Dashboards aren’t dead because charts are useless. They’re dying because asking people to stare at charts and work out what matters is increasingly an absurd way to use data.
We will continue to have charts, graphs, indicators and visual summaries, but what is dying is the idea that the best way to help someone understand an organisation is to give them a screen full of numbers – and expect them to find the story.
For years your organisation and vendors have proudly added more dashboards: executive dashboards, quality dashboards, clinical dashboards, risk dashboards, and compliance dashboards (just to name a few). And, because apparently nobody had suffered enough, dashboards containing links to other dashboards.
We made the human the analytics engine
Traditional dashboard process was: DATA → CHARTS → HUMAN → INTERPRETATION → QUESTION → MORE DATA
The manager has to notice that something looks unusual. Then decide whether it matters. Then drill down. Then correlate it with something else. Then perhaps ask Quality what happened. Then look at incidents. Then complaints. Then staffing. Then maybe discover the thing that actually matters.
In other words, we gave computers the easy job: displaying numbers. And gave humans the hard job: finding meaning.
A manager doesn’t really want a dashboard showing:
27 incidents
8 medication events
complaints ↑ 14%
falls ↓ 6%
19 overdue actions
staff turnover ↑
What they actually want to know is:
“Is there anything happening here that I should be worried about?” And then: “Why?” And then: “What should I look at?” And perhaps: “What should we do about it?”
Now computers can understand the stories
Care makes the dashboard problem worse, because the interesting things happening within a care organisation rarely live neatly inside one metric.
Imagine:
- Falls haven’t increased significantly.
- Complaints haven’t increased significantly.
- Medication incidents haven’t increased significantly.
- Agency usage has risen somewhat.
- Progress notes contain more references to agitation.
- There have been several seemingly unrelated family concerns.
- Nothing necessarily turns red on a dashboard.
But together they may be telling you something.
And conventional dashboards are poor at that because they tend to organise information according to the structure of the system: e.g. Incidents | Complaints | Risks | Workforce | Clinical | Compliance
Real life doesn’t respect those software module boundaries. People experience care; software experiences modules. Falls don’t live separately from staffing and complaints don’t necessarily live separately from incidents. Progress notes don’t live separately from changes in behaviour, either.
AI potentially lets us bridge that distinction.
Care doesn’t fit neatly into boxes
For 30 years we’ve been turning stories into numbers so computers could understand them. Now computers can begin to understand the stories, and with AI are now able to combine quantitative information with qualitative sources, which is one of the limitations of conventional dashboards.
That’s enormously significant in care because some of the richest care information isn’t numeric, it’s narrative. Things like incident descriptions, complaints and investigation findings. Progress notes, family feedback and staff observations. Audit comments, meeting minutes, and corrective actions, too.
Historically we’ve tried to convert these stories into classifications, dropdowns and numbers because that was what computers could analyse. But AI changes that equation – we can increasingly analyse the narrative itself.
‘Show me the numbers’ to ‘tell me what’s happening’
This is where the AI and Large Language Models transformation enters, from “show me the numbers” to “tell me what is happening”. The future isn’t merely, “how many falls did we have last month?” That’s useful, but hardly revolutionary.
It is more like: “what has changed in the last three months that I should know about?”
Or “are there any emerging quality of care issues that aren’t obvious from our headline measures?”
Or” why have complaints about meals increased?”
Or “is there anything across incidents, complaints, progress notes, staffing and actions that suggests these issues may be related?”
Now we have moved from retrieval to reasoning. We have shifted to DATA > TECHNOLOGY IDENTIFIES WHAT MAY MATTER > HUMAN JUDGEMENT
The important nuance is: AI does not have to provide a definitive answer. Sometimes the most valuable output is simply, “there is something here worth looking at.” However, identifying patterns is one thing. Deciding whether those patterns are important is another.
AI doesn’t get to decide what good means
But – and this matters – AI can’t simply make up what ‘good’ means.
Different executives can legitimately ask for the same metric in different ways and get different, individually correct answers. AI can calculate something correctly and still interpret its importance differently from your organisation. There’s a strong distinction between correct calculation and consistent organisational interpretation. Which, for care organisations, means governance.
AI still needs to understand:
- what measures the organisation has agreed matter
- how those measures are defined
- thresholds and tolerances
- regulatory obligations
- organisational policies
- historical context
- who is responsible for what, and
- what constitutes an exception requiring attention.
Otherwise, we haven’t eliminated dashboard chaos. Instead, we’ve created AI chaos faster.
So….are dashboards really dead?
Dashboards still have a role, just not necessarily the one we’ve historically given them.
The dashboard of the future might not be something you visit, but rather something the AI knows. It represents:
- These are our important measures.
- This is how we calculate them.
- This is how we normally view them.
- These are the thresholds we care about.
- This is what good looks like.
Then, AI can go beyond it. So the chart hasn’t disappeared, but might stop being the sole intelligence.
Human attention is the scarce resource
The deeper subject isn’t really dashboards: it’s human attention. Care organisations have finite human attention and increasingly, the challenge isn’t collecting more data or producing more charts, it’s knowing what deserves attention and applying human judgement to it.
For decades we have essentially said: here is the data → here are some charts → YOU work out what matters.
Now we can start saying: here is the data → the technology identifies what may matter → YOU apply judgement.
For years we’ve used technology to collect more data and build more dashboards so humans can understand what is happening. Now, AI’s job is ideally to reduce the amount of human attention wasted on finding what deserves human attention.
We don’t want to remove human judgement, but stop wasting it and instead move human judgement to where it’s actually valuable. How? By letting the technology search, correlate, compare, question and surface what may matter. Then, by letting people do what people should be doing: judging what it means and deciding what to do about it.
Dashboards may not be dead – but the dashboard as the destination probably is.
