Maidstone Borough Council, a local authority in Kent, England, deployed a predictive analytics system called One View to identify households at elevated risk of homelessness six to nine months before crisis, enabling proactive outreach and early intervention by council housing officers. The system was developed through a partnership between the Council, Ernst & Young LLP (EY), and Xantura, a UK-based public sector data analytics firm. The initiative was launched in 2019 as a strategic response to a 58 per cent rise in homelessness applications over the five years preceding England's Homelessness Reduction Act 2018, which placed new legal duties on local authorities to prevent homelessness at an earlier stage.
The One View platform operates by integrating and linking historically disconnected datasets held across council services and partner organisations. The system consolidated over 15 different data files from internal council departments and external agencies. Data sources include the housing register, council tax records, housing benefit records, tenancy debt data from Golding Homes (a local housing association), domestic abuse sanctuary scheme records, and 'troubled families' data shared by Kent County Council under the Kent and Medway Information Sharing Agreement (KMISA). The platform uses supervised machine learning risk-scoring models trained on these linked datasets to generate predictive alerts when a household crosses agreed risk thresholds. Warning indicators processed by the model include missed utility payments, housing assistance history, tenancy debt accumulation, and other vulnerability signals drawn from the integrated data.
A distinctive technical feature of the system is its use of Natural Language Generation (NLG) capabilities. When a household is flagged by the predictive model, the system automatically generates textual case summaries and safeguarding alerts that are presented to housing officers through dashboards integrated into the council's existing housing case management software. This NLG component supports, rather than replaces, professional judgement by providing officers with a consolidated narrative view of the household's situation across multiple service areas that would otherwise require manual cross-referencing of separate databases.
Data protection and information governance are central to the system's design. Maidstone Borough Council conducted a formal Data Protection Impact Assessment (DPIA) prior to deployment, as documented in a publicly released FOI response. The system employs IG-Bridge technology, developed by Xantura, which performs on-site pseudonymisation and encryption of personal data. All data is pseudonymised before being processed by the analytics platform, and if a household is flagged by the system, only the caseworker specifically assigned to that case has access to the de-pseudonymised personal information. Data sharing between the council and partner agencies is governed by formal Information Sharing Agreements, including the KMISA, which establish the legal gateways for data exchange. The system processes personal data and selected special-category data, including information relating to health and social care needs, age, disability, and marital status, where this processing is justified under the UK General Data Protection Regulation (UK GDPR) and the Data Protection Act 2018.
The system's infrastructure is hosted domestically within the United Kingdom. Pseudonymised data is encrypted both in transit and at rest, held at Xantura's data centre in Newbury, which holds ISO 27001, Public Services Network (PSN), and ISO 9001 certifications, as well as Cyber Essentials accreditation. Role-based access controls restrict data visibility to authorised personnel. The system operates with a human-in-the-loop oversight model: predictive alerts surface to housing officers who review each case individually and decide what intervention, if any, is appropriate. The model functions as an advisory triage and priority-setting tool and does not itself determine eligibility for services, impose sanctions, or make binding decisions about individuals.
During the initial pilot year, which coincided with the height of the COVID-19 pandemic, the system generated over 650 alerts for at-risk households. The council's housing team, consisting initially of one dedicated officer, contacted approximately 260 of these households due to capacity constraints. Of the households that received proactive contact, only 0.4 per cent subsequently presented as homeless. By contrast, among the approximately 390 alerted households that were not contacted due to resource limitations, 40 per cent later presented as homeless, with an additional 30 per cent presenting as threatened with homelessness. Overall, approximately 100 households were prevented from becoming homeless during the pilot year, and the rate of homelessness in Maidstone fell by 40 per cent. The council reported actual cost savings of approximately 225,000 pounds, with estimated potential savings of 578,000 pounds had staffing capacity allowed engagement with all flagged households, equivalent to approximately 15 per cent of the Housing budget. Broader societal savings were estimated at 2.5 million pounds, with a return on investment exceeding 190 per cent and projected potential of 660 per cent with broader rollout. Administrative task time was reduced by 61 days per worker. Following the pilot's success, the council recruited a second dedicated officer to expand response capacity.
The system was developed through a third-party partnership model, with EY providing advisory and implementation support and Xantura providing the One View data science platform and predictive modelling capabilities. The council has explored plans to integrate its system with wider Kent County Council operations to create a more holistic understanding of homelessness risk across the broader area. Challenges identified during implementation include limited staffing capacity to respond to all alerts, the need to retrain frontline caseworkers in proactive engagement approaches, public sensitivities around cross-service data sharing, and legacy IT systems within the council that still relied on paper-based and fax-based processes.