated, which "reached 1,425 learners in 65 schools. In total, over 9,800 tuition sessions were delivered between 15 June and 28 October 2020." Its verdict on the experience is unambiguous: "Online tuition lacked some of the benefits of in-person delivery. While the offer of online tuition was highly attractive during the Covid-19 pandemic, most learners would prefer in-person tutoring if given the choice."

Notice what that finding is and is not. It is a preference finding, gathered from learners and tutors, and it is the strongest one we have. It is not an attainment finding — it does not say the online learners came out with lower grades. Effectiveness and preference are different things, and they are frequently confused in both directions on this topic. A student who would rather have a tutor in the room may still learn perfectly well on a screen; a student who says they prefer the screen may be preferring the fact that they can keep the camera off.

There is also a piece of Department for Education guidance that gets misused in this argument, so we will use it carefully. The DfE states that "Remote education should not be viewed as an equal alternative to attendance in school." True, and important — but it is about remote education replacing school attendance, not about tutoring that runs alongside school. Quoting it as evidence against online tutoring would be exactly the kind of source-stretching this page was rewritten to remove.

What did Ofsted find when it watched tutoring happen?

Ofsted's independent review of tutoring in schools is the only body of UK inspectorate evidence in which inspectors actually sat in tutoring sessions, remote and in person, and compared them. No other page answering this question cites it, which is odd, because its central finding is the most quotable sentence in the whole field: "The weakest tuition provision we saw involved remote tutoring through the tuition partner route."

It goes further on engagement: "Pupil engagement in tutoring sessions tended to be better when the sessions were in person." And phase 1 of the same review recorded school leaders being blunter still where tuition had been delivered remotely — "Leaders in these cases were clear that remote learning was ineffective for the vast majority of pupils."

If you sell online tutoring, that is uncomfortable reading, and the temptation is to skip it. It is more useful to look at what was being described. The tuition partner route means a school buying blocks of tuition from an external organisation, for pupils the tutor has never met, often outside the school day, against a curriculum the tutor is not party to and with little contact between the tutor and the class teacher. That is a specific delivery arrangement that happened to be remote. What the finding cannot tell you — because the review was not designed to separate the two — is how much of the weakness came from the medium and how much came from the arrangement.

Which is why the same report's other observation deserves equal billing: "Six secondary schools chose to use one-to-one as their preferred method (through online tutoring)." For those schools, online was not the compromise. It was the mechanism that made genuine one to one possible at all, because a school cannot conjure a specialist physics tutor out of its own staff room at four o'clock. That is the same logic that makes online work for a private family: it is not a cheaper version of the room, it is access to a subject specialist who does not live near you.

Ofsted also caught something about who gets tutored once exams returned to normal: "Typically, more Year 11 pupils were being selected, particularly those who were borderline GCSE grade 3/4/5." Schools concentrate their tutoring around the grade 4 threshold in English and maths, because that is where a school's published measures move. A family's priorities are usually different — a grade 6 pushed to a 7 for sixth form, or a single subject dragging everything else down. Worth knowing, because if your child sits comfortably above that borderline, school-arranged tutoring may never reach them however good it is.

How much can tutoring realistically move a GCSE grade?

Here is the finding the tutoring industry does not print. The Department for Education's year three impact evaluation of the National Tutoring Programme is the largest attainment study of tutoring ever run in England, and on the GCSE cohort it reports this: "The effect sizes seen at KS4 were consistently smaller than those seen at KS2 and in all cases equated to less than one month's additional progress, although some did still reach statistical significance."

Less than one month. Against a Toolkit headline of approximately five additional months, and a secondary-specific figure of +4. There is no honest way to reconcile those except to say that a national programme delivered at enormous scale, through mixed routes, to pupils selected by schools, is a very different thing from a specialist tutor working with one student on one specification for a fixed number of weeks. Averages across a national rollout tell you about the rollout. They do not tell you what a particular tutor and a particular student can do together. But they are a corrective to any promise of grades, and anyone quoting the Toolkit headline at you without mentioning this evaluation has either not read it or has decided you should not.

