Aug 25, 2026 · 12 min read
What Does This Actually Cost?
The invoice is the smallest of five numbers. How to find the other four before you sign, and why funding in terms leaves a cliff nobody costed for.
There is a quote on your desk. It says a number, the number is large but not impossible, and somebody wants a decision by the end of the month.
The number on the quote is the smallest of five numbers, and it is usually the only one anybody writes down.
That is true everywhere, and it bites harder in a nonprofit for a reason that has nothing to do with anyone being careless. Funding arrives in terms. Software does not end when the term does. So a system can be funded correctly, run well, report brilliantly, and still arrive at a year in which nothing pays for it, by which point the grant has closed and the person who wrote the bid has moved on.
The five numbers
Find all five before you sign. Four of them take a morning between them, and the fifth is a conversation with your finance lead.
1. The invoice
How to get it: it is on the quote.
Why it misleads: it is the number both sides are most comfortable discussing, so it absorbs all the attention in the room. In most projects I look at, it is somewhere between a third and a half of what the thing costs over five years, because it is the only part that fits neatly on a page.
Check the tax before anything else. If your organisation cannot reclaim tax on this purchase, and many nonprofits cannot on most of what they buy, then the quote is not the invoice. Add the tax to both the implementation figure and the annual one, and do it before the number goes into any funding bid, because a bid written at the pre-tax figure is short on the day it is approved. This is the most common way I see a project start life underfunded, and the rules differ by country and by what you are buying, so ask your finance lead rather than assuming.
2. Your own staff hours, priced
How to get it: count the hours the work takes today, then price them at the fully loaded cost of the people doing it, salary plus employer costs. Your finance lead has that figure and will produce it in ten minutes.
Why it misleads: salaries are already committed, so staff time gets treated as free. It is not free. It is the largest thing you spend money on and the only one nobody itemises.
Do it once with real numbers and it changes arguments. Four people spending forty minutes a day moving information between two systems is about fifty-six hours a month across the team, once you allow for working days. Over a year that is roughly six hundred and seventy hours, which at a modest loaded rate is comfortably into five figures on a workaround everybody has been describing as free. That is often more than the software costs.
Then say out loud which kind of saving it is, because this is where business cases get taken apart. If nobody leaves and no post is cut, you do not bank that money. The hours get absorbed into other work, and your cash position at year end is identical. That is a real benefit and it is not a cashable one, and the two belong in different parts of the paper. Put the hours in the case for doing it, and never net them against the cost, because the first trustee who has run a budget will spot it and you will lose the argument you should have won.
There is one exception worth looking for. If those hours let you avoid a hire you were about to make, that is genuinely cashable and you should say so loudly, because it is the strongest version of this argument and it is the one people forget to check.
The number cuts both ways, and you should let it. Sometimes it says buy the thing immediately. Sometimes it says the manual process is genuinely cheaper than the licence and you should stop feeling bad about the spreadsheet.
3. The three year total, not the first year
How to get it: ask the vendor for a three year cost including licences, support, hosting, and any upgrade they know is coming. In writing. Then add the internal cost of whoever will administer it, which the vendor will not include because it is not theirs.
Why it misleads: first year pricing is frequently a discount, sometimes an aggressive one, and comparing two options on year one compares a deposit with a purchase price. It is also the number that has to outlive the grant, which is the subject of the next section.
4. What it costs to leave
How to get it: ask about the contract first and the data second, because most people do it the other way round and the contract is the expensive half of the two.
The contract question: what is the minimum term, what is the notice period, and what do we owe if we exit in year two? Signing three years is not an annual cost, it is a commitment for the whole term, and it belongs in your commitments note as one. Negotiating a three-year term down to twelve months rolling will usually de-risk a deal more than anything else you can do to it, and it costs you nothing to ask.
The data question: who owns the data, and in what format can we get all of it out. Get it in writing and treat vagueness as an answer in itself.
Why it misleads: this cost is zero until the day it is enormous. A system you cannot leave is not a purchase, it is a position, and the price of the position appears at renewal, when the increase arrives and everybody in the room already knows you are not going anywhere.
5. Who pays when it breaks, and out of which budget line
How to get it: name the budget line, not the department, and check it exists next year. Then take the recurring figure to whoever owns your reserves policy and ask what an unfunded commitment of that size for the length of the term does to your free reserves position. That is a five minute conversation and it is the one that decides this.
Why it misleads: this is the one that actually closes projects. Everything above is a number. This one asks whether the number has anywhere to live, and in an organisation funded in cycles that is a different question with a different answer.
