Delivering a Government Incubation Scheme
What changes when the money is public: the audit trail, the approval chain, and the reporting that decides whether the scheme is renewed.
The audience for the record is different
Running an incubation programme on public money is not harder than running one on private money. It is a different job in one specific respect: the record has to convince someone who was not there and has no reason to trust you.
A private funder who likes your programme will accept a reasonable account of what happened. A public audit will not, and cannot - the person conducting it is accountable for accepting it.
That single difference drives almost everything below.
Write the reason down at the time
The recurring finding in scheme audits is not fraud. It is decisions that were probably fine and cannot be shown to have been fine, because the reason existed only in a conversation.
The decisions that need a recorded reason, every time:
- Why this applicant was admitted and that one was not
- Why this amount, rather than the amount originally sanctioned
- Why a tranche was released when a milestone was partially met
- Why a deadline was extended
- Why a commitment was stopped
None of these is difficult to record. All of them are nearly impossible to reconstruct six months later in a way that satisfies someone sceptical.
Separate selection from disbursement
This is the control auditors look for first, and its absence is a finding on its own even when nothing went wrong.
The person or committee that decides who enters the programme should not be the person who authorises money leaving. In a small centre that feels bureaucratic - there are four of you and everyone knows everything. Do it anyway, because the alternative is explaining, under audit, why one person could both admit and pay.
Practically: a selection panel decides admission, and a different named authority approves each release. Record both.
Eligibility, checked and evidenced
Every scheme has eligibility conditions, and the audit will test them. Two habits make this painless:
Check eligibility mechanically, before scoring. An ineligible application should never reach a panel. Record what was checked and what the answer was, not just that it passed.
Keep the evidence, not the assertion. "Incorporated within the eligible window" is an assertion. The incorporation certificate with its date is evidence. Collect the document at the point the condition is checked, because collecting it later means asking a company that is no longer in your programme.
Work from the scheme's own guidelines and your sanction letter for what the conditions actually are - they differ by scheme and they change between rounds.
The disbursement chain, recorded end to end
Public money needs to be traceable in both directions: from the sanction down to the startup, and from a payment back up to the source.
Sanction / funding source
↓
Programme allocation
↓
Startup commitment
↓
Tranche, against a milestone, with evidence and an approver
Against every release, five fields carry the audit: amount, date, milestone, evidence accepted, approver. The one most often missing is the last, and it is the one an auditor asks about first.
Over-allocation deserves a specific control: the system should make it impossible to commit more from a source than the source holds, checked at the moment of commitment rather than at reconciliation. Discovering it at reconciliation means it already happened.
Utilisation reporting: assemble as you go
Schemes require periodic utilisation reporting, and formats differ. What does not differ is that the report is far easier to produce if the underlying records were kept as events rather than reconstructed at quarter end.
If each tranche already carries its milestone, evidence and approver, the report is a query. If it does not, the report is three days of archaeology and the result is only as reliable as somebody's memory.
Two practices worth adopting regardless of format:
- Track committed-but-unreleased separately. It is the number that produces year-end surprise, because slipped milestones mean you can be far below disbursement targets while having no headroom to commit.
- Record variations as events, never by overwriting the original figure. The original, the revision, the reason, the approval, the date. Overwriting is how a clean set of records becomes an unexplainable one.
Access control is part of the audit
Who could see and change what is a legitimate audit question, and "everyone was an administrator" is a poor answer.
What is expected:
- Permissions granted by person and by function, not one shared level
- Finance separated from programme management
- External panel members scoped to the programme they evaluate, with access ending afterwards
- A record of who accessed what, particularly for funding
The practical version of this is simple: if someone leaves the panel, their access should end because the programme ended, not because somebody remembered.
Publishing outward
Many schemes require the incubator to publish data - cohorts supported, funding disbursed, outcomes - to a funder dashboard or a public portal.
The thing worth setting up early is a way to publish that does not involve a monthly spreadsheet. A keyed, rate-limited data feed with an access log is more work once and far less work every month afterwards, and it removes an entire category of transcription error.
What renewal actually turns on
Uncomfortable, and worth saying plainly: schemes are renewed on the strength of the record more often than on the strength of the outcomes.
An official signing a renewal is accepting risk. A centre with modest results and impeccable, complete, promptly-filed records is a low-risk renewal. A centre with better outcomes and reporting that arrives late and needs reconciliation is a high-risk one, whatever the underlying programme achieved.
That is not a reason to care less about outcomes. It is a reason to treat reporting as a core function with an owner, rather than as something done in the week before it is due.
A short readiness check
- Every admission decision has a recorded reason
- Selection and disbursement authority are separate people
- Eligibility evidence is stored, not just asserted
- Every tranche carries milestone, evidence and approver
- Over-allocation is blocked at commitment
- Variations are recorded as events, not overwrites
- Committed-but-unreleased is tracked and reviewed monthly
- Access is per person and ends when a role ends
- Reporting has a named owner, not a rota
Frequently asked questions
- What is different about running a government-funded incubator?
- The evidence standard, and who it is for. A private funder wants to know the programme works. A public funder needs a record that would survive audit by someone who was not there - which means the reason for every decision has to be written down at the time, not reconstructed.
- What do government scheme audits usually focus on?
- Eligibility of recipients, whether the selection followed the stated process, whether disbursement matched sanction, and whether utilisation is evidenced. Programme quality is rarely the issue; the record of the process almost always is.
- How should approval authority be structured?
- Separate the person who selects from the person who releases money. It is the single most effective internal control and the one auditors look for first. If one person can both admit a startup and disburse to it, that is a finding regardless of how carefully they behaved.
- What causes schemes not to be renewed?
- Reporting quality far more often than programme outcomes. A centre with modest results and impeccable records is renewed more readily than one with good outcomes it cannot evidence, because the second creates risk for the official signing the renewal.
