What Running a Tutoring Company Taught Me About Tracking Student Progress
How do you track student progress in a tutoring company?
The methods that scale capture progress as a by-product of work students already do: marked homework and practice questions that roll up into a per-topic picture of what each student knows, rather than notes a tutor writes from memory after the fact. For my first few years I didn’t really track progress at all. I hoped, I asked how the tests went, and I re-tested a hard question the next session to see if it had stuck. That was fine with a handful of students, and it fell apart the moment one cancelled for a fortnight or arrived brand new. What a real progress-tracking system has to do is narrow: show mastery topic by topic, pull from actual marked work rather than memory, be visible to parents, and stay comparable across every tutor.
This is the practical companion to a separate argument I’ve made, that session counts are a vanity metric and learning signals are what predict retention. That post is the why. This one is the how: what I actually did, why it broke, the step most owners reach for next, and what finally held.
Method 1: memory and hope (asking how the tests went)
I’ll be honest about where I started, because it’s where most tutors start: I didn’t track progress at all. I hoped. I’d ask a student how their grades were going, how they went on the last test, and I’d take the answer at face value. Then, at the start of a session, I’d get them to solve a question they’d struggled with the week before, to see whether they’d actually understood the concept or just nodded along at the time. That felt like tracking. It wasn’t. It was remembering, backed by whatever the student happened to remember too.
It held together while I was seeing the same few students every week and the last session was fresh in my head. The cracks showed the moment that stopped being true, and there were three of them.
The first was cancellations. A student would cancel for a fortnight, and by the time I saw them again the thread was gone. I couldn’t reliably remember which question they’d stumbled on or what the underlying gap was, so I’d spend the first part of the session re-diagnosing something I’d already worked out two weeks earlier.
The second was new students. With someone I’d never taught, there was no starting picture at all. I was working blind for the first few sessions, feeling out the weaknesses by trial and error, which is a slow and expensive way for a parent to buy a diagnosis.
The third is the one that quietly undermines the whole approach: students forget their own weaknesses. If your tracking system is “I’ll ask them what they found hard,” you’re relying on the least reliable record in the room. Memory isn’t tracking, and a student’s memory of their own weaknesses is even worse than mine. A record that lives in two people’s heads, and leaks from both, was never going to survive more students or a second tutor.
Method 2: the spreadsheet most owners reach for next (and why it caps out)
I skipped this step, but it’s the one nearly every tutoring company reaches for once memory stops coping, so it’s worth understanding before you sink a month into building one. The instinct is sound: a shared sheet, a row per student, columns for topics or weeks, tutors filling it in. It’s a genuine upgrade on memory. The data is in one place, comparable across students, and in principle anyone on the team can see it.
The trouble is that a spreadsheet fails slowly, through neglect, rather than all at once. It’s only ever as good as the discipline of the people entering it, and busy tutors at the end of a long day are not disciplined data-entry clerks. Cells go blank. One tutor uses “good/ok/poor”, another a 1 to 5 scale, a third writes a sentence. A couple of weeks of half-filled rows and the sheet stops being trustworthy, and a tracking system you don’t trust is one you quietly stop opening.
The deeper problem isn’t discipline, it’s duplication. The spreadsheet asks tutors to re-enter information that already exists. The student did the homework. The homework has answers that are right or wrong. The progress is sitting there in the marked work. Asking a tutor to then transcribe a summary of that into a sheet is asking them to do the measurement a second time, by hand, after they’ve already done it once with a red pen. At any real scale, that second step is the one that doesn’t happen. That’s the trap I’m glad I never had to climb out of.
Method 3: marked-work data (progress as a by-product of teaching)
The shift that actually worked was to stop treating progress tracking as a separate task at all, and to capture it from the work students were already doing. If a student does homework and that homework is marked, the result of that marking is the progress data, question by question and topic by topic. The job isn’t to write it down afterwards. It’s to capture it at the point of marking and let it roll up on its own.
