An honest list, in order, of what would add the most value — with every figure counted from the software rather than estimated, and the two places this list turned out to be wrong left on the page.
This is an honest list of what would most improve commvita, in order. It is written for someone deciding whether to buy the platform or where to invest in it — not for the people building it.
Every number below was counted by a program that reads the software itself, not estimated by anyone. Where something could not be checked, it says so rather than guessing.
This is the question buyers most want answered and demonstrations are worst at answering. commvita has 414 screens. Some show live information from the system. Some show worked examples so you can see what a screen would look like with data in it. The platform now measures which is which, and says so on the screen itself.
138 of the 414 screens show example figures only. This is the biggest gap between what commvita demonstrates and what it does, and it is now measured rather than guessed at.
The work is one screen at a time: give each one a real question to ask of information the system already holds. It is not a quick connection job — see the correction below, where we thought it was.
commvita does not predict anything today. There is no machine learning running anywhere in it. That is a plain statement of fact, and it is worth stating plainly because the product once said otherwise: a screen described a prediction model, the size of the data it had supposedly learned from, and how accurate it was. None of that had happened. It was removed and replaced with a description of what such a model would have to achieve before anyone turned it on.
One carefully chosen prediction — built on real data, tested properly, and reported with its limits — would be worth more than several claimed ones. It is also the only item on this list that would change how commvita compares with other suppliers, several of whom do this today.
Health organisations merge, split and change their boundaries constantly. commvita now holds that as a proper record: who sat inside whom, on any given date, and which change made it so.
Two things already use it — deciding what a member of staff can see, and stating which organisational shape a reported figure was based on. Three more still work it out their own way. Each is now a small change rather than an open question.
Of 429 areas compared against other suppliers, commvita is ahead in 261, level in 148, and behind in 20.
One of those twenty is newly conceded rather than newly discovered: a pharmacogenomics rating was downgraded on evidence when the guidance it claimed to surface turned out never to have been read from source. A score that only ever improves is not a measurement.
Six of the twenty should stay as they are. Some are things another organisation does because that is its job — a national reporting service commvita feeds rather than competes with, an identity system it deliberately relies on, a partner whose product it carries. Some sit outside what a software supplier should be doing at all, such as decisions that belong to a licensing authority. Closing those would mean building the wrong thing.
The genuinely closable remainder is small, and needs product work rather than a change of wording.
Half of this exists already. Every summary figure the platform produces names the field it was calculated from, states which organisations it covers, and refuses to display a figure it cannot account for.
Finishing it means any number on a board report can be followed back to the individual records behind it. That is as much a documentation job as a software one.
Something else the measuring turned up. One compliance figure was reporting 100% against an empty record. Nothing had been recorded, so nothing had failed, so the screen showed perfect compliance. A board would have read that as good news. It now says plainly that the figure cannot yet be worked out — while a genuine nought is still shown as nought, because those are different things.
The comparison figures above are the same ones the platform reports to anybody who looks. Improving a score in order to show progress on a roadmap is exactly the behaviour the platform's own checks exist to catch.
Nothing in the system measures effort, and a confident-looking date typed here would be impossible to tell apart from a calculated one. Sequencing and cost belong in a conversation, with the people who would do the work.
Several national and standards websites cannot be reached from the environment these documents are produced in. Every figure above was counted from commvita's own code, so none of it depends on a source that was not read.
A screen showing example figures may still sit on top of working machinery underneath — and sometimes does. The count above measures what a user sees, which is the honest thing to publish, but it is a measure of the surface.
| Figure | Value | Produced by |
|---|---|---|
| Screens in the platform | 414 | read from the application's own list of screens; checked on every build |
| Showing only example figures | 138 | read from each screen's source |
| Reading everything from the live system | 80 | read from each screen's source |
| A mixture of both | 196 | read from each screen's source; deliberately not split further |
| Areas compared against other suppliers | 429 | the platform's own comparison record |
| — ahead of every supplier compared | 261 | one shared rule, used by the figure and by the list behind it |
| — level somewhere, behind nowhere | 148 | the same rule |
| — behind at least one supplier | 20 | the same rule |
The Community Edition is £1 per instance with optional paid support. It is open, self-hosted, and built on published standards.