Some years ago, I was responsible for finance in a trust of thirteen special and alternative provision schools. I had a suspicion I could not shake about the pupil to teaching assistant ratio in one of them. The number looked wrong to me.
I raised it with the headteacher. What came back was a single sentence: you can’t compare us, we’re different to other schools.
That ended the conversation. Not because they were being difficult, but because they were right.
I had not gone in empty-handed. We held comparison data across all thirteen schools, and I could see exactly where each one sat against the others. I thought that was evidence.
It was not, and it took me a long time to understand why. Every one of those schools could reasonably claim to be different, because every one of them was. With that much variation, any school I challenged had a legitimate answer ready, and I had no way of telling a genuine difference from a habit nobody had examined.
There was a deeper problem I could not see from where I was standing. I was measuring each school against the average of the other twelve, and nobody had ever tested that average against anything outside our own walls. If we were collectively staffed unusually, then my baseline was the outlier and every judgement I drew from it was wrong. I was checking our schools against ourselves and calling it proof.
It went quietly away, the way these things do. I never found out whether I was right. Looking back, I suspect the headteacher was.
What I got wrong was not the suspicion. It was assuming that more comparison would settle it. What that conversation actually needed was a peer group we could both accept, which for a special school means schools with a similar profile of need rather than simply a similar size or place. Put that in front of a headteacher and “we’re different” stops being the end of the discussion and becomes something you can test together. Either the school sits with its genuine peers or it does not, and both answers move you somewhere.
I now spend my working life on this problem from the other side, and I have made a version of the same mistake there too. Last year we produced a benchmarking output I was pleased with. Every dimension, every level of detail anyone could ask for. The feedback was that people could not find the answer in it. We had given them depth when what they needed was a starting point they could hold up in a meeting. That is the error I made as a CFO wearing different clothes: treating more data as though it were the same thing as a shared basis for a decision.
So, I have stopped thinking of benchmarking as an analytical exercise. What failed in that room was not analysis. It was that two people with different expertise, both acting in good faith, had no common ground to stand on. I had numbers that meant something to me. The headteacher had knowledge of their children and staff that meant something to them. Neither of us could reach the other, so nothing happened.
Comparison data earns its place when it gives those two people something they can both accept. That is a relationship problem before it is a data problem, and it is why the audience for benchmarking is so rarely the person who produced it. It gets used in trustee meetings, in conversations with heads about staffing, in boards asking whether central services earn their keep. If it only works in the hands of the person who built it, it is not working.
I am not going to tell you there is a right number of teaching assistants for a school. There is not, and anyone offering you one is overreaching. What I would ask instead is the question I never properly asked myself: how do I know whether resource in this trust is allocated in the most effective way, and could comparison prove or disprove it? If you cannot answer that, the suspicion goes back in the drawer, and you never find out who was right.
- Warren Porter is Head of Product at IMP Software. IMP published this year’s MAT Finance Insight Report at the CST's 2026 Annual Conference, where it was a headline sponsor.
We welcome perspectives from a diverse range of guest contributors. The opinions expressed in blogs are the views of the author(s), and should not be read as CST guidance or CST’s position.
