I’d never seen a corpse before. I stepped carefully to avoid blood and body parts as people ran from the terrorists, who proclaimed their allegiance to a banned far-right group. The Armed Response Vehicles arrived within 6 minutes of being called and snipers scaled the roof, but already several people had been injured and some killed.
Fortunately, my experience was as an observer of a simulated terrorist attack conducted by the Metropolitan Police, Ambulance Service and Fire Brigade. Operational and Tactical Firearms Commanders observed closely, trying to find ways to improve and to save more lives.
The planning that went into that simulation, one of several annual exercises, was meticulous, as you’d expect from highly trained professionals working to protect one of the busiest and most complex urban environments on earth. But it did give me pause for thought: what is the equivalent for everyday crimes?
The breadth and scale of the challenges of policing a city like London are staggering: the Metropolitan Police respond to an average of 7,000 emergencies every 24 hours, police 100 large public events every single day, and begin 250,000 new investigations every year, in a city that speaks 270 different languages – and that’s just a part of the picture. A service like this requires an astonishing variety of skills and capabilities. As such, the Police cannot afford to run simulations for all its activities. However, the Met keeps an enormous trove of data – detailing every incident, offender and reported crime in its vast jurisdiction. With the right skills, this data can be of immense value in explaining the world around us and finding new and better ways to police it.
For the last year, I’ve been embedded within the Metropolitan Police to establish and lead a new function – the Strategic Insight Unit (SIU) – whose remit is to fight crime with cutting edge analytical techniques. Powered by our lead academic partners at LSE’s Centre for Economic Performance (as well as UCL’s Jill Dando Institute and Cambridge University’s Institute for Criminology), SIU brings together data-science expertise with behavioural insights and the operational know-how of frontline cops. We report to the Commissioner – Dame Cressida Dick – and Board of the Metropolitan Police in rapid cycles before working with the frontline to trial new ideas.
For operational reasons I’m not able to share much of our work, but just as our report for the Mayor of London’s Violence Reduction Unit showed, to really get under the skin of crime patterns, we need to conduct analysis at the hyper local level. Take burglary as an example: fewer than 1 per cent of addresses account for almost a fifth of all residential burglaries and the most burgled Lower Super Output Areas had 26 times more burglaries than the least.
Burglaries per 100 households over the last 8 years, by LSOA
Though there been some recent, high profile burglaries related to overseas crime gangs, looking at 8 years of data we find that a third of burglaries occur within just one mile of where the burglar lives (over half occur within 2 miles). The behavioural principles of ease and familiarity seem to apply as much to burglars as anyone else.
The findings of the SIU’s work are relevant for the Met’s community partners and the wider Criminal Justice System too. For example, we can identify young people most at risk of committing a serious offence within a few years, the first time they are arrested. Using this information to prioritise support to those young people who need it most should generate a good return on investment since, in 2018, almost 1 in 2 people arrested in London had previously been arrested as a child. As a first step, we will launch an RCT to assess the impact of additional support to parents or guardians of children in Police custody – watch this space for the results.
There is much to gain by bringing our best analytical minds and latest tools into contact with Police data, which is why many research institutions are keen to access it. But the best ideas for policing will often come from within policing. To exploit them we need to create the right environments within forces: blended teams where data scientists work closely with officers who are trained to understand analytical methods; can ask the right questions; interpret and challenge their output; then work with behavioural scientists, engineers or entrepreneurs to design new approaches, tactics or even products to trial. Pooling data from multiple forces and other organisations will present even more opportunities.
There are still bureaucratic and cultural hurdles to overcome but London’s Metropolitan Police have made an important step towards unlocking the full potential of its data and its people. I’m proud to have been a part of it.
 i.e. a London neighbourhood of 400 – 1200 households.