More than your accounts show. The real cost of an SME's software subscriptions adds up a visible cost — the sum of monthly invoices, which almost no one tracks all the way through — and four invisible costs: teams adapting to the tools' limitations, annual price hikes absorbed at renewal, data that has become hard to extract, and exit costs no one budgets for. One hour of stocktaking is enough to arrive at an honest first figure. Here is the method I use.
The visible cost: the sum nobody adds up
A company's software subscriptions almost never live in one place. Some run through the company card, others are billed annually and buried in operating expenses, still others get reimbursed as expenses because an employee pulled out their own card to unblock a project. Add to that free tiers that became paid without any formal decision, and per-seat pricing that scaled with headcount without anyone ever re-validating the scope.
The result: accounting sees line items, never a system. The annual total exists nowhere.
I did this exercise at my own scale before building FiscalDoc, my local tax-filing application. The market solutions I had shortlisted ranged from CHF 25 to 60 per month per entity, i.e. CHF 600 to 1,400 per year for two tiny scopes — a one-person practice and a household. An SME with twenty employees multiplies that kind of line item by the number of functions covered: customer relations, project management, e-signature, storage, accounting, marketing. The total then runs into the thousands of francs per year, excluding VAT — often without any budget line carrying that name.
What are the four invisible costs of a subscription?
My working definition, as of 2026: the real cost of a software subscription is the sum of five items — the invoice, the team's adaptation to the tool's limitations, price hikes absorbed at renewal, data captivity, and exit cost. Only the first appears in the accounts. Each of the other four deserves scrutiny.
Adapting to the tool's limitations. A rented piece of software is generic by design: it serves thousands of customers with needs different from yours. When it doesn't fit the way you work, it isn't the tool that bends — it's you. Exports to spreadsheets, double entry between two tools that ignore each other, a "notes" field turned into a catch-all: every workaround eats up hours that show up on no invoice. It's the question "who adapted to whom?" that I already raised in SaaS or custom-built, and it's the first item the inventory needs to make visible.
Absorbed annual price hikes. Among established vendors, raising prices on the installed base has become an openly embraced growth lever. It arrives by email, a few weeks before the deadline, at the exact moment switching tools would cost more than accepting the increase. Taken in isolation, each hike looks reasonable. Compounded over several years, they result in paying noticeably more for a service that has stayed largely the same — without any decision ever having been made inside the company.
Data that's hard to get out. Export almost always exists on paper. In practice, it often produces a flat file where the tool actually held links between records, histories, attachments, and automations. The more your data piles up inside a subscription, the more it becomes the vendor's best retention argument: you're no longer just paying for a function, you're paying for access to your own history.
The exit cost nobody budgets for. Migrating data, retraining the team, running the old and new tools in parallel during the transition: this expense is very real, and nobody budgets for it at signing. Yet its absence weighs on everything else — it is precisely because the exit is never costed out that the price hikes from the previous paragraph get accepted. An unbudgeted exit is a negotiation lost in advance.
How do you do the inventory in one hour?
No audit, no mandate needed. A bank statement, a spreadsheet, one quiet hour:
- Pull twelve months of payments — company card, annual invoices, expense reports — and highlight every recurring payment to a software vendor.
- For each line, note the real annual cost excluding VAT, the number of paid seats, and the number of users actually active last month.
- Add the renewal date and the contract's cancellation notice period.
- Note where the data is hosted and whether a full export has actually been tested — tested, not assumed.
- Flag with a cross the tools the team is already working around with spreadsheets.
The result fits on one page. This table doesn't decide anything for you; it makes the decision possible, which changes everything compared with the starting situation, where every subscription renewed itself silently, alone. On first read, two columns are usually the most instructive: paid-but-inactive seats, and the crosses — the tools your own teams are already working around.
What do you do with the total once it exists?
Above all, don't cancel everything. A high total doesn't say "cut"; it says "look." The trade-off is made tool by tool, and it needs a grid: bounded versus evolving need, data sensitivity, network effect, functional complexity, reversibility. I detailed it in Buy or subscribe? The decision grid, which is the direct sequel to this inventory.
Some subscriptions will come out of the review stronger — a genuine network effect, compliance carried by the vendor, a function that just works without drawing attention: I gathered these cases in the situations where keeping your SaaS is the right call. Other lines, on the contrary, combine high cost, narrow use, and sensitive data. That's the zone where an owned tool, built close to the actual need, advantageously replaces renting. FiscalDoc is the proof at my own scale: three evenings of building through dialogue with a coding assistant, zero francs of operating cost, and the CHF 600 to 1,400 in annual subscriptions avoided — the case is documented from start to finish.
This inventory is, in fact, the first thing I ask for before any custom-built project: without it, tools get replaced by guesswork. And to place the question — what to rent, what to own, what to automate — within the full set of digital decisions facing an SME, the roadmap is laid out in What should a Swiss SME do about AI in 2026?
Key takeaways
— The invoice is only one item out of five: adaptation to the tool's limitations, absorbed price hikes, captive data, and unbudgeted exit costs appear in no accounting record. — One hour is enough for the inventory: annual cost excluding VAT, seats actually active, renewal date, tested exportability, workarounds by the team. — The total doesn't say what to cancel; it says where to look — the trade-off is then made tool by tool, using a grid.
FAQ
Is there a "normal" software budget for a Swiss SME? Not in any useful form: differences across sectors and equipment levels make averages misleading. The benchmark that matters is internal — your annual total excluding VAT, set against the functions actually in use. That's what the one-hour inventory provides, not a market statistic.
Should underused tools be cancelled immediately? No. First check the renewal date, the cancellation notice period, and data export; some lightly used tools also carry an obligation — archiving, compliance — that justifies keeping them. Cancellation is a second-pass decision, never a first-read reflex.
How do you estimate the exit cost of a subscription? Test the full export, measure what's missing on arrival — links, histories, attachments — then estimate the work to recover those gaps, the training, and the dual-operation period. An order of magnitude is enough; what matters is that it exists before the next renewal negotiation, not after.
Should free tools be included in the inventory? Yes. A free tier hosts data and creates dependency exactly like a paid tier; its cost is measured in captivity rather than in francs. And it's often the one that becomes paid with little warning, once the team has grown used to it.
Did the total surprise you? The AI Usage Diagnostic: sixty minutes to map your actual workflows, identify what deserves a custom build, what stays SaaS, and what needs no AI at all. Book a diagnostic
Jérôme Deshaie is CEO and founder of MCVA Consulting SA, an augmented agency based in Valais. Fifteen years serving major international brands, now working directly with Swiss SMEs. Background.