Ofsted has said that, from next term, a computer algorithm will be used in deciding whether to inspect good and outstanding schools.

In a methodology note published today, the inspectorate says that rather than using predetermined thresholds on performance data, as in previous years, it will use “supervised machine learning” for the first stage of decision-making from the summer term of 2018.

It says it has created an algorithm that “has effectively produced a probability of a forthcoming inspection being less than good”, which is known as the “raw risk score”.

“We believe [our new methodology] will improve our capacity to identify concerns about performance,” the note states.

The machine-learning algorithm was created by examining data on progress and attainment, school-workforce data and parent-view responses to see how it fitted with known inspection outcomes.

After the first-stage judgement, a desk-based review will explore any concerns brought to the inspectors’ attention.

The watchdog adds that “in no way do the algorithm results impact on inspection judgements”.

“We use risk assessment to ensure that our approach to inspection is proportionate and to focus our efforts where they can have the greatest impact,” states Ofsted. “We also use risk assessment to identify good and exempt outstanding schools for which we have concerns about their performance.

“It is important to note that the risk-assessment process is not used in any way to prejudge inspection outcomes, and inspectors do not have access to the risk assessments when inspecting schools.”

Outstanding schools are exempt from inspection unless there are concerns about performance or issues such as safeguarding. But the inspectorate has said it is aware concerns exist that some schools are then going for long periods without inspections.

Want to keep up with the latest education news and opinion? Follow Tes on Twitter and Instagram, and like Tes on Facebook