Legibility
You get standardised fields, comparable categories, a number a trustee can read on a page.
It costs frontline time, and the condensing that makes a number readable destroys the trail behind it.
Working example · Impact Desk
Record the evidence when it happens.
The records that are easy to keep are rarely the ones a review asks for.
Impact Desk is an operating foundation for small teams delivering real services on very few paid hours. It records the work against the questions a funder will actually ask, as the work happens — so nobody has to rebuild the evidence a year later.
Once I prepared a homeless service for a funding review. The question was whether its staffing was more than the need justified. The records could not answer it either way, and it took over a year to build the picture after the fact.
The problem, named
Every service asked to prove itself is solving three problems at once.
They pull against each other. Almost every system — and almost every funder requirement — quietly picks one and sacrifices the other two.
The evidence trilemma
Name it and it stops being a vague frustration. Most of what feels like bad software, or an unreasonable funder, is one of these three demands winning at the cost of the others.
No system resolves this, and this one does not claim to. Impact Desk is built with all three in view, and makes the trade-off visible instead of hiding it where only an audit would find it.
Usability
You get few fields, plain language, capture at the moment the work happens.
It costs comparability. Almost nothing aggregates, and leadership is handed anecdote.
Provenance
You get attributed, dated, sequenced records that hold up when somebody finally digs.
It costs capture effort, and the full record is unreadable to anyone above the work.
Where the usual routes land.
This is not an argument that other software is bad. It is an argument about which side of the triangle each route optimises, so the choice can be made deliberately rather than by default.
Funder-shaped systems
Optimise legibility. Comparable numbers arrive, and so does heavier capture. The reporting requirement becomes the workload.
Spreadsheets and light tools
Optimise usability. Fast to start and genuinely pleasant, until you need an aggregate — and the record leaves with whoever was keeping it.
Enterprise platforms
Optimise capability and provenance. They can be built into almost anything you can describe — with a platform budget and somebody to administer it permanently.
If a team has a platform budget and an administrator, a large platform is a reasonable route. Impact Desk is the alternative worth considering when they have neither, and the work still has to be evidenced.
In practice
What this looks like in a team of three.
Not a feature list. Four situations that decide whether a small service can show what it did.
01
Three hours on a Friday
Someone arrives after a fortnight away. They open one screen and see the queue: who needs a call, what is waiting for review, which follow-up is overdue. They do not spend the first hour working out what to do.
02
Every hat, not just case work
COSHH records, daily building checks, risk assessments. Each one arrives as another form somebody has to design, another procedure, another pile of evidence for when an inspector asks — and a handover problem on a revolving rota.
03
The step nobody records
A plan says someone needs help with life skills. Then nothing carries it. The need, the steps agreed with the person, who delivered what, and whether it worked all have to sit in one chain.
Someone who could not plan a grocery budget or cook for themselves, who now can. That is an outcome, and it is invisible in a system that only holds case notes.
04
Delivered is not effective
The register says the session happened. It does not say whether the person can now do the thing. Completing an activity should schedule a check owned by a role rather than a name, because whoever delivers rarely sees the effect.
Four rules the system holds to.
- 01Capture once, at the point of the work, by the person doing it.
- 02No collation layer, no second system, nobody re-keying one spreadsheet into another.
- 03Nothing collapses upward. A referral sent never becomes a placement achieved.
- 04Every number traces back to an attributed record in one step.
Experience behind it
I have run these services, not only built software for them.
Fifteen years in UK support services, starting as a project worker, then project manager, then area manager. I began in a small independent non-profit, went through Stonham, and ended inside Home Group — one of the country’s largest providers of housing, health and social care.
A good deal of that work was taking services that were failing, or had funding question marks over them, and turning them around so they passed Best Value Reviews — the statutory regime local authorities were charged with auditing against strict criteria.
What I kept finding was not that the work was bad. The value was real and nobody was recording it, because there was nowhere to put it and because putting it somewhere would have been one more burden on a team with none to spare.
So the honest expectation is not that your numbers get worse once they are recorded properly. In my experience they usually get better, because you are already doing work you have never been able to count.
15 years
Support services, multi-site
Budgets, safeguarding, inspection and services for adults with complex needs.Worker to area
Both ends of the record
I have written the case note and I have had to report on it upward.Best Value
Audited turnarounds
Failing or at-risk services brought through statutory review against strict criteria.Education
The assessment spiral
Where I learned how fast an evidence requirement becomes an unpaid workload.The demo
Look at it working before we talk.
There is a demonstration instance open now, on fictional data with no sign-in. It is a partial demo, not a platform, and it says on each page what is working and what is only a direction.
Open the demo ↗What you can see in it
- A staff dashboard and the queue of work waiting
- An intake reviewed and promoted into a record
- A referral through to a recorded outcome, each move attributed and dated
- Interaction history with follow-up tasks that stay visible
- A programme report, and what every figure in it traces back to
Carrying a service on fewer hours than it needs?
I would like to hear which record nobody has time to keep, and which question your funder actually asks. One conversation about one workflow — not a pitch, and no follow-up sequence.
If it turns out a smaller change would do it, I will say so. The point is a better operation, not an app for its own sake.