EXCL: DWP explores data science to inform Jobcentre ‘interventions’ for benefit claimants

Department retains specialist firm to help explore how data science models and insights could inform the content of mandatory weekly meetings between work coaches and new claimants of out-of-work benefits

The Department for Work and Pensions is undertaking a six-figure programme of work to experiment with using data and analytics to help direct the “interventions” made by Jobcentre staff working with benefit claimants.

The DWP last week entered into an initial 18-week contract with Cardiff-based firm Butterfly Data. The specialist tech provider has been brought in to deliver the initial discovery phase of a “project [that] seeks to explore the potential of greater usage of data analysis and data-science techniques to enhance the targeting of Jobcentre Plus (JCP) interventions more effectively”, according to recently released commercial documents.

This work will involve “gathering a range of data from across DWP systems, using this to identify common characteristics among claimants at 13 weeks of their claim, and uncover their main barriers to employment”.

PublicTechnology understands that the initiative will focus on those included among the DWP’s “intensive work search” regime. This cohort is comprised of about 1.4 million recipients of Universal Credit and includes claimants – who are either out of work or earning only very small amounts – who are “expected to take intensive action” to seek employment, according to government guidance.

As part of which, for the first 13 weeks of their claim, those on the intensive regime are required to attend weekly Jobcentre meetings – otherwise known to as ‘interventions’ – to review their search for employment with a DWP work coach. After this initial 13-week period, the interventions may switch to fortnightly intervals for some claimants.

The ultimate intent of the engagement with Butterfly Data is to support the possible future implementation of “data-driven improvements” to these regular meetings between work coaches and claimants.

The project will involve the supplier collating and providing to the DWP “structured data sets (e.g. SQL tables) at the claimant level encompassing a range of characteristics and other variables that have been explored during the project”, according to the text of the contract signed by the two parties.

“These data sets should be stored in a common location e.g. DWP DataWorks environment, accessible by permanent staff and able to be recreated using code,” the document adds.

Related content

The data science firm will also be tasked with delivering “one or more models created from the use of statistical or machine learning techniques [and] applied to the data sets” during the course of the discovery exercise.

Finally, the DWP expects to “receive a summary of the work, conclusions, key insights and actionable recommendations”.

The contract adds: “This should be suitable for a range of technical and non-technical stakeholders, including project sponsors, DWP decision-makers, and operational teams, and is designed to inform and guide the implementation of data-driven improvements to Jobcentre interventions. The reports should include a review checkpoint following the completion of the initial data discovery, with the intention of enabling the project stakeholders to decide if the project should progress to further stages.”

An initial £200,000, excluding VAT, will be spent via the discovery engagement – a figure which could rise as high as £500,000 if the department takes up the option of a further 18-week extension.

Once this initial research stage has been completed, the DWP will assess the options for further pursuing the project and whether and how to implement data science techniques on a permanent basis, PublicTechnology understands.

According to the company’s service listing on the government’s Digital Marketplace platform, Butterfly Data aims to “enable customers to exploit their data and gain intelligence from the patterns and trends within it”.

The listing adds that the firm’s consultants and technical experts can support initiatives including: “data exploration to extract information and actionable insight; predictive analytics for forecasting and propensity modelling; use of algorithms including random forest, gradient-boosting machines; unsupervised machine learning and neural network models; [and] unstructured data analysis and text mining”.

Sam Trendall

Learn More →

Leave a Reply

Your email address will not be published. Required fields are marked *

Thank you! Your subscription has been confirmed. You'll hear from us soon.
Subscribe to our newsletter