When a licence-waste audit isn’t worth buying — and the arithmetic that decides it
We turned a company away this summer because our own guarantee would not have been honest at their size. Here is the gate, the arithmetic behind it, and the question it hands you for your next vendor call: at what size does your promise stop being true?
A licence-waste audit is not worth buying below roughly €40,000 a year in visible SaaS and AI licence spend — about €3,300 a month, or 65-odd paid seats. Below that line the recoverable money is usually too small to cover a serious piece of audit work, and any supplier guaranteeing you a multiple of their fee is guaranteeing something the arithmetic cannot reach.
We know that number precisely because it is ours, and because it cost us a deal this summer.
The call we made in July
A company came to us wanting the audit. We wanted the work. The arithmetic said no, and we told them so.
They were not a bad prospect. They were a small one — well under the gate — and the promise we attach to that audit would not have been honest at their size. So we said the sentence that is rare in this market: our best offer is the wrong size for you, here is the smaller piece of work that actually fits.
We publish the arithmetic below because a gate you cannot see is not a gate. It is a filter the seller applies privately, and you are meant to trust that it exists.
What is a 3× audit guarantee, and where does it stop being true?
Our guarantee is that a scoped AI and Tool Waste Audit finds at least three times its own fee in recoverable spend, or you do not pay for it. That is a real commitment, and like every real commitment it has a size at which it breaks.
The chain is four steps and you can check each one:
- The scoped audit has a floor of €2,000 (the anchor is 50,000 Kč; the euro figure is quoted as €2,000–3,000).
- Three times the floor is €6,000. That is the minimum recoverable spend we would have to find for the guarantee to hold at the entry price.
- We assume a conservative 15% waste rate. This is our own planning assumption, not a market statistic — deliberately below what industry data typically reports, because a guarantee sized on an optimistic assumption is a guarantee designed to fail.
- €6,000 ÷ 0.15 = €40,000 a year in visible licence spend. That is the gate.
Two independent checks land on the same place. Audit fees in this category tend to run 3–5% of visible annual spend; 5% of €40,000 is exactly €2,000, so the gate sits precisely where the floor stops being a discount. And if you prefer the other direction: €6,000 of estimated annual waste is the same test stated as an outcome rather than as spend.
What should a smaller company do instead?
Buy something smaller. Not nothing, and not the big thing at a discount.
For a founder-led or genuinely small business, we run a Small Audit at €1,000–1,500 — the same shape of work at human scale, ending in a roadmap the owner can actually run themselves. It carries no 3× guarantee, because at that size the guarantee would be theatre.
What we will not do is take the scoped audit fee from a company whose licence estate cannot mathematically return it and quietly hope. That version of the sale works exactly once per client, and the market is too small for once.
We applied the same test to our own free guide
In August we rewrote a free document we give away, for the same reason.
The guide had columns asking the reader to estimate "minutes with AI" and "hours saved" for tasks they had not yet tried. We had written a document that forbids invented numbers and then asked the reader for one. So the columns were relabelled to be filled in after a test rather than before it, and the prompt was instructed not to estimate savings pre-pilot.
The same pass withdrew two other things: an unconditional claim that the audit pays for itself — now stated conditionally, because most readers of that particular document sit below the gate above — and a promise that the work needed no new tool, which is not something we can promise to someone who does not yet pay for a single AI licence.
None of that made the document more persuasive. It made it true, which is a different and slower asset.
How do you tell an honest guarantee from a sales line?
Ask for the boundary. Any guarantee is a claim about a range, and a supplier who has done the arithmetic knows where their range ends.
The question, verbatim, for your next vendor call: at what size does this stop working?
Then watch what happens. Three responses, and they tell you different things:
- They give you a number and the reasoning behind it. They have modelled their own offer. You can now check whether you are inside it.
- They say it works for everyone. Either they have not done the maths, or they have and would rather you did not see it. Both are useful to know before you sign.
- They reframe the question as a lack of ambition. This is the one to walk away from.
A guarantee with no stated size is not a guarantee. It is a sales line wearing one.
Why we would rather publish the gate than optimise the funnel
There is an obvious commercial argument against this article: it tells a share of readers not to buy the thing we sell.
We publish it anyway, for a reason that is not altruism. We do not resell licences and have nothing to upsell, which is what makes it possible for us to tell a client to cancel a tool. That independence is the whole product. It is worth very little if we then apply a different standard to our own offer than to theirs — a supplier who will not name the limit of their own promise has no standing to audit anyone else's spend.
So the gate is published, in a place a prospect can read it before a call rather than discover it during one.
Where you should actually start
If you do not know your visible annual licence spend, that is the first number to get, and it is usually a morning's work: the last twelve months of software invoices, including the subscriptions on personal cards, added up.
If it is comfortably above €40,000, the guaranteed audit is a reasonable thing to consider — from us or from anyone who will state their own boundary as plainly.
If it is below, you do not need us to tell you what to cancel. You need a short piece of work that fits, and a roadmap you can run without a retainer.
Either way, get the number before anyone sells you a solution sized for it. The free Tool Waste Self-Check gives you an estimate in about two minutes, and it will tell some readers that their number is too small for the audit — which is the point.
A guarantee with no stated size is not a guarantee. Ask any supplier where their promise stops being true, and watch whether they have a number.