School leaders are still facing problems with the government鈥檚 AI-generated attendance reports, despite them being re-issued after errors 鈥 as a government leader cautions 鈥渘o model is perfect鈥. Leaders who spoke to Schools Week said they had been told to learn from schools hundreds of miles away, their own school, or in some cases were not linked with any settings at all. It comes as polling shows 80 per cent of headteachers think pairing up schools won鈥檛 make a difference to their attendance levels. How do reports work? Schools were issued new versions of their government attendance baseline improvement expectation reports last week, after the first reports were found to contain mistakes. The reports include an AI-generated target to raise attendance levels, and rank schools within a group of 21 similar settings. Schools are then given four 鈥渂etter-performing鈥 schools that they are encouraged to network with and learn from. Schools are grouped together based on factors including the proportion of pupils eligible for free school meals, with social, emotional or mental health needs, English as an additional language, or in an urban or rural setting. A mock report shown to teachers by the Department for Education The told teachers earlier this week this 鈥減rovides an intelligent alternative to judging attendance solely against the national average鈥, which 鈥渋ncludes schools in very different circumstances鈥. 鈥淣ot only does this provide you with greater transparency on the schools we鈥檙e comparing you against, but it also provides an opportunity for you to reach out to those schools and create networks,鈥 the DfE added. ‘It’s quite valuable to see in person’ Stephen Ferguson But Stephen Ferguson, deputy headteacher at Kingsmeadow Community School in Gateshead, said although the similar schools was a 鈥渞eally positive feature鈥, most schools listed in his report were in London or the south. They also all belonged to multi-academy trusts, whereas Kingsmead is a local authority school. 鈥淚t鈥檚 difficult to really do any sort of collaborative work with schools around a feature like attendance, if actually the similar-schools report generates schools that, once you get under the surface, aren鈥檛 quite similar 鈥 geographically, it鈥檚 hard to get there and work with them. 鈥淵es, we can do it online or over the phone, but often the systems have multiple layers and depths to them. It鈥檚 quite valuable to see in person.鈥 Other schools, including Dartmouth Academy in Devon and Trinity Academy St Edward鈥檚 in Barnsley, were ranked among the top of their similar school groups. They were told to go to their own school for advice. Mark Allen, principal of Trinity Academy St Edward鈥檚, said: 鈥淏ecause it鈥檚 AI-generated, you lose the kind of human intuition 鈥 that鈥檚 disappointing, you鈥檙e invested in this and yet it鈥檚 telling me to look at my own school for help.鈥 ‘That’s no context’ Andy Goodwin, safeguarding and attendance lead at Dartmouth Academy sponsor Education South West, said the reports 鈥渁re on the journey toward what I think a lot of us have been crying out for, which is 鈥榩lease tell us who contextually we are similar to鈥欌. However, he added the reports have 鈥渟ome massive weaknesses鈥 and leaders 鈥渄on鈥檛 fully understand how they鈥檝e made the decision about the context鈥. Another of the trust鈥檚 settings, a technical college for 14- to 19-year-olds, was told to learn from other 11-18 secondary schools. Robert Coles 鈥淭hat is no use, that鈥檚 no context,鈥 Education South West trust chief executive Robert Coles said. Nigel Attwood, headteacher at Bellfield Junior School in Birmingham, said his latest report did not include any schools to go to for support. The attendance data on the report was also wrong, he said. In the school鈥檚 initial report, Attwood said he had been recommended another school with about 60 per cent of its pupils receiving pupil premium funding 鈥 the same as his school. But others recommended had half the proportion. 鈥淚 think it鈥檚 another document that鈥檚 costing money that we could have used somewhere else,鈥 Attwood said. ‘No model is perfect’ According to Teacher Tapp, the sector agrees with the leaders鈥 concerns. In a poll of 10,000 school staff, 80 per cent of heads, 82 per cent of teachers and 76 per cent of senior leaders thought pairing schools wouldn鈥檛 make a difference. Sir Kevan Collins Sir Kevan Collins, the DfE鈥檚 lead non-executive director, told a teacher webinar about attendance on Tuesday there is 鈥渘ever, ever a perfect match鈥, and that the scheme was still in its test-and-learn phase. 鈥淥ne of the things we want to do in this first year is test this with you. We want to find out whether we鈥檙e matching you with schools you recognise as your own, or whether we need to make adjustments. No model is perfect.鈥 The DfE said while school region is given 鈥渁ppropriate weight鈥 when forming the similar school groups, technology allows schools to 鈥渢ranscend location鈥. It added that if a school sees their own name as a high performing school, it 鈥渋s not a suggestion they should contact themselves for support鈥.