How learning from end users delivers data-driven improvements with real impact

Louis Pattison

A little data can be a powerful thing. These days, though, organisations often find themselves dealing with more than just a little data.

Day-to-day operations create mountains of information – some of it useful, much of it just noise. Though, the art of learning from data requires more than separating the wheat from the chaff. It’s about working out how to turn data into something that’s of real value to end users. 

We tackled this issue head-on in a recent project. Pegasus is a UK-based care home network providing care to adults with disabilities and other complex needs. Their app, Bloom, helps document and plan the care they provide to patients across their care homes.

Every day, Bloom collects huge amounts of data around meals, medication, sleep, hygiene and hydration, as well as logging information about unexpected incidents, like falls or challenging behaviour.

Pegasus came to us because they wanted to learn how they might make data-driven improvements to Bloom – including ways they might use AI to reduce the administrative burden on staff by automating time consuming activities.

Laying the foundations for AI integration

To understand where we might find high-value improvements, we took a two-pronged approach, focused on data and user experience. 

Pegasus granted our data scientist access to Bloom’s anonymised datasets, giving us the raw material to develop proofs of concept. We explored this data, looking for:  

We also wanted to learn about what Bloom’s users needed when it came to data. To do this, our user researcher and I conducted one-on-one interviews with care home managers in the Pegasus network. As well as managing the workers who input data into Bloom all day, managers also use Bloom to create care plans, reports, audits and respond to critical incidents.

Through our interviews with care home managers, we gained a deep insight into their everyday processes, pain points, and unmet needs that might define what solutions – AI or otherwise – might look like.

Turning data and user research into insights

Both our data exploration and research findings revealed issues with the data quality. Too much data in Bloom was collected via free text fields, which led to inconsistent or incomplete data, limiting its suitability for powering analysis or AI solutions.

Bloom is far from an outlier in this regard. Many organisations experience data quality issues as they move beyond basic use cases. A key recommendation was the importance of improved data entry, with defined fields as the default – a change that would transform the quality of their data and quickly unlock value.

Following the delivery of our final report, Pegasus took immediate steps to improve the way they captured data – an important step in strengthening their data foundations and laying the ground for future AI integration.

Care home managers talked about the positive difference that Bloom made to their previous paper-based processes. But they also experienced difficulties. One user described the “detective work” they had to do, jumping from page to page to find and cross-reference information.

All managers we spoke to expressed a desire to bring information together in one place in a visual and easy-to-digest format.

Our user research gave us valuable information on how to present information that would be useful to care home managers.

Using automation and AI to cut admin

We identified other ways that data could assist care manager workflows. The user research revealed significant pain points around certain admin-heavy tasks.

Care home managers said that reporting was still a heavily manual process, involving lots of double or even triple-keying. One manager suggested she spent up to 50% of her working day on creating reports. We created designs to show how aspects of report-building could be automated. Report-building functionality that automatically ingests data would both save managers significant time and reduce the possibility of human error.

We also heard how long it takes care home managers to create and update care plans – a personalised, written document that details a resident’s specific health, social, and personal needs.

We showed how LLMs could turn rough notes and stored data into material compliant with Care Quality Commission (CQC) standards. We also flagged the importance of keeping human operators in the loop to ensure outputs are accurate and free from  hallucinations.

Ready to find out how AI-ready your organisation really is?
Take our AI Readiness Assessment and get personalised insights to accelerate your AI journey from our AI experts.

Getting the best out of data

Good data can have a transformative impact on the way you and your organisation works. It’s not just about making the data available. It’s about learning what your users need, and presenting the data in a way that helps them do their jobs better.

That way, you don’t just get a shiny dashboard that won’t get used – you get a tool that saves time, improves quality and unlocks new workflow opportunities.

To learn more about this work, read the case study on the Zaizi website.

Worried you’re not getting the best out of your data and want to know what you can do with it? Unsure about what your users really need? Speak to us.

If you’d like to explore what AI readiness looks like in practice, you can read our AI Readiness Whitepaper, which sets out the 7 steps organisations need to address to adopt AI responsibly and at scale.

Related content

Thanks for joining us! We’ll keep you informed with regular updates.

Subscribe to our monthly newsletter, including updates about our content and events

Consent(Required)
This field is for validation purposes and should be left unchanged.