I’ve spent a large part of my working life in Microsoft business intelligence. I was at Microsoft when we announced the acquisition of ProClarity on 3 April 2006, so my interest in helping people make sense of data goes back a fair way before today’s Power BI.
That background runs through Edequity. Getting data out of systems, bringing it together securely and producing numbers people can trust is our DNA. We care about the definitions, the calculations, the privacy controls and the quality of the reports.
But I’ve come to think of reports and dashboards rather like exams: necessary, but no longer sufficient. The result tells you something you need to know. Understanding what it means, deciding how to respond and finding out whether that response helped all require further work.
I want Marginal Gains to help colleagues carry that work through, from making sense of the information to agreeing a response and learning from what happens. That has shaped what we have built around the reports.
Getting the data together is only the beginning
Things are still too hard. The work of Open Education AI has helped give the sector shared foundations for bringing data together, and that matters enormously. Even with those foundations, someone still has to understand what the information means in their school.
Take a report showing rising absence. You need to know which pupils are affected, whether the pattern is new, what has changed and what colleagues already know. An unusual result might be a recording problem, a change in circumstances or something that needs attention quickly.
You also need to understand the measure itself. Who is included? Which period does it cover? Are two schools calculating it in the same way? A polished chart cannot resolve a difference in definitions.
This is why being able to find the right information, look into an unexpected result and check how a number was calculated belongs alongside the report. It gives the people discussing it a sound starting point.
Meaning needs knowledge, experience and conversation
Once you understand the issue, you still have to decide what to do about it.
That means drawing on evidence about approaches that have worked, the experience of colleagues and knowledge of the pupils and families involved. You need to consider whether an approach makes sense in your context, what it will take to put it into practice and who should be involved.
For an attendance concern, that could mean bringing pastoral staff, the attendance lead and someone who knows the family into the same conversation. Each may hold part of the explanation. The people closest to the issue need a way to contribute, and everyone needs to understand what has been agreed.
Software can make that work easier to organise. It can keep the relevant information, evidence and discussion together, so colleagues can spend more time thinking about the response and less time assembling the background for it.
A decision needs a plan someone can carry through
A useful discussion should lead to a clear next step: what are we trying to improve, what will we do, who will take responsibility and when will we review it?
The measures that matter will sometimes come from the systems you already use. Sometimes they won’t. A school improvement priority might require you to track whether an agreed support plan has been put into practice, or whether reviews are happening when they should. Those measures need a definition and a way to collect the evidence, even if no existing system produces them.
We also need to agree what improvement we expect to see for pupils, and when. Knowing that a plan was carried out is different from knowing that it helped.
We have built Marginal Gains to connect that work to your school improvement priorities, with the ability to define your own key performance indicators rather than limiting the plan to the numbers already available.
The plan also needs evidence from the work itself. That might be an observation, a note from a conversation, a document or a photograph captured while someone is carrying out an activity. Being able to capture it where the work happens, attach it to the relevant plan and share it with the right colleagues makes it easier to keep a useful record.
Communication continues throughout. Colleagues need an update they can act on, leaders need to see where help is needed, and governors need enough context to ask useful questions. This is why we have included tools for producing documents in a school or trust’s own templates. In governance and SEND work, keeping the evidence, agreed actions and reviews together helps the people responsible prepare their next update. The aim is to use the information you have already gathered without having to reconstruct it for each audience, while still reviewing what is shared and who should see it.
AI should help us investigate and challenge our thinking
Marge, our generative AI assistant, sits within this work too.
One of the most important things behind Marge is the Logic Library. It contains the definitions and calculations behind our measures: what a number means, what goes into it and how it is worked out. Using those shared definitions is what makes the figures consistent and their calculation available to check.
The Logic Library supplies the calculations; Marge helps you investigate and explain the resulting figures. With that context, it can help you explore a question, extend the analysis and challenge your own interpretation.
I want that to make us more thoughtful about the evidence. An AI explanation still needs checking, especially where a decision affects a pupil or family. The local knowledge and responsibility remain with the people doing the work.
We need to learn from what happens next
The cycle is incomplete until we review the impact of the action and bring what we learn back into the next decision.
I’m grateful for Inspiration Trust’s thought leadership on attendance prediction. In a CST article by Open Education AI’s Amy McJennett, the Trust’s work shows how data can help identify pupils who may be absent the following day, giving staff an opportunity to respond earlier. The article also describes their exploration of bringing intervention records back into the data architecture.
That follow-through is central to what we are building. We want to connect the insight to the intervention, record what was done and follow how the data changes over time. We also need to understand why an approach seems more effective in one setting than another.
An improvement after an intervention is a reason to investigate, rather than proof on its own that the intervention caused it. Who received the support? Was it carried out as intended? What else changed? What do staff, pupils and families tell us? The record of the action needs to sit alongside the numbers if we are going to make sense of them.
That gives the next review something useful to work with. We can decide what to continue, what to adapt and what to stop, and feed that learning into the next plan.
Why we have kept building
My aim is for the next conversation to start with a clear account of what we understood, what we did and what changed. The report, the reasoning, the plan and the evidence all need to stay connected for that to happen.
I’ll unpack these aspects of Marginal Gains in subsequent articles and posts, and show how we are building around the work people do to improve outcomes for young people.