Data Story — The Peterborough Social Impact Bond
Thursday, 24th August 2017
By Dan Miodovnik, Associate Director at Social Finance.
We launched the Peterborough Social Impact Bond to break the relentless cycle of re-offending by short sentenced offenders. To do this, we built a service that addressed their acute need, such as addiction, housing, and health.
However, we soon faced a peculiar challenge. Once we met their initial needs, people stopped working with us. This would happen regardless of whether they had stopped re-offending. Initially, only 37% of people accessed support for more than 1 month even though we offered it for 12 months.
We knew that if we were going to have a meaningful impact on their lives, we needed to improve the programme. This is when the CEO, David Hutchison asked me the question — ‘How can we improve the Peterborough Social Impact Bond?’
To answer it, I turned to data. I started with a simple philosophy — make data useful and never lose sight of the people we were trying to support. To do this, I needed to:
· Identify where data can help: I shadowed staff, met with service users, and visited the prison to understand what decisions the team made and where data and analysis could play a role.
· Decide what data to collect: I watched the team use our case management system and started tweaking to reduce the time they spent on the computer while still collecting the data we needed.
· Enable secure access: I worked with the Police, the prison, and others to enable our case management system to be accessed as and when needed (eg. car, prison, council office etc.).
· Build trusted algorithms: I wanted the team to understand every number, so we sat together and co-designed each calculation.
· Create a data culture: I spent countless hours helping the team incorporate our tools into how they worked and adjusted our analysis and models based on staff and service user feedback.
For people like Bryan (name has been changed) — our data philosophy meant a more personalised service. Bryan started drinking at 13 and three years later became homeless. He used prison for accommodation, the police for support, and the emergency room as a GP service. He had been in-and-out of prison over 150 times. Now at 57, he could not just stop offending.
We used data to identify important patterns in Bryan’s reoffending behaviour. These previously went unnoticed because no one collected the right data. Bryan lost his wife at a young age and never recovered. His reputation and fondness for cider meant that no one provided him bereavement support. So every year, he dealt with the anniversary by drinking himself into a bender and ended up in prison for anti-social behaviour.
In the weeks leading up to the anniversary, for example, we made sure that Bryan had access to increased support, particularly psychological counselling. This kept him out of prison and on the journey to breaking his cycle of offending.
Data became part of the service. We were able to track who was doing well and who wasn’t. We could tell if someone’s needs had changed.
Unsurprisingly, it earned me the nickname — ‘Data Dan’. As we continued to analyse what each person needed, the service improved — for example, the percentage of service users who accessed support for 1 month improved by 50% and the percentage of those who accessed it for 3 months increased by 100%.
Armed with data insights, the Peterborough SIB evolved to include over 10 different services such as legal and immigration advice, employment training, behaviour change support, a mental health service and more.
The result — a reduction in re-offending of 9% across the programme against a target of 7.5%.
The Peterborough SIB showed us the power of data. However, this was just the beginning. At the time, the social and public sectors were skeptical of data and technology. We kept hearing the same stories about products that did not start with the social issue. There needed to be an alternative.
We launched Digital Labs in 2015 to combine our understanding of social issues with digital, data science and product expertise. Together with governments and social organisations, we are developing person-centred approaches to capturing and structuring data and building data science tools to better understand and improve social and public services. Over the next couple months, we are:
· Children’s Services Learning Community: Launching a platform that enables Children’s Services departments to build, share and use tools with the National Pupil Database to improve decision-making and services.
· Leaving Well: Building a digital tool that helps authorities provide the right support to care leavers. It will enable increased accountability and communication between young people and the local authority so that young people can take ownership of their care leaving experience.
· Early Help Targeting and Triaging: Developing a model to help authorities to better target and triage Early Help provision at young people and families who are most at risk of entering care.
The last six years have been amazing. Who knew that an occasional trip up to Peterborough in 2011 would chart my journey at Social Finance — from the American intern to ‘Data Dan’ to setting up our Digital Labs. I can’t wait to see what impact we’ll have next.
