Few professionals are observed as frequently as teachers. Teachers are always watching each other - reciprocally, in pairs or in groups, often to research a particular aspect of practice. Increasingly, they also engage in self-observation and analysis using video and other techniques.

According to the OECD, four-fifths of teachers in England have been involved in observation and feedback for professional development. Nine in 10 have been observed for formal appraisal. The three-hour limit to classroom observations for state schools has been shelved and there is now no limit to how often or for how long a teacher can be formally observed.

Which approaches are the most effective?

We have scant evidence on this front. The Education Endowment Foundation has commissioned trials, but a demonstrable impact on pupil outcomes is yet to be proved.

The wider research suggests that training is essential to ensure people look at and see the same things. A tight focus should be agreed for explanation or questioning, and this should include evaluating pupil progress.

What else do we know?

Observers need to be cautious. Even the best-trained observers agree only half of the time on what great teaching looks like, according to one review. That’s as a good as tossing a coin. Great teaching is a complex craft. We see activities and interaction, but learning is usually invisible to the human eye.

If it’s so difficult, is it worth doing at all?

It is tempting to conclude that observations are not worth the effort involved. This is particularly true if you consider the opportunity cost. What else could those involved be doing that might be more beneficial for pupils?

Well-delivered and sustained approaches to professional learning can, though, improve teachers’ effectiveness. Observations should primarily be used as a formative, developmental process to create self-regulating professional learners. Inferences from observations must be triangulated with other evidence - pupil progress data and other information, such as student surveys or consultations - to provide an overall picture. You should avoid grading specific lessons, as Ofsted now emphasises. Observations are not reliable enough on their own for high-stakes performance reviews or appraisals.

Observation is most effective when undertaken as a collaborative and collegial exercise, but there should be effective challenge - constructively involving expert teachers, senior managers or external experts.

And finally…

It’s important to see lesson observations as the start of the learning journey. Teachers should adopt the same principles for teacher feedback as they do for providing feedback to pupils. Some advocate “professional learning cycles”. But the focus must be kept firmly on pupils’ learning.

Lee Elliot Major is chief executive of the Sutton Trust and Steve Higgins is a professor of education at Durham University. Together, they authored the teaching and learning toolkit, now the Education Endowment Foundation toolkit