Where should we launch Ask the NHS first?
The team is working hard to get Ask the NHS launched before the end of this financial year. Real-world information about how the public might use this service is essential to establish its potential costs and benefits as well as identify previously unanticipated user behaviours.
These will help us firm up a longer-term roadmap that can prioritise building out the most valuable parts of this capability and mitigating user challenges.
Ask the NHS is not meant to serve a new user base. It is a multiplier for existing digital and in-person services in the NHS; we aim to match people quickly to the help they need. So for our first release it is important to consider where to position it within existing channels to prove it can effectively support a wide variety of user needs.
There are three likely settings where Ask the NHS could make its debut: the NHS App, NHS website, and NHS 111 online. Each of these represent slightly different propositions for our service, and all will be important for scaling up this capability. But for now the operative question is which setting should we support first to give us the right test-and-learn data?
The team has already held conversations with each of these stakeholders to gauge the art of the possible. And we intend to keep those dialogues going to agree the collaboration necessary to go live with our first test-and-learn pilot.
NHS App
The App mostly assists users with administrative matters: booking appointments, managing medications, reading records. There are ambitions to expand this scope, outlined in the 10-year plan, but it is uncertain when a substantial portion of users will start to view the app as a tool that supports other needs.
Hypothesis: For an initial pilot, Ask the NHS would mostly receive user queries relating to existing capabilities in the NHS App because users are unlikely to access the App for help with other matters.
This means launching first as a capability within the NHS App would be unlikely to prove the ability of Ask the NHS to support a broad range of users’ needs. This is not meant to minimise these user needs; the App has succeeded nationwide because its development has rightly focussed on supporting tasks that are common for a very large number of patients. Ask the NHS will need to ensure that it, too, offers good recommendations on these topics.
We fully intend for Ask the NHS to be part of the NHS App in the near future. And adding this capability, once proven, could be a valuable addition that will help reshape the way users think about what the NHS App is for. In that light, perhaps we will need to consider bespoke success criteria when it comes time to roll out in the App.
NHS website (NHS.UK)
The NHS website serves many more user needs. It is a vast estate with content relating to conditions, services, medicines, procedures and treatments. I’m sat next to NHS.UK colleagues as I write; a team board beside advertises a target of 1,219 topics to transform and modularise.
The website offers the breadth Ask the NHS should target, but our offering would need to be included thoughtfully to ensure it supports users when they are struggling to find help. The size and scope of NHS.UK lends itself to fly-in-fly-out interactions; many users go directly to a relevant topic after a web search (or LLM query). But it probably wouldn’t help to add Ask the NHS to these individual pages.
If it appeared, for example, on a condition page for asthma, our service might only receive queries limited to asthma. Or worse, it might get very few enquiries because users had already navigated to the page they needed.
The value of Ask the NHS is in directing users when they are uncertain about the service or help they need. So it needs placing where website users go when they can’t immediately find the guidance they seek. This makes it a better fit for, say, the NHS.UK homepage.
Hypothesis: Placing Ask the NHS in general-interest parts of the NHS website, would invite a diverse range of user queries and target users who need our support the most.
We are in the early phases of conversation with NHS.UK teams about how we can do this. There is a need for cooperation on both design and technical development. Ask the NHS needs to be integrated into the current homepage in a way that adds value to the page without disrupting its core objectives. And it needs building in a way that makes it semi-independent of NHS.UK, especially in the test-and-learn phase. Ask the NHS should only appear to a subset of users and it should be implemented flexibly enough to allow easy removal or updates to help manage costs or clinical risks.
Collaboration with NHS.UK can be a two-way opportunity. Website teams are alive to the need to support a public that increasingly engages with LLMs when looking for health advice. Adding an AI-powered service like Ask the NHS to NHS.UK can provide all of us with insights into how people use such services. It can also be an in-house test-site for LLM-compatible content solutions like an MCP server.
NHS 111 online
Many colleagues working on Ask the NHS previously supported 111 online so are familiar with the public’s struggles to navigate the NHS. The service gets reams of frustrated feedback, and we plan to use an anonymised portion of these as a starting point for real-world evaluation data that will test the performance of any AI components within Ask the NHS.
There is an opportunity for Ask the NHS to transform the “guided entry” process, already active on the 111 online homepage. Indeed this work began as an attempt to offer a smarter version of that multiple-choice interaction.
By capturing people’s problems in their own words, the service should identify, for example, whether a user is struggling to finding a local dentist or debating if their toothache needs urgent attention. The first can be handled by a service finder; the second can be achieved with a 111 symptoms assessment.
This “guided entry” process has been in place since early 2021 and it consistently diverts a small but significant number of homepage visitors to more relevant services. It makes a marginal but helpful contribution to demand management. Yet for testing Ask the NHS, the guided entry stats show that, in the main, people visit 111 online when they need help with symptoms.
Hypothesis: Positioning Ask the NHS at the start of 111 online, would mean user queries overwhelmingly ask for help with symptoms.
Trialling Ask the NHS as part of 111 online, therefore, is unlikely to prove how well the new service handles a wide breadth of user needs. Almost every journey would end with a recommendation to answer questions about symptoms on 111 online, perhaps with another option option beside it to read health advice and guidance on NHS.UK.
Importantly, the team has already done user testing about offering AI-enabled assistance on 111 online. Adding this option immediately raised users’ expectations beyond the bounds of a minimum viable product. They assumed any symptoms assessment that came after a free-text submission would take account of all the information users supplied, like medical history. Introducing AI tools also suggested that 111 online had integrated access to information like prescribed medicines and GP registration.
In future, we want Ask the NHS to tailor user recommendations based on patient context and to pass along relevant user-provided information, like sending their symptom descriptions into a separate online assessment engine with adaptive triage processes.
These features will arrive later as other Access, Triage, and Navigation (ATN) capabilities come online. For a test-run in the present, the results of user testing suggest we might introduce a clinical risk of patients withholding important information or taking the wrong action because they have incorrectly assumed digital systems know more than they do.
A wider view
As highlighted above, Ask the NHS can play a part in making each of these products better: redefining what the NHS App can do, reshaping NHS.UK for AI access, directing people into symptoms assessments with improved accuracy. We want to be good neighbours to the products closest to us.
But beyond these thoughts about the very first, most-informative place to introduce Ask the NHS and some ideas on where to roll out next, our team should consider the ways we are supporting the entire health service.
Introducing an open portal that lets anyone in England ask the NHS for help offers an incredible view of how, where and why people struggle when they try and manage their health. That data can inform real improvements in all kinds of operations, whether they are digital, administrative, or clinical.