We left classroom teaching in 2007 to set up a tuition business in Bath. We taught from a converted shop, which became a small classroom equipped with the best maths resources I could lay my hands on. The business quickly took off, and soon we were teaching more students than we could manage, so the natural next step was to employ a tutor. And then a couple more. By 2008 the small classroom was full every evening and every Saturday morning, but parents were still knocking at the door.
We opened a second centre on the other side of Bath, and ran it remotely, employing tutors and a centre manager. Not quite as successful and the first, it still saw over 100 students per week pass through its doors. But by 2010, the business was stagnating. For a while we couldn’t understand it: we’d experienced 3 years of rapid growth, and we envisaged opening a third and perhaps fourth centre – after all, we were still only hitting 1% of the addressable market in Bath. Why had we stopped growing?
The answer is simple: the students being tutored in 2010 were receiving good tuition, but not the outstanding tuition that our first customers had received. As the business scaled, it became harder and harder to find really good tutors, and harder and harder to ensure our initial exacting standards were being met. And you are never under more scrutiny than when you are teaching a child maths. If we had opened a third centre, it could have been catastrophic for our business.
As I have said in previous blogs, the single most important factor when it comes to raising standards in maths is to ensure that every child is working at their threshold of understanding. And this is harder than you may think. Tutors do not do this reliably – even when you put processes in place, they need to be followed. The biggest issue we faced with tutors was the constant defaulting to teaching children what they were doing at school, which, by-and-large, was not what the child needed because it wasn’t their threshold.
As we build DoodleMaths, the opposite is happening. Increased revenues mean we can invest more and more back in to creating the resources that work best for a child. The vast data we collect is vital to driving this: last month, 600,000 questions were answered. We know exactly what the average time taken to answer each question is, which questions are too hard, which ones are too easy. This allows us to refine our work program with greater and greater accuracy. And outcomes are immediately tangible: we can demonstrate progress transparently to parents, teacher and children in real time, so if a child is finding something difficult, there should be no reason for them to sweat on it for weeks at a time
As our technology-based tutoring system evolves, it will become better and better through scaling in a way that our human-based equivalent never could.