The same evaluation gives the only published UK answer we know of to "how many hours is enough": "our best estimate for the number of tutoring hours per tutored pupil that maximised school level outcomes across all pupils was 31 for maths and 20 for reading". Treat that carefully — it is a key stage 2 estimate, not a GCSE one, and it is a school-level optimum rather than an individual prescription. Even so, it is a far better anchor than the figure a sales page will give you, and its shape is informative: maths needed more hours than reading, and both are larger than the four or five sessions families often imagine will do it.

Put the two EEF findings together and you get something usable. Sessions of about 30 minutes, three to five times a week, over up to ten weeks, is the pattern with the best evidence behind it. Twenty to thirty hours is the order of magnitude at which a programme starts to shift outcomes. If your budget of hours is much smaller than that, spend it narrowly — one subject, one paper, the specific topics the mock exposed — rather than spreading it across three subjects where it will simply disappear.

How do you make online tutoring actually work?

If you do choose online, the evidence supports three specific instructions, and they are considerably more concrete than the "make sure they have a quiet space" advice that fills most articles on this subject.

First, the device. Ofsted's guide to what works well in remote education says: "Quality of teaching is far more important than how lessons are delivered. But there is some evidence that the medium does matter, especially in digital remote education. Pupils tend to spend longer accessing a remote lesson when they are using a laptop than when using a phone (tablets are in between)." That is an unusually actionable finding. A GCSE tutoring session in which a tutor is annotating a past paper needs a screen big enough to hold a question, a diagram and the working at the same time. A laptop is the sensible minimum; a tablet with a stylus is genuinely good for maths and science because the student can write rather than type; a phone is the weakest of the three, and if a phone is the only device available then the case for in person gets stronger immediately.

Second, the length. The same Ofsted guidance notes that "because evidence suggests that concentration online is shorter than the length of a typical lesson, filming a classroom lesson may be ineffective." That points the same way as the EEF's optimum session pattern of about 30 minutes, three to five times a week. If you have bought an hour, ask the tutor to break it up: a stretch of teaching, a stretch of the student working under something like exam conditions while the tutor watches, then feedback on what they actually wrote. An hour of continuous talking at a screen is the format most likely to fail, and the one least likely to be noticed failing, because the student's face stays politely in the frame throughout.

Third, the subject. Ofsted's remote education research found particular subjects were consistently harder: "Art, science, physical education and design and technology were commonly cited as problematic." The common factor is practical work. For GCSE that argues for splitting the subject rather than the child. Required practicals, apparatus handling and studio work need a room; the extended written responses, the maths inside physics and chemistry, the exam technique and the write-up of a practical already done at school transfer to a screen with very little lost. The same reasoning applies further up the school: it is why much of our A-level tuition in the sciences runs online without difficulty while the practical endorsement stays firmly with the school. You can see the full range on our subject list.

What about students with SEND, or students who lose focus?

Most articles assert that in-person suits students with additional needs and leave it there. There is a real UK figure behind the intuition. Ofsted's remote education research reports: "The YouGov survey found that 59% of parents of a pupil with SEND said that their child has been disengaged with remote learning, compared with 39% of parents of children without additional needs." That is a gap of 20 percentage points (59 minus 39).

Two caveats keep that honest. It describes remote schooling during a national closure, not one to one tutoring, and the difference between those two situations is enormous: a tutor is watching one face and can stop the moment attention goes, which a teacher facing a grid of thirty cannot. And 39% of families without additional needs reported disengagement too, so this is not a SEND problem with a neurotypical control group that sailed through. It is a general risk that lands harder on some students than others. Either way, it makes engagement the thing to monitor in the first fortnight, whichever format you pick.

The same research also corrects the thing parents tend to worry about. "The data highlights that 11% of parents saw access to an appropriate device as a challenge, compared to 40% who responded their child's focus on studying was a worry." Roughly four times as many parents were worried about focus as about kit (40 divided by 11 is about 3.6). Most of the online tutoring checklists you will read are device checklists. The evidence says the harder problem is motivation, and no laptop specification fixes it.

