Service Innovation – Segment and Conquer
Supermarkets use data to sell us more of the things we want, and even things we don’t yet know we want – a real world example of service innovation through segmentation that we can learn from.
In social policy, we all know that there is no one programme or service that will work equally well for everyone in the target cohort. Even if it is having an impact across the board there will be some people for whom it works better than others and that’s where extra value can be squeezed out.
We might roll our eyes at the buzz-phrase ‘customer segmentation’, and of course there’s a difference between tailoring public services and selling sausages, but both require a similar approach to gathering data, analysing it, and in a sense letting it lead the way.
In the case of Tesco it’s about working out what shoppers want and selling it to them – a far easier job than convincing them to buy things in which they have no interest, a win for both parties. With public services it’s a matter of thinking in broad terms where we want people to end up – or not end up, as the case may be – and then letting what bubbles up from the data determine the most efficient route, and even the specific end point.
For example, working with one client that specialises in tackling youth offending, our data analysis found that though their intervention was effective overall, it was less effective at reducing offending among 12 to 13-year-olds than among young people of 15 and 16. By treating these two segments differently the overall impact of the intervention can be improved and more young people can be set on the right path at the right moment in their lives.
This approach challenges current orthodoxy which would have us determine our theory of change and set out clearly how we will achieve a given outcome before starting work. This can lead people to impose an analysis on the data after the fact, forcing it to fit the predetermined course. It also implies that all service users need more or less the same thing and we know very well that they don’t. The orthodox approach has its place, of course, once data has been collected and analysed, when we can start to make predictions based on prior knowledge.
Equally, it’s not efficient to design a bespoke service for every single end user, but there is a sweet spot in which we can identify sub-groups and thus wring out more value from programmes with relatively little additional time, manpower or funding. I’ll finish with another example: we have been designing approaches to impact management with a number of providers of universal services for young people and adult disability services. These agencies work with different sorts of people, with varying needs, and for whom different outcomes are desirable. Advanced statistical analysis can help us identify groups within that complex body and lead to service innovation which is both tailored and general.
