Some eight years later, he is an enthusiast for the school improvement policies of Birmingham's charismatic chief education officer, Tim Brighouse, backed by a Blairite Labour council so committed to raising standards that they increased their education budget this year while all around were being cut.
If school improvement is to make an impact then it has to work in schools like Shenley Court in Birmingham's southern suburbs which, Mr Dennis is the first to admit, still has problems even after eight years of steady improvement. It is a 1,660-strong mixed comprehensive which loses many of its brightest potential pupils to the city's independent and grammar schools and a lot of girls to single-sex schools.
Even so, it is oversubscribed and takes in the full ability range. "I think we are beginning to be seen as an improving school," he says. Last autumn's OFSTED inspection confirmed that Shenley Court was providing a "good education for its pupils", in an environment where "the joy of learning is evident in a significant proportion of lessons".
But it has been a long hard struggle. The school draws a high proportion of its pupils from housing estates where, Mr Dennis says, there is worse deprivation than in many inner-city areas but no corresponding government help. Getting on for a quarter of the intake arrived this year with reading ages of below nine years - something which in future should not happen under the city's school improvement programme.
Five years ago, when Mr Dennis first began to collate statistics systematically, all of Shenley Court's difficulties were reflected by exam results which put the school way down most league tables. In 1989 only 16 per cent achieved five good GCSE grades. In l990, the proportion had dropped even further to 14.5 per cent. But what worried Mr Dennis most was the fact that more than a quarter of the pupils achieved no grades at all because they did not bother to attend school often enough to either enter or take their exams.
"Our major difficulty is a huge non-attendance problem, which reflects the low aspirations of many families in this area, and the feeling amongst the youngsters themselves that if they are not going to get good grades then there is no point in bothering at all. That is what our school improvement policy is aiming to change."
By last summer, the proportion achieving five good grades had been boosted to almost 26 per cent; those with no passes at all reduced to 20 per cent. Not impressive by the standards of the leafy suburbs, Mr Dennis agrees, but a significant improvement. The target for this summer is 30 per cent with five A to Cs, which will put the school in sight of the national average, and a further drop to 17 per cent in the proportion of non-attainers.
How is it done? Mr Dennis is generous in his praise of the Tim Brighouse approach, which, he says, has raised teacher morale and provided practical policies to help schools improve standards.
There are essentially five prongs to Birmingham's strategy. The first is to establish a common language about the seven school processes which the local education authority believes need to be addressed: leadership, management and organisation, the school environment, collective review, teaching and learning, staff development and parental involvement. Governors and parents are central to the whole approach and are included in many of the meeting and training sessions.
The strategy has also set up an "innovation exchange" to enable schools to learn from each other about effectiveness and improvement; it has supported national and local initiatives to improve monitoring and measure standards of achievement, and is building up networks of support for school improvement in the wider community.
But central to the whole strategy is the development of "guarantees" - there is some discussion about the exact nomenclature: some prefer commitments or entitlements. Guarantees are in place for early years, primary and secondary pupils. Essentially these balance LEA commitments to improved funding, information and support against specific school commitments to pupils and measurable targets for outcomes at each stage.
The primary guarantee, for instance, promises every child residential experience, the opportunity to engage in a "public performance", and support for every child's talent in the expressive arts. In return the school must audit the percentage of its pupils reaching given stages in reading and maths, and set targets for improvement.
At secondary level, the guarantee offers all pupils the opportunity of involvement in a range of curriculum areas. For their part, schools are expected to audit attainment at the age of 12, to set improvement targets in the core subjects at the end of key stage 3 and in public exams at 16, to have enabled all pupils to produce a record of achievement and enabled them to make the transition successfully to the next stage of education or training.
The targets are undoubtedly the most controversial aspects of this policy, raising spectres of government ministers imposing impossible norms on struggling schools. The Birmingham approach is not quite like that. First, the targets are set by the schools themselves and are not compulsory. Second, they are set against its previous best performance - in other words, they are realistic and take account of historical and social factors. And third, the LEA is putting substantial statistical effort into enabling schools to measure their "value added" input, to monitor trends and to compare their performance with schools in similar circumstances rather than with inappropriate averages.
All this, Keith Dennis thinks, will help make target setting a useful tool in the real improvement of schools like his which started from a low base in terms of exam results. Even so, he says, he set a target for GCSE results last year in his own mind which he did not communicate to his staff. Which is just as well, perhaps, as they didn't reach it.
This year he is being more up-front. The 30 per cent of the pupils he hopes will get five good GCSEs is high for an urban comprehensive with neighbouring grammar schools. To get there, Shenley Court has identified pupils who seem to be on the C-D borderline, has discussed the value of good results with them and their parents and offered extra help where it is needed.
At the same time, the school is hoping to push down the proportion of youngsters who fail to get any qualifications through the introduction of GNVQ courses more likely to motivate those who find it hard to do well at GCSE.
Mr Dennis still regrets the loss of vocational courses when the national curriculum was introduced and welcomes their reintroduction now. Like Tim Brighouse, he is an enthusiast for Professor Howard Gardner's definition of different intelligences, and does not believe a rigidly academic approach suits all children.
In some ways target-setting is relatively easy to accept in secondary schools which are now used to having their exam results published. For primary schools, this is a whole new world, but one to which some Birmingham primaries are enthusiatically adapting.
Ron Morton, head of West Heath Junior School, on the southern edge of the city with a catchment area which spans middle-class suburbia and a council estate so dire that it is about to be demolished, is a keen supporter of the LEA's policies.
In drawing up profiles of his 350 children he is using both seven-year-old national test results and the baseline assessment data which is compiled in nursery schools and classes three years before the children reach him.
"What I would really like to do is see if we can take the end-of-year attainments of all the children and see if we can improve them year on year. But the important thing about all of this really is the debate which has come out of the suggested mechanisms. The whole city is now talking constructively about how we can move forward. Expectations are becoming much more rigorous. "
The process is not simple, Mr Morton says. Discussing reading attainment with his head of English, they concluded that teachers needed in-service training to help them define attainment in a consistent way at the same time as children's performance was targeted for improvement. The primary guarantee expects the schools to set improvement targets for reading and maths but at West Heath this has been extended to include all the national curriculum subjects. Even that, Mr Morton says, is not enough because, as the National Primary Centre which has given the school a good practice award argues, attainment is even wider than that.
Meanwhile the city-wide campaign on school improvement already has parents asking for hard evidence of progress which many teachers might quite recently have found threatening, but are now beginning to accept as perfectly normal. David Wood, the city's chief adviser, admits that it has not all been plain sailing.
Quantitative measurement of attainment does worry teachers, he says, and there has been some resistance to the improvement programme.
"But in the end it is all about creating a climate of change, winning hearts and minds, and I think we are succeeding in that. After all, if you say to any school you can improve on your previous best, no one can really claim that it is out of reach. We are not bashing the schools with impossible norms. We are asking them to set their own targets.
"And in the end, if a city like Birmingham could get all its 11-year-olds up to their chronological reading age, that alone would make a dramatic difference."