'Change NHS' ideas: getting clarity by soliciting views
(Maps viewable on desktop only)
In 2024, the NHS launched an initiative soliciting change ideas from the public. The initiative was based on open-ended questions, soliciting views on the “best things” as well as the challenges.
The problem comes when organizations try to make sense of the resulting haystack of responses. Reading a long list of verbatims isn’t empirical. Using NLP or topic models to organize the content provides only a narrow view of the responses – without a sufficient line of sight to reasons why.
Instead, we took 2,770 of the responses and ran them through narrative analytics - quickly revealing clear, actionable themes in context. The result is a "map" of the ideas, based on axes and clusters that are bespoke to the dataset.
To view the map, just hover over the dots with your cursor. First, look at the 5-cluster view below. This shows the axes: The talent/people vs The process on the horizontal axis, and How the system works vs What patients need on the vertical axis. All of the dots are placed based on these axes. Make sure you explore the different levels of narrative detail, by using the slider on the lower right and viewing the 34-cluster level.
What the press is getting wrong: This is not a ‘Boaty McBoatface’ moment
It is brave to do a “Let’s hear from the public” exercise in the UK. The media noted this – with The Guardian and other publications suggesting the effort became an exercise in ironic humour. (see link) The data shows the reality was the opposite. In hovering the dots and viewing the narratives, it is easy to see just how serious and well-reasoned the ideas were.
If a large number of verbatims were characterized by irony or sarcasm, narrative analytics would include a cluster specifically identifying them within the map. There is no such cluster, because there are very few such responses. People are serious about their NHS.
That said - If you would like a laugh or two, a higher concentration of light-hearted comments can be seen in the Culture/media cluster. You can find it by selecting the 34-cluster view in the map above. It is a small cluster near the middle, with dark green dots.
What types of ideas are most frequently posted?
From the 34-cluster view in the above map, the 10 largest are shown in the bargraph below. Resourcing choices, Simpler/better info handling, and NHS funding were the top 3. Have a look at the others. How do the narratives in the graph below compare against the top priorities as you see them? Further cuts of data such as this can reveal how the clusters rank in different regions, or how they vary by age, income, gender, or political affiliation. This can be highly actionable in the formulation of new policies and reforms - particularly when changes and trends are tracked over time.
Narrative traction - what ideas are getting more vs fewer likes?
Another way to view the data is by narrative traction. This view shows the distribution of ideas which get more vs fewer likes - viewed by narrative. In the below graph taken from selected clusters in the map, we can see that Redefining hospital care and Fix inefficiency/waste have a higher proportion of ideas earning six or more net likes. Though NHS funding has the 3rd highest count of idea submissions - the chart below shows it drops to #7 as measured by likes. This suggests that - though there are many ideas submitted on funding - getting the taxpaying public to advocate funding changes is another matter. (Net likes = the total number of likes minus the total number of dislikes).
For healthcare organisations and policy makers alike, narrative analytics is the fastest, easiest way to measure the 'outside-in' stakeholder realities related to a particular issue or challenge. This case illustrates how survey blindspots are eliminated through open-text questions, rather than closed ended surveys - an emerging practice is known as open-ended quant. In addition to providing a scaled, empirical understanding of stakeholder views, the results themselves can be leveraged for content and engagement.