The real cost of a SaaS subscription exceeds its sticker price

This article's thesis is simple and falsifiable: the subscription price is only the visible part of what a SaaS really costs, the rest — integration, training, administration, forced upgrades, exit cost — staying off the price list. If, over three years, your tools cost exactly what the public price list announced, the thesis is wrong for you: the final model lets you check.

This is not a niche topic: 52.7% of EU enterprises used paid cloud services in 2025, up from 17.8% in 2014 (Eurostat, February 2026). Subscriptions have become the default way to buy software.

The analytical framework is more than thirty years old: procurement literature has formalized total cost of ownership (TCO) since Ellram (1993) — pre-transaction, transaction and post-transaction costs, including indirect costs: training, peer-to-peer help, lost time. Few SMEs apply it to their subscriptions.

Four patterns make a tool budget drift, all readable on public price lists checked on August 12, 2026.

These mechanisms stack: the "SaaS inflation rate" measured by Vertice from more than two million price points reached 16.4% in June 2026, nearly four times the US price index; 79% of IT leaders faced a renewal increase in the past twelve months and 61% cut projects to absorb it (Zylo, January 2026).

The TCO definition places training and administration among indirect costs — exactly what SMEs underinvest in. In 2024, 22% of EU enterprises trained their staff in digital skills: 73% of large companies, but only 21% of SMEs (Eurostat). SMEs therefore pay full license price without the investment that determines the tool's return.

The symptom is measurable: 36% of licenses go unused on average in the organizations analyzed by Zylo — 305 applications per organization, 81% of spend decided by business teams. A base dominated by large US organizations: an order of magnitude rather than an exact figure for a European SME.

Then comes administration — accounts, permissions, departures, internal support: hours nobody invoices, so nobody sees. Costing them before signing is part of our method.

The strongest argument comes from the law. The European Data Act (Regulation 2023/2854), applicable since September 12, 2025, structures how you leave a cloud or SaaS service: a maximum two-month notice, a 30-day transition — extendable to seven months where technically necessary — then at least 30 days to retrieve your data. Its article 29 goes further.

What the regulation says — From January 12, 2027, no more cloud or SaaS switching charges in the EU (art. 29); until then, only reduced charges, capped at actual migration costs, remain allowed.

If the legislator caps these fees and then bans them, it is because they were real and systemic. The market has followed: Google Cloud removed outbound transfer fees for departing customers (January 2024); AWS credits them on request, with migration completed within 90 days (September 2025).

The honest nuance: the regulator removes the invoiced part. The migration project — re-integration, re-training, parallel run — stays with the customer; it should be provisioned at signature, not on departure.

What it costs to do nothing — A 100,000-franc tool budget renewed at the SaaS inflation rate measured by Vertice (16.4% in June 2026) reaches roughly 116,000 francs in year two and 135,000 in year three; at the average rate measured by Zylo, more than a third of that sum funds empty seats.

Fill it in before signing, then at every renewal:

The total to compare between two solutions is never the "subscription" line: it is the cumulative three-year figure. That reasoning drives how we build: building blocks whose full cost, exit included, is known before signing. More models appear in our insights.

TCO is a comparison tool, not a verdict against subscriptions. A well-adopted SaaS often beats an in-house build on full cost — training and administration exist whatever the solution. The model measures only costs: an expensive tool can remain profitable if the value created outpaces the price drift. Finally, the waste and inflation figures come from customer bases of large US organizations: for a Swiss or European SME, they indicate a trend, not a prediction.