How to ensure class quality in a language school

When the school had a single teacher, it was simple. You taught the class, you knew what had been covered, you knew whether the student spoke or stayed quiet. Today you have three, five, ten teachers running classes in parallel, and the only thing that reaches you is whatever the student complains about on WhatsApp. Usually when it is already too late.

This piece is about how to ensure class quality at a language school without turning it into a surveillance operation and without depending on a polite student's hint to tell you when something is wrong.

Why class quality is so hard to ensure in a language school

A language class is one of the hardest services to audit. There is no written deliverable, no invoice for "1 hour of English," no physical product coming off a shelf. There is a teacher, a student, a one-hour window, and what happened in that hour is a black box to everyone who was not inside it.

And that black box hides a lot. A teacher can skip lesson-plan content to finish early. They can stay superficial, flipping through the page without truly digging in. They can leave the student talking alone the entire class because "it is a conversation class." They can do the opposite and dominate the talking, leaving the student no room to produce anything. They can use the class to share personal stories. They can show up late, cut 10 minutes off the end, repeat the same exercise from three classes ago.

None of this shows up in a financial report. All of it directly affects how your school's quality is perceived.

The owner cannot be in every room. Neither can the academic coordinator. And by the time the student's complaint finally arrives, it has usually been 4, 5, 8 classes with the problem. At that point, the student's trust in the school is already shaken.

The symptoms that show your quality standard is slipping

Before any tool, it helps to recognize the signs. Most online language schools feel several of these at once:

• A student complains about teacher X but not Y, and you cannot objectively describe what is different between the two.
• Renewals drop in a particular group and no one can explain why.
• The teacher swears they are covering the content and the student swears they never saw that topic.
• The coordinator gives the teacher feedback based on a hunch or on what another student mentioned.
• The class pattern varies so much between teachers that a student switched to a new teacher feels like they changed schools.
• Initial training is strong, but six months later each teacher is already doing it their own way.

It all comes down to the same root: the class is invisible once it ends, and all that remains is subjective memory.

How most schools try to solve this today

With no way to see inside the class, schools try to attack it from the outside. Each of these approaches has merit, but they all hit the same ceiling.

Initial training and a teacher handbook. Works well in the first few months. It erodes as time passes, as new teachers come in, as each one adapts the method. Without reinforcement, it becomes a document gathering dust on Drive.

Mandatory lesson plan. The teacher fills in what they are going to teach. You have no way to know whether they actually taught it, whether it was superficial, whether 20 minutes of chit-chat were left at the end. A plan is intention, not execution.

Student feedback form. It captures perception, not fact. And it captures only the students who respond, who tend to be either very satisfied or very upset. The middle of the chart, where most people live, answers nothing. When a student does respond poorly, it is usually the last warning before cancellation.

The coordinator dropping into class by surprise. It creates a bad atmosphere, distorts the teacher's behavior in that specific class (they prepare more when they know they might be visited), and even then you can only cover a tiny fraction of the classes that happen each week.

Monthly academic meeting. Good for aligning guidelines, bad for catching problems early. By the time the problem reaches the agenda, it has already become a crisis.

Asking the teacher to record the class on Zoom and send it. Almost no one does, and those who do send the link once a month. You never have time to watch 30 hours of recordings per week.

The common denominator of these alternatives: they all try to supervise before or after the class, without touching what happened inside it. And what happened inside it stays opaque.

What is missing to truly ensure consistent classes

To have a real quality standard in an operation with several teachers, three things need to exist at the same time, without turning into surveillance:

Objective indicators of what happened. Not opinion, not memory. Raw data: how much each person spoke, which words came up, at what moment, for how long. This data answers questions that subjective feedback cannot, like "does teacher X give the student time to produce?" or "did today's class actually cover new vocabulary?"

Sample access to the full class. The coordinator needs to be able to review an entire class without arranging it with the teacher beforehand, without dropping in by surprise, without relying on a manual recording. Ideally, any class from last week is available to be reviewed by whoever has permission to do so.

Continuous visibility, not occasional. An indicator looked at once a quarter does not change behavior. An indicator that shows up at the end of every class does. If the teacher knows that at the end of each class there will be a report with speaking time, vocabulary covered, and teaching suggestions, they self-correct without needing a coordinator breathing down their neck.

And the politically important point: none of this should become a tribunal. Good data serves to show the teacher where they stand and open a conversation, not to create a witch hunt. A school that becomes a surveillance environment loses its good teachers before it improves the average ones.

How Noladi solves it

Noladi was built around a simple idea: an online class does not have to become a black box the minute it ends. It can become data.

Every class taught in Noladi's live class room is recorded in full and stays available for later review. The post-class AI generates a transcription that identifies who said what, and from it computes objective per-participant indicators: teacher speaking time vs student speaking time, unique words used, speaking pace, pauses. Teaching suggestions are generated automatically for each student based on what they produced in the class.

For the coordination team, this becomes a supervision tool with no weight. Instead of barging into a class, the coordinator opens the lesson review of the class that already happened, sees the indicators, and if they want, hits play and watches only the segment that caught their interest. Instead of feedback based on a hunch, the conversation with the teacher is now based on what actually happened. And with roles and permissions, you decide who on the team accesses what, without exposing everything to everyone.

The side effect is that the teacher also sees their own numbers class by class. Whoever is letting the student speak too little sees the unbalanced chart and adjusts on their own. Whoever is covering little new vocabulary notices it. Most course correction stops needing a coordinator, because the data is already staring the teacher in the face when the class ends.

Get to know Noladi

If you are setting up or already running an online language school and want to stop depending on student complaints to know what is happening in your classes, get to know Noladi at noladi.app/teacher. Live class room, automatic AI review, objective per-class indicators, and team management in a single system.