Is AI Worth Paying For in a Small Tutoring Business?

Coding Liquids tutorial cover featuring Sagnik Bhattacharya for Is AI Worth Paying For in a Small Tutoring Business?
Coding Liquids tutorial cover featuring Sagnik Bhattacharya for Is AI Worth Paying For in a Small Tutoring Business?

Yes, usually, but only one general AI plan at $8-$20 a month per tutor who plans lessons, not a stack of specialist tools. It earns its fee if it saves a tutor about an hour a week on lesson plans, worksheets and parent reports after checking time. Per-student AI platforms are hard to justify below roughly 50 students.

The subscription price is the smallest number in the decision. A $20 plan pays for itself if it saves one hour a month at almost any tutor's rate. The costs that decide it are the minutes spent checking what the AI produced, and the risk of a wrong answer key or a clumsy parent report going out under your name. A tool that saves 30 minutes of drafting but needs 25 minutes of checking isn't earning much.

Follow me on Instagram@sagnikteaches

Where a tutor's unpaid hours actually go

Before comparing plans, list the work that happens outside sessions. Here is an illustrative week for one tutor with 18 students across maths and science, and how much AI can realistically take off each task once you include checking.

Connect on LinkedInSagnik Bhattacharya
TaskHours a week nowAI's roleRealistic saving after checking
Lesson plans for each student3.0First draft from your notes on each student1.0-1.5 hours
Worksheets and practice questions2.0Draft questions at set difficulty, with answers0.5-1.0 hour (answer keys need checking)
Parent progress updates1.5Turn bullet notes into a readable update0.5-1.0 hour
Enquiry and scheduling emails1.0Reply drafts and templates0.3-0.5 hour
Marking1.5Limited: handwritten and multi-step working is unreliable0-0.3 hour

That tutor could reasonably save two to four hours a week. Marking barely moves, and it's where owners are most tempted to over-trust the tool: a chat assistant reading a photo of handwritten working will misread a digit or accept a wrong method that happens to land on the right answer.

Subscribe on YouTube@codingliquids

Which single plan to pay for

Most small tutoring businesses need one of these, not several. Prices are list prices in USD as of September 2026.

PlanPriceSuitsWatch for
ChatGPT Go$8/monthSolo tutors doing light draftingMay show ads; lower allowances than Plus
ChatGPT Plus$20/monthSolo tutors planning most lessons with AIConsumer plan: turn off model training in data controls
Claude Pro$20/month, or $17 billed annuallyTutors who write a lot of parent reports and long feedbackConsumer plan: check the training setting
Google AI Pro$19.99/monthTutors who already work in Google Docs and SlidesPersonal account, not a business workspace
MagicSchool Plus$12.99/month, or $8.33 billed annuallyTutors who want ready-made teaching tools (question sets, rubrics)Built for classroom teachers; less flexible for business admin
ChatGPT Business or Claude Team$25/seat monthly, $20 annually, minimum 2 seatsCentres with three or more tutors sharing materialsYou pay for at least two seats

Notice what isn't on the list: your tutoring management software. TutorBird lists one plan at $16.95 a month plus $4.95 for each extra tutor, and Teachworks prices by plan plus a per-lesson fee; neither lists AI features on its pricing page at the time of writing. So the AI budget in most tutoring businesses is a chat plan, and the scheduling and billing tool stays as it is. If a management platform later adds AI to reports or scheduling, compare it against the chat plan you already use rather than paying for both.

A solo tutor, a small centre and a larger one

A solo maths tutor with 14 students

She plans each lesson from last week's notes and writes a short monthly update to parents. With ChatGPT Plus she drafts plans from bullet notes and generates practice sets, saving about two hours a week. Her time is worth $45 an hour in sessions she could add, so two hours a week is $360 a month of capacity against a $20 fee. Verdict: pay, and the Go plan would probably do if budget is tight.

