How Advances in Crime Prevention Could Be Used to Optimise Paid Ad Campaigns

Last week, I attended an interesting presentation from Toby Davies who works in the Crime Science department at University College of London. It was about analysing crime data to predict and prevent future incidents of crime. UCL has been working closely with the police on high-frequency crimes (such as burglary) and have found that even with small improvements of 2-3%, the changes to policing have been very low cost, and have had a big impact in high crime areas.

We decided to attend the presentation, not because of a dormant passion to police the streets, but because of the technology’s potential to optimise paid ad campaigns through to the improved targeting techniques.

Crime Theory

Analysing crime data is a relatively new practice starting around 2005. To improve crime analysis, there are a few tools which are in use. One such tool is PredPol, a platform to prevent crime, keeping affected communities safer. PredPol significantly outperforms crime analysts, and only requires three data points: crime type, location and date/time.

  • Space Time Analysis

Crime theory demonstrates the link between time and space. Related incidents will often be concentrated close together and close in time. For example, if two crimes occur in a 200m radius within 30 days, it is 1,000 times more likely that the crimes were committed by the same offender.

crime theory in programmatic

  • Street Networks

Crime Pattern Theory is a combination of two elements: awareness space and opportunity. The first element, awareness space, consists of locations that a high number of people use. The second, opportunity, is the chance that a crime can take place.

The street network defines many of the fundamental structural features of an urban area with building arrangements and paths between places. A heatmap of high footfall areas across cities was produced to create a street network discovering that burglary rate was 40% higher for busier streets due to increased ‘awareness space’ and, therefore, opportunity.

Street networks can be used as part of town planning by analysing whether footfall will be concentrated across a few streets which may increase the crime rate to decrease concentrated footfall town planning can incorporate multiple options for journeys, spreading the footfall.

  • Spatial and Temporal Accuracy

The crime theory used two models for accuracy; Spatial and Temporal models. The data’s influence decays across time. Data which is closer together and most recent is most accurate. The research uses Birmingham as an example: the residential data shows that at higher levels of spatial aggregation, there was a positive effect of crime connectivity and higher counts of victimisation happened on the streets with the highest footfall levels.

Online Advertising Application

This next generation data capture process could help in the digital marketing world, giving advertisers the chance to identify where potential customers will be, based upon their recent web activity.

marketing programmatic

Programmatic advertising plays a key part in the digital revolution and represents a new element in the traditional media model for media buyers to improve optimisation using automation and targeting users based on behaviour. Using the key points of the crime pattern theory we can see how the theory can be applied to online advertising success.

  • Awareness Spaces and Online Real Estate

The challenge of programmatic has been to reach the right consumers at the right moment in the right place. User data is collected across different devices and touch points, and plays a critical role in ensuring the right audience is served high-quality advertisements.

There are important similarities between the theory and the large awareness spaces online. These can be used to increase the chance of a conversion with advertisements appearing in high-traffic placements.

  • Clusters and Similar Customers

As with clusters of crimes occurring by the same person, the same can be applied to online activity and advertising, with people sharing similar traits and habits more likely to make similar purchases. User segmentation is key, as the system will automate bidding and targeting based on those who are most likely to make a purchase.

  • Spatial/Temporal and Real-Time Bidding

Real-time bidding is like Spatial and Temporal accuracy, with the most recent data having the most significant impact on optimisation.  More recent purchase data will be given greater weighting when creating a bidding strategy. To produce a successful campaign, the system will analyse what happened previously in the same space creating more purchases and increasing return on investment.