It is worth ending this section on the sentence school leaders gave Ofsted, because it is the least fashionable and most useful line in the whole literature: "remote education, as the medium for teaching, does not change how pupils learn." Retrieval practice, spacing, worked examples, feedback on a real attempt — these work the same way through a webcam as they do across a kitchen table. What changes is how easily a student can hide, and how easily a tutor can tell.

Can an online tutor really be "Ofsted accredited"?

No, and this is the single most checkable claim in the online tutoring market. The Department for Education runs an Online Education Accreditation Scheme, and Ofsted genuinely is involved — the guidance states that "Ofsted will act as the quality assurance body for the scheme and will carry out visits in line with its inspection handbook." So the phrase sounds entirely plausible, which is why it spreads.

But the scheme is non-statutory and covers full-time online education of compulsory-school-age pupils in England, and the DfE's own explanation of how it works says: "Supplementary providers such as online tutoring services or agencies will not be eligible to apply to the scheme." Not "have not yet applied". Not eligible at all. And the DfE anticipated the loophole, instructing accredited providers that they "must not state or imply that their accreditation covers other services that they may offer, such as part-time provision, full-time provision for pupils that live outside England, or tutoring services."

The DBS line in that table is the second question worth memorising. The DfE's online education standards say that "if a person works for an online provider for more than 3 days in 30 where they have access to children (albeit remotely) and are there to ensure the establishment can provide its services, they will be in regulated activity with children." A weekly tutor clears that threshold comfortably. There is a real difference between a basic DBS certificate and an enhanced one with barred list information, and an agency that cannot tell you which it holds within a sentence is telling you something. This applies to in-person tutors identically — the distinction is about the check, not the camera — and it is the same standard we apply across entrance exam preparation and every other part of the service.

Does where you live decide this more than the format does?

There is a constraint that outranks every argument above, and it is geographical. The Sutton Trust's Private Tutoring 2026 report breaks uptake down by where families live, and the spread is wider than any effect size on this page.

A London pupil is 26 percentage points more likely to have been tutored than a rural pupil (45 minus 19). It is not plausible that London parents care 26 points more about their children's GCSEs. What differs is supply: the density of specialist tutors within a reasonable drive, and how many of them can teach a particular subject at a particular level on a particular evening. For a family outside a large town, "online versus in person" is frequently a false choice dressed up as a real one — the actual choice is an online specialist in the right subject, or a local generalist, or nothing at all.

That is the strongest argument for online tutoring at GCSE, and it is worth noticing that it has nothing to do with the medium being better. It is about who you can reach. A student sitting a less commonly taught subject, or needing a tutor who knows the specification in detail, or wanting a Saturday morning slot in a household with two working parents, is choosing between a real option and a hypothetical one.

And for balance, the same Sutton Trust report ends up recommending something a purely online reading would not predict. Its policy recommendation is to "Re-introduce funding for in person tutoring in schools, with ringfenced funding for the long-term, targeted at disadvantaged pupils." The organisation that produced the geography data, having weighed the evidence, names in-person delivery in schools as the thing worth funding. We could have left that out. It is more interesting in.

So what should you actually choose?

Nothing in this evidence base says one format produces better GCSE grades than the other. What it says is narrower and more useful: teaching quality dominates, delivery mode is a second-order variable, students on average prefer the room, and the failures inspectors observed in remote tutoring looked like failures of arrangement — disconnected from the curriculum, bolted on after school, with no contact between tutor and teacher — rather than failures of the camera.

So choose in person if the student is in the group the evidence flags: prone to disengaging, working from a phone, needing someone physically present to keep the session honest, or studying a subject whose practical core cannot be simulated. Choose online if the constraint is supply — if the right specialist for the specification is not within reach — or if the schedule only works in short, frequent slots, which the evidence says is the better pattern in any case. Choose a split for the sciences and art, keeping the practical work where the equipment is.

Then, whichever you pick, ask the three questions that actually predict quality. Which exam board specification is my child sitting, and can you name the assessment objectives it is marked against? What is the plan for the first fortnight, and how will you find the topic gaps? And what evidence will you show me that it is working — marked past-paper questions, or an impression of growing confidence? A tutor who answers those well will be effective in a room or on a screen. A tutor who cannot will be ineffective in both.

Frequently Asked Questions