The one that ends things
| Number | Who usually knows it | Survives the grant ending? |
|---|---|---|
| The invoice | Everyone | One-off, so yes |
| Staff hours, priced | Nobody, until asked | Never was funded |
| Three year total | The vendor | Only to year three |
| Cost of leaving | Nobody | Falls on reserves |
| Who pays when it breaks | Nobody | This is the problem |
Look down the right hand column. Everything a grant is comfortable with is one-off. Everything that recurs has to survive a funder moving on, and keeping a system alive is not a project, has no launch, and photographs badly in a report.
I want to be careful here, because the lazy version of this argument is that funders will not pay running costs and fundraisers have not thought to ask. Neither is true. Full cost recovery has been ordinary practice for well over a decade, most serious funders expect a properly apportioned overhead line, and your fundraising lead has almost certainly been putting one in for years.
The real problem is not refusal, it is the term. Funders fund three years. Software runs for ten. So a bid that correctly includes the running cost still produces a cliff in year four, and everybody involved did their job properly on the way to it.
Which means the fix is an accounting one, not a fundraising one. Put the recurring cost into your support cost pool from year one and apportion it across the whole restricted portfolio, on the same basis you already use for finance, IT and premises. Then it is carried by many grants for as long as you have grants, instead of depending on one funder volunteering to keep a system alive that has nothing to do with the outcomes they care about. That conversation is with your finance lead, it is unglamorous, and it is what actually prevents the pattern below.
Do not treat the pool as the whole answer, though. Recovery is usually partial, because plenty of funders cap overhead at ten or fifteen percent or apply a flat rate, and the pool only recovers against the restricted income you actually win. So it softens the cliff rather than removing it. Work out what is left over and where that comes from before you rely on it, because an unfunded residual you have sized is a budget line and an unfunded residual you have not is exactly the surprise this article is about.
Because the pattern is reliable. A funder pays for a new case management system. It goes live, it works, everyone is pleased and the report is genuinely good. Two years later the licence renews, the person who ran it has left, there is no line for support, and the organisation is running critical operations on something nobody owns. The system did not fail. The funding model did what it always does, and nobody moved the cost anywhere it could survive the grant ending.
If the recurring cost will not fit in the support pool and no funder will carry it, that is a real decision rather than a delayed one. Find it from unrestricted funds before you start, or build something smaller you can afford to keep. Smaller and kept beats ambitious and abandoned, and it is a much easier conversation now than after a launch event.
The three questions to ask any vendor
Everything above compresses into three questions, and you can ask all of them without knowing anything technical.
What does year three cost, in writing? Not year one, and include the increases you expect. A vendor who answers plainly is telling you they intend to still be there. One who cannot or will not is telling you something too.
What is the minimum term, and what do we owe if we leave in year two? This is the one people skip and it is the one with money in it. You are asking what you are committing to, not what you are paying this year.
How do we get our data out, in what format, and what do you charge for that? Nobody enjoys this question and good vendors answer it without flinching, because they are not relying on you being stuck.
Send all three in one email. The pairing is what makes it work, because the first is about their pricing and the other two are about their confidence.
When the answer is that you cannot afford it
Sometimes you do the five numbers and the answer is no, or at least not this version of it.
That is a good outcome and it rarely feels like one, particularly if the project has already been announced internally, or a funder is enthusiastic, or somebody has spent months getting it this far. There is real cost to stopping and most of it is social.
Finding out in August that you cannot sustain something is worth far more than finding out in year three, when it is running, when people depend on it, and when stopping means telling your staff that the system they have built their week around is going away.
A good deal of what I am asked to look at is somebody else’s abandoned project, and those cost more to sort out than starting from nothing would have. So when I say the running cost decides this, that is where it comes from.
The person who will pay for this in practice
Every number above eventually lands on somebody, and it is usually not the person who signed.
It is the operations manager who becomes the unofficial administrator of a system nobody budgeted to administer. They learn it because somebody has to, they end up the only person who understands the configuration, and none of it appears in their job description or their review. When they leave, and they will, the knowledge goes with them, and the organisation finds out it has been running on one person’s goodwill for three years.
You can see this coming from the day you sign. Ask who will run this thing, name them, and then either give them the hours or admit you are asking for a favour rather than assigning work. Both are legitimate, only one is honest, and only one survives that person moving on.
Five numbers, one morning, and the awkward one is the fifth.
The rest of this is easier if you are asking it before you sign rather than after. If you are further along than that, I wrote separately about evaluating a proposal when you are not technical, about whether to build at all, and about what to do when a project you have already signed is going badly.
If you want somebody technical on your side of the table for this, that is what a technology assessment is. I go through what you already pay for, what is still done by hand and how many hours it takes, and what you are being sold. You get back what is worth doing, what to drop, what you are paying for twice, and what to leave alone, with the numbers attached to each so your finance lead can check my working.