A three-tutor maths and English centre

The owner wants shared worksheet banks and a consistent tone in reports. Three Plus accounts cost $60 a month but nothing is shared. ChatGPT Business at $20 a seat annually (three seats, $60 a month) gives shared projects, keeps business content out of model training by default, and lets the owner set up one project holding the centre's report style and marking rubric. Same price, better fit. Verdict: pay for the business plan.

An eight-tutor centre looking at a student-facing AI tutor

A vendor offers an AI practice companion for students, priced per student. With 60 students, even a few dollars each is a meaningful monthly bill, and the centre must then answer parents' questions about what the tool says to their children and where their data goes. Verdict: pay for staff-side plans first, and only trial a student-facing product with a small, consenting group and a clear written policy.

A parent report prompt, and what to fix in its output

Progress updates are where paid plans earn their keep fastest, because the input is your own notes and the output is easy to check. A prompt you can reuse:

You are writing a short progress update from a tutor to a parent.
Tone: warm, specific, plain English, no jargon, no exaggeration. 120-160 words.
Use only the facts in my notes. Do not invent test scores or praise.
Structure: one line on what we covered, one strength with an example,
one area to work on with what we'll do next, one thing the parent can do at home.

Student: first name only.
Notes: fractions to decimals secure now, 9/10 on quick quiz. Still rushing
word problems, misses units. Next 3 weeks: two-step word problems, show working.
Home: 10 mins of times tables 3x a week.

An illustrative output, trimmed:

This month [first name] and I focused on converting fractions to decimals, and it has really clicked: [first name] scored 9 out of 10 on our last quick quiz and can now explain each step. The next thing to work on is word problems. [first name] tends to rush them and sometimes drops the units, so over the next three weeks we'll practise two-step problems with working shown in full...

What you would fix: "really clicked" and "can now explain each step" go slightly beyond the notes, which only said "secure". Parents quote these reports back to you, so trim anything you didn't observe. Keep the first-name-only rule even on a paid plan. For a full workflow across a centre, see how tutoring centres use AI for parent progress reports.

A parent enquiry, answered before and after

Enquiry emails are a small job with a big effect, because the first reply often decides whether a family books a trial session. Here's an illustrative before and after for a centre offering maths and English.

The parent's email: "Hi, my son is 14 and struggling with maths, especially algebra. Do you have anything on Tuesday or Thursday after school, and how much is it? He's quite anxious about it."

  • Before (typed in a hurry between sessions): "Hi, yes we have Tues 5pm. It's $40/hr. Let me know. Thanks."
  • After (AI draft from the centre's saved reply prompt, then checked): "Hi, thanks for getting in touch. Algebra is one of the most common things we help with, and lots of students feel anxious about maths at first, so our first session is always relaxed: we find out where he's confident and where the gaps are. We have Tuesday at 5pm or Thursday at 4:30pm available. Sessions are $40 an hour, and the first one comes with a short written summary for you. Would either slot suit?"

The second version answers all three questions (subject, times, price), picks up on the anxiety, and ends with an easy choice. What you'd check: the times must come from your real timetable, and any promise ("a short written summary") must be something you actually do. Drafting it took under a minute with a saved prompt; typing it from scratch takes most tutors five.

The answer-key mistake that costs trust

Here's a realistic example. A tutor asks for ten quadratic equations with worked answers. Nine are fine. Question 7, x² - 5x - 14 = 0, comes back with the answer "x = 7 or x = 2". The correct roots are 7 and -2. The student gets it right, sees it marked "wrong" against the key, and loses confidence; the parent hears about it that evening.

Two habits prevent this. First, ask the assistant to check each answer by substituting it back, and to show that check. Second, solve two or three questions yourself before the sheet goes out, choosing the hardest ones. The checking routine for AI-made worksheets sets this out step by step. Build the checking minutes into your saving estimate; they are part of the real cost.