This is the difference between tracking being extra work and being no work. When the data comes from marked homework and practice questions, nobody has to remember to update anything: the work gets marked (or auto-marked), and the student’s topic-by-topic picture updates itself. It also closes the exact holes that sank memory. The record doesn’t live in a tutor’s head, so a cover session or a change of tutor doesn’t reset the student to zero, and it doesn’t depend on anyone’s data-entry discipline, because the data is generated by the act of marking, which was always going to happen anyway.
What a workable progress-tracking system actually needs
Trying things that broke taught me the requirements the hard way. Each one is a lesson named after the method that failed without it. If you’re choosing or building a way to track student progress, these are the four things it has to do before it earns the name.
1. Mastery topic by topic, not a grade and not a session log. The useful unit is the topic, not the session and not an overall mark. “Doing well” isn’t trackable; “solid on simultaneous equations, shaky on the discriminant, hasn’t touched probability” is. A system that only records that a session happened, or only stores a single overall score, can’t tell you what a student knows, and “what does my child actually know now” is the question the whole exercise exists to answer.
2. Captured from real marked work, not re-entered from memory. This is the lesson both memory and the spreadsheet teach. Any system that depends on a person recalling or transcribing a summary after the session will degrade, because that step competes with everything else at the end of a tutor’s day and loses. The progress data has to come from work that already happens (marked homework, practice questions) so capturing it costs no extra effort and survives a busy week.
3. Visible to parents, because the parent is the one deciding whether to stay. A progress record only the company can see solves half the problem. The person paying renews when they can see the child improving, not when you tell them the child is improving. So the same data that helps you run the company has to be something a parent can open and read for themselves: topics covered, where their child is strong, where they’re behind. (The why this drives retention is the subject of why tutoring clients really cancel; the short version is that they leave when they can’t see progress, not when admin is bad.)
4. Comparable across tutors, so quality doesn’t depend on who you got. This is the requirement memory can never meet, because whatever’s in one tutor’s head leaves when they do. Once you have more than a few tutors, your biggest hidden risk is variance: one tutor’s students genuinely progress, another’s plateau, and from the outside the company looks identical. A progress system has to roll up the same way for every tutor so you can see that difference. If progress is recorded differently by every tutor, or only in their heads, you don’t have quality control. You have hope.
Where I landed
I won’t pretend I arrived at a system through sheer discipline. I ran out of road on hoping and remembering, and I set out to build the thing I wished I’d had. That became ClassQuill, and it’s what my company runs on now.
What changed first was the start of every session. I no longer have to ask a student what they want to cover, or spend the first ten minutes re-diagnosing a gap I’d already found a fortnight ago. I open the session already knowing where they’re up to. Students also save the questions they get stuck on during the week, so nothing they struggled with is forgotten by the time we’re back together. Sessions became far more targeted as a result: students got noticeably more out of each one, their results improved, and that improvement showed up where it matters most for a tutoring business, in student and parent satisfaction, which in turn drove more referrals.
Underneath it, the mechanism is the four requirements made real. Tutors set homework from a curriculum-aligned question bank; it’s auto-marked, and every result feeds a topic-by-topic mastery profile that’s attached to the student, not the tutor, so cover sessions and handovers don’t wipe the history. That profile rolls up two ways: parents get a progress view they can open themselves, and I get a cross-tutor view of who’s progressing and who’s slipping.
One honest note on the “AI” part, because it’s where a lot of tools oversell. ClassQuill isn’t a black box that invents questions from scratch. I found out the hard way that raw AI generation isn’t specific enough for a given curriculum, and it sometimes gets the maths wrong, which is the last thing you want in front of a student. So the model is different: a curriculum graph that understands each topic and what it involves, sitting on top of a vetted bank of base questions, with AI used to tailor those questions to the individual student’s gap. Standard, very good material plus tailoring, not a generator left to its own devices.
I’m biased, I built the thing. But the four requirements stand on their own whatever you use to meet them. If you’re tracking progress in your head or a spreadsheet right now, it’s probably working today. The real question is whether it survives the next five students and the next two tutors. Mine didn’t, and the pattern in how it broke is the most useful thing running a tutoring company has taught me.
Related reading:
