The tyranny of classroom observations

Classroom observations and school visits are essential for providing high-quality teacher coaching. However, these visits also impose high transaction costs.

Over the past academic year, our team has conducted 3,000 classroom observations. With most of our teachers receiving at least 2 observations each month. This frequency of class visits and high-touch support is way beyond what any other coaching programme offers anywhere in the world. For example, in the UK, most schools conduct a maximum of 3 observations per teacher per year [1].

Why do we need classroom observations?

While elsewhere, classroom observations are geared towards standards enforcement and to meet requirements for head teachers, our visits are designed to collect data to make teachers lives easier and happier.

Our coaches visit classrooms and observe lessons to help teachers promote student centered learning. This is mostly a behaviour change initiative and involves a lot of habit formation. As with every new habit, adapting to new teaching methods takes time, practice, patience and feedback. So, high-frequency teacher and coach touchpoints make a lot of sense.

Classroom Observations act as manual, in-person push notifications to enable implementation.

- Overheard during our internal brainstorming

Furthermore, these visits also serve a key role to help us internally improve the secret-sauce of our coaching continuously with immediate feedback from each classroom in our partner schools.

What are the transaction costs?

1. Scheduling conflicts, teacher absences and other school related issues lead to significant time and productivity loss, not to mention general annoyance and irritation.
2. Almost 70% of our coaches’ time every week is consumed by classroom visits, the rest of the time is spent on feedback sessions, data analysis, stakeholder management and reporting.

At scale, these bottlenecks can lead to exhaustion and tedium among coaching staff and low team morale which could in turn result in low-quality observations and feedback. If ever there was a problem to be fixed…

At this point, any self-respecting management consultant would start arguing about pareto principles, low-hanging fruits and quick wins. Clearly, any improvements to the frequency and quality of our classroom observations would result in elephantine gains.

Potential solutions

Our team has expended considerable time and effort to identify solutions to this problem. Naturally.

We have researched alternatives applied in school districts in the US, ideas proposed at a seminar hosted by Her Majesty’s Chief Inspector (pour quoi pas? I guess.), coaching models from Brazil and also white papers from independent thinktanks. We have adopted Human Centered Design principles and have generated copious lists of original ideas during the ideation phase.

Despite all this labour, solutions are still far from clear nor easy to implement consistently.

Over the course of the next few months, our team will be trialing combinations of the following ideas:

1. Using existing “expert teachers” in our schools to conduct some peer-observations. These are teachers who have completed our Level 1 program and have shown willingness to mentor newer and less experienced teachers.
2. Availing services of school leadership. We already train school leaders to take more ownership of teacher performance and some school managers are already conducting observations using our mobile tools. Scaling this to all schools could ease our load.
3. Student led observations. This is an idea implemented in a few school systems around the world. There could be many externalities (+ and -). We need to be exteremely careful how we roll this out.
4. Self-reporting by teachers. One main problem of self-reporting of anything anywhere would be the eternal principle-agent problem of misaligned incentives. Designing this would need a bit of trial and error and the right incentive framework for accurate reporting.

We also have some additional interesting ideas on the table that are difficult to quickly prototype. These interventions involve more resources, time and technical capacity, additional complexity.

1. We hire specialised data collection teams to conduct observations
2. Schools hire/assign specialised data collection teams
3. Predictive statistical extrapolation of teacher performance and goal-setting (this will be the future!)

Any others we might have missed?