Student and parent details: what goes in, what stays out

Tutors hold information about children, which raises the bar. Whatever plan you pay for, set these rules and write them down for every tutor:

  • First names or initials only in prompts. No surnames, schools, dates of birth or contact details.
  • No special educational needs, medical or family details in consumer plans. If a report genuinely needs them, write that sentence yourself.
  • No photos of students' work with names visible. Crop or cover the name before uploading.
  • Training switched off on any consumer plan, in the privacy or data controls settings. Business plans such as ChatGPT Business and Claude Team don't train on business content by default.
  • No recording or transcribing sessions without written consent from parents, and a clear note of where recordings go and when they're deleted.

Data-protection law applies to children's information whichever tool you use. If you plan to use AI with anything beyond first names and learning notes, ask a data-protection adviser before you start rather than after a parent asks.

Per student, per month: the quick sum that settles it

Owners sometimes find it easier to think per student. A solo tutor with 14 students on a $20 plan pays about $1.43 per student a month. The three-tutor centre on ChatGPT Business, with 45 students, pays $60 a month, or $1.33 per student. If either charges $35-$45 an hour for sessions, the AI costs less than three minutes of one session per student each month.

Put that way, the question isn't "can we afford it" but "does it make each student's sessions better or our week shorter". A centre that can't point to either after a month should stop paying, however cheap it looks per head.

Two thresholds worth knowing when you grow:

  • Moving from individual plans to a team plan makes sense at about three tutors who share worksheets or report styles. Below that, two individual plans cost the same as the two-seat minimum and give each tutor more independence.
  • Adding a second, specialist tool makes sense only when one named task still takes more than two hours a week after the chat plan is in use, and the specialist tool demonstrably cuts it in a trial.

Three purchases that rarely pay at this size

Once one plan is working, vendors will offer more. These three come up most often in tutoring, and each is easy to overspend on.

  • AI marking services. Useful for large volumes of multiple-choice or short typed answers, which small tutoring businesses rarely have. For written working and essays, you still read every script, so you pay twice: once for the tool and once for your checking time.
  • AI-writing detectors. Owners buy these to catch students handing in AI-written homework. They misfire often enough that you should never confront a student or parent on a detector score alone. A two-minute conversation where the student explains their answer is a better test, and it's free.
  • Done-for-you curriculum packs "made with AI". A full year of generated lessons sounds efficient, but it isn't matched to your students and still needs every page checked. Your own notes plus a chat plan produce materials that fit the student in front of you.

A typical pattern, sketched as an illustration: an English tutor with 20 students tries a $15-a-month marking tool for essay feedback. The comments come back generic ("consider varying your sentence structure") and she rewrites most of them, so she cancels after a month and drafts feedback in her chat plan from her own margin notes instead, which takes less time and sounds like her.

A four-week trial that tells you whether to keep paying

Pay for one month, then decide with numbers rather than impressions. Each tutor keeps a simple log for the tasks in the first table. Here is how one tutor's log might look at the end of week four, as an illustration:

TaskTimes usedMinutes saved (drafting)Minutes spent checkingErrors caught
Lesson plans525201603 (wrong level)
Worksheets143501402 (answer key)
Parent updates18270704 (overstated praise)
Emails25125200
Total1,2653909

Net saving: 875 minutes, about 14.5 hours in the month, for $20. That is a clear keep. Keep paying if net saving is above two hours a month per tutor and errors caught are falling by week four. Cancel, or drop to a cheaper plan, if checking time is more than half of drafting time on the tasks you care about most, or if tutors stopped using it after week two. If lesson planning is the biggest line in your log, planning a week of lessons in an hour shows how to get more from it.

A final check before renewing: ask each tutor which single task they would miss if the plan disappeared tomorrow. If nobody can name one, the subscription is a habit, not a tool.

Further reads

Sources: OpenAI, Anthropic and Google plan pages; MagicSchool pricing page; TutorBird and Teachworks pricing pages (no AI features listed), all checked September 2026.

Want to know which AI jobs pay off in your centre?

On a 1:1 call we'll list where your tutors' unpaid prep hours go, pick the one or two tasks worth handing to AI, and set up a plan and data rules your team can follow.

Book a 1:1 call with me