Nearly a third of all schools are now academies or free schools - a fact that is striking enough. Two-thirds of academies are part of the 818 multi-academy trusts (MATs) that have two or more schools - a fact that is even more striking.

I’ll come clean: I have always believed that MATs present a great opportunity to combine deep school-to-school collaboration with formal governance, along with hard-edged accountability.

Previous school partnership initiatives have delivered varying degrees of added value but too many have been overly reliant on the individual heads leading the initiative and have faded or fractured when a key player moved on. Crucially they have, as a rule, also lacked a framework for partners to hold each other to account.

MATs should be the vehicle to address these shortcomings while consolidating the advantages of school-to-school working.

But the evidence suggests that too many are still struggling to fulfil their promise. You can argue that we should not expect too much too soon. After all, the vast majority of MATs are small and most only got going during or after 2013. Yet such evidence as we have provides cause for concern - and I say that as a friend of the MAT movement.

Yes, there are some excellent high-performing MATs, but the evidence - taken as a whole - points to problems in three key areas.

First, there is governance. Although many MATs govern well, there are weaknesses that are too extensive to be dismissed as exceptions. The unbridled expansion of early MATs, well-documented failings in overseeing trust finances and academic standards, an imbalance of skills on MAT boards and, in some cases, a lack of clarity about board and local governing body responsibilities indicate that all is not right.

Secondly, there has been a worrying number of compliance issues, such as fraud, deficits, employment of family members and inappropriate payments and contracts. In 2016, a total of 23 trusts were involved in inappropriate related-party payments and contracts totalling over £4 million - mostly to companies owned by their trustees.

The third problem relates to performance. Researchers, in general, have struggled to identify a positive “MAT effect”. The performance tables for MATs show that, on average, MATs are pretty average. At key stage 2, they do a bit better than other schools in terms of progress on writing and maths but are below average on reading. At KS4, over half of MATs are seriously below average on the Progress 8 measure. Those averages conceal another truth: there are significant variations within and between MATs.

Furthermore, the advent of regional schools commissioners (RSCs) has led to “rebrokering” becoming a mini-industry, as they grapple with what to do with those academies declared “inadequate” by Ofsted - the so-called “untouchable” schools.

Some might conclude that we should abandon the academy experiment. But that would be a monumental mistake. It would be both hugely disruptive and a waste of all the effort and lessons learned so far. It would fail to acknowledge the plus side of the MAT account - take, for example, the work of MATs in turning around schools that have for generations failed young people.

Consider how many MATs are systematically developing and deploying leadership talent. Or look at how the best MATs are managing school improvement at scale by testing, refining and applying teaching and learning approaches across groups of schools.

Politicians and policymakers should, therefore, stick with MATs but reboot their strategy. Here is my five-point plan:

1. Shared vision

Build a shared vision of what great MATs look like - particularly in terms of improving teaching and learning, but also effective governance. MATs exist to improve outcomes for young people. Continually improving their theory on how to achieve this across groups of schools should be their all-consuming mission. Too little of the discourse about MATs is about their core business - better teaching and learning.

2. Research best practice

Commission and establish the research evidence to underpin that vision. We have a fair amount of guidance on effective governance, but is there a case for learning from the corporate sector and increasing the number of MAT members and strengthening their remit as a means of holding MAT boards to account? We urgently need to address the scandalous failure of the Department for Education to commission any systematic research on the strategies of the most successful MATs.

3. Build solid foundations

There must be much more organisational capacity-building available for MATs and their leadership teams. The DfE has come to this party far too late in the day. There are now a number of programmes that help MATs meet the challenges of developing as sound organisations. However, there is still a huge amount of wheel-reinvention going on.

4. Take things slowly

Adopt a more measured pace in growing MATs. I understand the need for MATs to reach a viable level of operation, but experience points to gradual growth being more likely to deliver the sound systems and depth of working necessary to make MATs effective in the long term. The government should only promote expansion prioritising performance.

5. Leadership

Foster a stronger voice and role for MAT leadership. MATs are meant to be part of a broader move towards a school-led system of school improvement. But the unrelenting focus on upward accountability to RSCs and the DfE is undermining the sector’s maturity. We need MAT leaders sharing their experience and resources and working together to shape the agenda and address problems. The MAT health checks that are being piloted are a good first step in this direction.

These ideas would help MATs to fulfil their potential and deliver consistent improvements across the whole sector. And they should also help to shift the perception of MATs. So now is the time for a real step-change in the performance - before it’s too late.

Robert Hill is a former ministerial special adviser. He is now a consultant, researcher and blogger on MATs