In order to tackle public sector wide problems like the increasing cost of social care and health and the lack of affordable housing or jobs councils need to switch from reactive / broad brush approaches to targeted actions often in the preventative space. In order to do this we need to understand who our citizens are by joining up data from across different service and organizational boundaries, combining county data, district data, NHS data and other data sets together. This is a complicated process both technically and from an information governance point of view, which means that creating a personalized detailed linked data record that can be legally used is extremely costly. We want to explore they value in sharing different data sets at different aggregation levels e.g. (ward, postcode, street, household, person) with different levels of details e.g. (“known to”, RAG, coded, full details) between different user groups. This will hopefully let us understand the most valuable data sets to share between different users and the most useful aggregations that can be shared which should provide a blueprint towards the creation of a minimal viable linked data set that provides actionable insight to enable change.
The users of data set would then be service managers within all branches of the public sector allowing them to undertake new ways of working and be able to track their effectiveness in real time.
This problem is caused by a number of different factors, related to: –
- Technology, the lack of good APIs means we are often forced to use flat extracts of data;
- Data, poor data quality in many systems and the lack of the globally unique identifier makes combing data from different systems hard at a personal level;
- Process, to share data between different organizations needs agreements in place and equivalence between data classification and controls;
- People, often people focus on what they know and is familiar e.g. the systems they use every day rather than thinking outside of the silos of services and organizations;
- Magnitude, large data projects are complex and costly hence often don’t get funded or are funded without a clear understanding of how the outputs will generate value;
We want to perform discovery with service managers as our core users, to understand what actions they can take and would like to take if they knew how to target them. This should generate a map of the value different levels of data sharing will bring and hence set a blueprint for a long-term data sharing solution that generates value every step of the way.
The project recently funded in round two of the local digital fund led by Stockport Council around providing better information to social works around children’s referrals partially demonstrate the validity of this idea, in that the collation of a very little data about a person can lead to great benefits.
Hypothesis: Not all data is equally valuable, and the value of data is not linear in relationship to its level of detail.
Assumption 1: It is easier and cheaper to share aggregated data than it is to share personalized data;
Assumption 2: If a linked data set existed services would use it to take actions to improve outcomes for citizens whilst generating savings;
Assumption 3: There are data aggregations that are useful to multiple stakeholders at once, rather than everyone needing their own unique way of slicing the data;
We plan a roughly four pass model of user research where we will engage with different user communities on four linked topics. During these engagements we aim to try and prove the hypothesis in general and specifically determine where a value curve may lie around the creation of a linked data set based on aggregated data with limited levels of detail.
Initially we will engage with the user base of service managers and data scientists within our partner organizations to find out what data sets they own and already have access to. For these data sets we will establish their level of detail / aggregation, any restrictions on their usage or further sharing and how they are used already to target actions and measure effectiveness of change activities. This will focus on interviews held within partner councils, the NHS (CCG and STP) and Cambridge Research institutions that have expressed interest in this work already. This will provide us with a map of the possible that we can use to inform our second round of information gathering on how data could be used.
Our second round of engagement will be run as a series of workshop involving multiple stakeholders from different organizations together. The objective is to understand what actions could be taken if there was the “perfect” data record with full details from every system, personally identifiable for all citizens across the region. The aim here will be to ideate as a multi-disciplinary team to create new types of actions that could be taken. These workshops will be held across the partners with an open invitation to non-partners councils and organizations to attend the workshops to get as wider set of views as possible.
Our third round of engagement will be with citizen themselves to test if the ideas we have generated as preventative approaches to see if they would work. This would be to seek feedback from both a control set of subjects and a set of citizens the preventative measures would aim to target.
The final round of engagement will be to examine the generated ideas to determine if they are still possible to implement with a degraded data set and the value they should generate both to the organization undertaking it and to the public sector as a whole. This should allow us to understand the value curve for each idea vs the quality of data used to target it. By then aggregating across the ideas we have generated and explored we hope to build a blueprint for data sharing where the value generated is always more than the cost incurred that allows us to build up a business case and roadmap that can be used across all of local government that improves citizens outcomes without require a large scale capital investment or a complex technology programme.
