Articles & Research

How to Train 8,000 Teachers and Make the Company Money

Nikita Berezin, head of the teacher training and development department at Skyeng, explains how to influence teachers’ performance metrics — and how to make the company extra money in the process.

Skyeng has 8,000 teachers, and they deliver some 15,000 lessons a day. All lessons run on our interactive platform, where content is packaged into courses on the widest range of topics and for every level.

The quality of learning depends most of all, of course, on the teacher. That is why, from the very start of a teacher’s collaboration with Skyeng, there is a multi-stage selection: resume review, an interview with a methodologist, and a basic training course. The topics in a new teacher’s training:

Once a teacher has passed the initial selection, is teaching several students, and delivers about a hundred lessons a month, they need active support. A teacher’s effectiveness is measured in several metrics:

lesson quality scored on a checklist; the share of planned lessons that actually take place (how often the student skips); the intensity of a student’s lessons; how long students stay with them; the number of lessons per month. The main question: how do you drive these metrics up? So, the task: organize regular training for 8,000 teachers, on whom product quality — and therefore profit — depends directly.

Teacher Training as a Product

At our company we treat everything as a product. Teacher training became a product too. A year ago, experienced teachers attended the occasional seminar, took courses on the internal platform, and received methodological support on request. That training was irregular, its effectiveness was hardly measured at all, and it had no effect whatsoever on teachers’ metrics. Everything had to be built anew.

The plan was this:

Assemble a Team

Research — which problems we can solve to grow the metrics Research the market and develop training that affects lesson quality Settle on didactics and training formats

Launch A/B Tests

Pick one successful find and scale it Survive it all :) The team was minimal: two project managers, a methodologist, and an expert who runs the training. So, let us begin with the very first one.

Developing the topics that affect lesson quality We gathered the data we needed — requests to methodological support, scores from the internal control department, student ratings, and more. We surveyed the teachers themselves, too. The result was a backlog of training topics; we prioritized it and began development.

Training Format

The classic formats the market knows:

A webinar — cheap, you can pull in a lot of people, but boring: pure theory and hardly any live communication. A course — you can package anything into it and pull in lots of people, but it takes long, and there is no live communication at all.

From our custdev interviews we understood what our teachers value in training: live formats, where you can take an active part in the discussion and practice on concrete cases. We tinkered a little with the didactics of the format — and that is how the seminar was born.

User story:

the teacher receives an assignment before the seminar (a short survey), the expert builds a presentation from the survey results and gathers the most frequent cases to be analyzed at the seminar, at the meeting itself there is personalized practice with every participant who completed the assignment, after the meeting participants receive a final assignment in which we ask them to rate the seminar and complete one more short task. The main question there — when, with which student, and which single action from the seminar will you apply? Assessors listen to part of a lesson and send feedback to the teacher + a certificate from ttc.

As the user story shows, the seminar (or workshop) is not the center of our training. The center is application. All of our training exists so that the teacher goes out and applies the action they have learned. Only then do we consider the training to have taken place.

That is how our product gained a leading indicator — the percentage of knowledge applied in the lesson. We began running seminars and developing our other products on the course map. Everything seemed fine — but not for very long!

Teachers who teach other teachers On the threshold of the third quarter, problems came pouring in. We had already built the entire mechanics of the training, but we were still not running the many experiments needed to prove impact on money. Our seminars were no longer as good as they had been — with 50+ people in them, effectiveness is low.

We had no analytics either, and several projects we had worked on for months never made it to the developers. On top of that, the team was given a condition: if you do not scale a single one of the four launched experiments within a quarter, it all ends for you. We went on building.

At the research stage, we identified the problem topics — solving them improves the lesson. It took us a while to hit on the idea that we could use, in training, the very teachers who had no problems at all. We decided to try packaging their expertise into training and handing it out to the other teachers!

The training mechanics were excellent, but the seminar itself had to change. We decided to cut the number down to six people per session, write a new methodology with a far heavier emphasis on practice, redo the design, and call it a workshop. That is where the real training began! As it happened, experienced teachers began training everyone else — and getting paid for it, too. In other words, our client is both the seller and the buyer.

Launching A/B Tests

In one of the experiments, the task was to show how lesson quality can increase the company’s profit. We needed to solve a teacher’s specific problem as quickly and effectively as possible. For example, when the teacher ends the lesson off the communicative method — no wrap-up, no stating the goals of the next lesson, no homework assigned.

This is a problem you solve neither with a course, nor a webinar, nor even a workshop. That means something very short, personalized, and precisely targeted is needed. We invent a new format — the case lesson. An expert meets the teacher for 15 minutes, and they solve this one case only.

In the first stage of the experiment, we proved that we can influence lesson quality through training. This means the teacher’s lesson got better in the methodological sense. An excellent result — but, as we found in the second stage of the experiment, it has little effect on profit.

The Experiment We Scaled

In another experiment, the task was to influence things so that every lesson the teacher and the student have scheduled actually takes place. This metric is called discipline; we would level it up — but with all the formats at once. We called it a training track: the teacher has to complete, at their choice, a case lesson, a workshop, or a self-study course.

The problem — the student skips. We launched an A/B test with a sample of 200 teachers. One group we trained; the other we left alone. The test was going well, but there were problems all the same. For instance, when we invited teachers to train through the channels we knew — an email newsletter, a message in a messenger, and on the platform — a month later we realized our reach percentage had fallen sharply. The teachers did not care, and that is a problem.

We had to save the experiment, and we did not know how. What if we asked the experts themselves to approach teachers with an offer to meet? We found a new technique for engaging teachers in the training track. The experts who run case lessons began writing to teachers on their own — and the teachers readily came out for a case meeting with them to solve the discipline problem.

Then the expert would book the teacher into a workshop and send over the remaining materials. Conversion grew sharply: teachers were completing the workshop after the case meeting. Because the teachers trained with experts in the case lesson and in workshops, their discipline metric grew by 3.3 percentage points relative to the test sample. We proved that rolling this training out to all teachers would grow the company’s profit by several million rubles.

Today we know how to successfully influence our teachers’ performance metrics and earn the company additional money. No less important, the training center team gained unique experience. No examples on the market of solving a particular case? The company’s current methods do not work? Time is limited?

Take a product approach — research everything in detail, prioritize every idea, plan every project, build a detailed action plan, design the experiment, count everything and make decisions at every step. It worked for us.

How to Get the Maximum

Set the goal of creating an internal training product (develop your employees and contractors on a regular basis). Study all the pains that keep your people from maxing out the key metrics (which specific business problem will you be solving). Find experimentally the most successful training format — the one that suits your particular employees and contractors (do not be afraid to invent your own solutions if the ready-made ones do not fit). Design the test in detail before launch (assess all the risks, think the analytics through, and calculate ROI). Launch several experiments per quarter (not all of them will be successful, so it is important to achieve a good enough result to scale at least one). Use internal resources and your employees’ expertise to create training content as relevant as possible to the company’s problems. Grow your company’s profit through training’s impact on the metrics.

Read more: https://rb.ru/columns/obuchit-8-tysyach-sotrudnikov/?utm_source=copy_out