Xantura OneView (predictive homelessness flagging)

OneView is a data-integration and prediction platform that Xantura Ltd builds and operates for several English councils. It combines data from 15 or more council and partner sources into one view of each household. Its risk models flag households at risk of homelessness months before a crisis, as ranked alerts with a written case summary.

How Maidstone used it

Maidstone Borough Council, in Kent, partnered with Xantura. It combined data from 15 services and outside agencies. They included the housing register, council tax, housing benefit, and tenancy and rent-arrears debt from a housing association. They also included the county council's data on domestic abuse and on families in its "troubled families" work.

When a household's score passed thresholds the council agreed, OneView sent an alert to the housing team. In the pilot year it raised more than 650 alerts for households three to six months from a crisis.

An officer with a background in financial inclusion, meaning helping people with money, reviewed and acted on the alerts. The officer had capacity to attempt contact with about 260 of them. Support included income maximisation, budgeting and debt advice, Discretionary Housing Payments, and mediation. The council later recruited a second officer.

What the reported results show

The council and vendor reported that 0.4 percent of the highest-risk households contacted later presented as homeless, meaning they came to the council as homeless. About 40 percent of the alerted households the officer could not reach did so. The contrast is widely quoted.

It is not a controlled comparison. Contact was rationed by capacity, not assigned at random. So the gap measures who was selected for contact, not what the help did.

Council and vendor materials also report a 40 percent fall in homelessness and about 100 households prevented. They report an initial accuracy of 84 percent in identifying imminent homelessness. They report savings of about £225,000, and up to about £578,000 at full capacity. The vendor states a return on investment over 600 percent and about £2.5 million in savings to society.

All of these figures come from one pilot year during COVID. They compare before and after, and the council or vendor reported them. No method, starting rate, or margin of error was published.

How Barking and Dagenham governs the data

Barking and Dagenham, a London borough, began working with Xantura and the firm EY in 2018. It bought OneView in 2019 to develop analytics across children's social care, homelessness, and adult services. In 2020 the platform produced a COVID-19 risk model.

There, OneView's Single View of Vulnerability runs under a statutory data-sharing agreement made under the Digital Economy Act 2017. It is number 376 on the public register and runs from 1 June 2023 to 31 May 2026.

The Department for Work and Pensions, the Ministry of Justice, the council, and the Metropolitan Police are named as controllers, the bodies responsible for the data. Xantura, INBest, and CareTech are among the processors, the firms that handle it for them. The data covers identifiers, health, social care, education, offending, benefits, and debt.

What outside reviewers found

The Ada Lovelace Institute, an independent research institute, studied staff using OneView in Barking and Dagenham by observing their work. Its fieldwork ran in 2020, and it published in July 2024. The study covered children's social care and the COVID-19 response.

It found that social workers and managers lacked a clear explanation of which factors drove the case summaries and alerts. Some frontline staff were not convinced the analytics were as objective, neutral, or accurate as described.

In 2021 the campaign group Big Brother Watch published an investigation called The Poverty Panopticon. It criticised Xantura's COVID-era OneView, used by two London councils. It said the model built on thousands of data points, including sensitive items unrelated to housing. It said the model operated without residents' knowledge.

How the claims are being tested

To test what the pilot figures could not show, the Ministry of Housing, Communities and Local Government commissioned a randomised controlled trial. It is called Using Data to Prevent Homelessness. The Centre for Homelessness Impact manages it.

Xantura supplies a pseudonymised list of households at risk, meaning names are replaced by codes. Verian Group and Simetrica-Jacobs are the independent evaluators. The councils are Barking and Dagenham, Newham, Stockport, and Test Valley, and the ministry's privacy notice adds Kensington and Chelsea.

Households are randomly assigned either to proactive support calls or to a control group for comparison. Calls began around April 2025. Full results were due in 2026, and the trial was still in progress into that year.

Why reach matters here

Almost every other predictive tool in the case files is judged on its accuracy. OneView raises an earlier question. Even a correct flag does nothing until someone acts on it, and here far more households were flagged than one officer could reach.

So the tool's real coverage was set by staffing, not by the model. The harm did not fall on the households the tool got wrong. It fell on the ones it may have got right that no one contacted.

Two more controls matter, and neither is about accuracy. The first is transparency. Where staff cannot see which factors drove an alert, an opaque alert is either trusted too much or quietly set aside.

The second is whether combining the data is legitimate at all. A statutory agreement and a data-protection officer govern it. In 2021 Big Brother Watch framed the COVID-era OneView as automated suspicion.

The intended output is proactive help, not a penalty. The open question is whether prediction widens that help, or widens surveillance faster than the capacity to act on it.

What this network is drawn from

This network follows the pattern the case file describes. It is not a reconstruction of the actual platform or its models. It draws on Maidstone for the officer and the pilot year, and on Barking and Dagenham for the kinds of data combined.

What the available tools can and cannot address

A failure pathway is a link between two parts of the network where a mistake made by one part can be passed on to the other. Closing a pathway means mistakes stop passing along it. The work along it goes on.

This case has a budget of 13 units. Each tool costs the same at every target level, except Understand the system, which pays for ongoing study of the deployment. It costs 3 units under Explore (No Targets) and Service Targets Only, and 4 under the two higher levels. While it is on, Gate vendor updates, Require sign-off, Vet connections, and Assign a challenger each cost 1 unit less. With its stronger setting, Deep research, at 6 units, they cost 2 less, but never less than 1.

Before any tool is used, five failure pathways are open. They are Ranked alerts to the officer, Council and partner data into OneView, and Household history used in scoring. The others are Officer records contact outcomes and Case history read by the officer. The network is at a tipping point, where mistakes could start building on one another. It is not self-correcting, meaning it does not clear mistakes on its own. The service is keeping up with its workload.

Explore (No Targets) sets no targets. Under Service Targets Only, the targets are not met before any tool is used. That level asks for the network to be self-correcting and for the system to be helping the work. Assign a challenger meets them on its own for 2 units, and so does Escalate checks. Understand the system does for 3, and so does Gate vendor updates at its stronger setting. Review on schedule at its stronger setting does for 4. Upgrade model and Store less data, each at its stronger setting, do for 5.

Lingering effects is a Dynamics setting in which damage outlasts its cause, on by default. With it off, Vet connections and Store less data each meet the targets alone for 3. In all, more than 1,200 different sets of tools within the budget meet them.

Under Service and Safety Targets and All Governance Targets, the targets can be met. Both levels ask you to close every failure pathway and keep the service from being strained, among other targets. Every set of tools that meets them includes the same four tools: Escalate checks, Understand the system, Store less data, and Vet connections. Together they cost 11 units at these levels. Nine different sets of tools meet Service and Safety Targets, and four meet All Governance Targets.

Escalate checks has the officer check alerts more closely when monitoring flags trouble. Mistakes stop passing along Ranked alerts to the officer. Store less data means fewer records written and kept, and mistakes stop passing along Officer records contact outcomes. Understand the system stops them passing along Case history read by the officer. Vet connections lets only granted data links run. With Understand the system on, mistakes stop passing along Council and partner data into OneView and Household history used in scoring.

More is not better here. Using every tool on offer, each at its strongest setting, costs 35 units, nearly three times the budget. It meets the targets at none of the three levels that set them, because the system then no longer helps the work.

No tool on offer reaches the households the officer never contacts. The harm to them lies outside this network.