The problem here can be decomposed into two costs, firstly the capital and revenue costs of each organization creating their own “perfect” master customer record data store and analytics tools. These can be estimated both by creating project outlines and by sourcing input from the wider public sector from councils that have attempted this approach themselves and asking them to estimate the total cost of ownership of such a solution and describe the nature of the resulting solution, in terms of data quality, data volumes and linked data sources. Ideally, we would also seek to gain an estimate on the value such solutions have provided to their hosts, though this might be hard to obtain as councils might be less willing to share this information. This would allow us to show the costs developing these “perfect” data systems on a national scale hence the scale of capital investment needed and the depth of expertise that would be needed within the wider eco-system of suppliers and consultancies alongside staff.
This would allow us to demonstrate how a more value, actionable insight led approach to solving this problem would reduce the needed capital to deliver this style of solution and make it more accessible to all councils.
The second cost is the escalating cost of social care, health care and housing related matters. Whilst we would not be claiming to solve the entire problem, we can provide an estimate of the cost of delivering preventative actions at different levels of targeting and hence demonstrate the value that can be gained and hence the costs mitigated by the solving of this problem.
In order to create a functioning data warehouse within a single organisation, we conservatively estimate that it would take 100 person days to create and at least 1 FTE to manage, on top of this you would need time from every major system owner to integrate their line of business system into the data warehouse. Presuming we run the system on a cloud IaaS solution this means we estimate £150k in year one with a run time cost of £75k per year. If 150 authorities create these data warehouse this would be a total 3 year cost of £45m to the public sector without any guarantees or roadmap of how to achieve value from this.
Considering just the cost of social care, based on a recent report on the costs of social care “Without additional government funding, there will be a gap between demand pressures and available resources of at least £2.7bn in 2023/24. This is in addition to the £6bn gap which opened up between 2010/11 and 2017/18.” Funding gaps of this size are not fixable via efficiency savings which means preventative actions will need to form a role in future delivery of social care.
The project will be coordinated and run out of Office 365 where possible, using a dedicated MS Teams site to support access by all partners into a shared collaborative space. Within the Teams space tasks will be collated and tracked using Planner with MI information generated automatically and accessible via a PowerBI dashboard to provide real time progress tracking and business intelligence. MS Teams allows for the sharing of documents and their collaborative authoring between multiple stakeholders, it also supports video conference calls and chat functionality.
A weekly huddle / checkpoint meeting will be scheduled for all currently active project participants where we can confirm that all tasks are on track and hear about any issues that have occurred in the previous week. The aim of this meeting is to allow for the rapid visibility to risks and issues to allow us to dynamically re-plan as needed within a phase. This is likely to be a virtual meeting. Not all participants may be actively involved in the delivery of all phases of the project as we need to focus our efforts on ensuring we get the depth of knowledge and well as the breadth of context.
Between each phase a joint retrospective and re-planning meetings will be held, ideally with key project partners coming together in person to discuss the last phase and the next phase of the project, other participants will be invited to join via Teams video conferencing to enable full participation. This meeting is likely to occur on a 4 – 6 week period depending on the speed of progress of the project. Here we will share all of the learnings from the previous phase and gain perspective from every partner on if what we have learnt is widely applicable or is to focused, this will allow us to understand the balance between what is widely applicable and what might be particular to an area or perspective.
The project will be led by a single project manager who will coordinate all day to day activities and the finances of the project. A board made up of a single senior key stakeholder from each partner will then receive regular (weekly) updates from the project manager on the progress of the project and any issues with resourcing and progress from each of the partners on the project. This key stakeholder will be responsible for ensuring effective delivery from that partners resources and informing the project of future resource shortfalls to allow for re-planning. Each phase will additionally have a nominated technical lead from the partner that is best placed to lead that element of the work, they will coordinate with the project manager on developing the detailed activities that will be undertaken. This joint leadership should keep all partners fully engaged with the project for its duration.
We would appreciate the support of the Local Digital Collaboration or other functions within MHCLG in providing us details of any conference, events or central government discussion forums that we might benefit from attending / making contact with. We know that the landscape of events and forums within the public sector is disjointed and large and whilst we know of some of these events and forums, we acknowledge that our knowledge is incomplete. Hence the request for timely additional information about useful opportunities to raise the awareness of what we are working on and then support in the publication of the learnings we arrive at, at the end of the process.