SaaS and enterprise software

Ask a chief information officer how a piece of enterprise software was chosen and you will hear about a selection process. Ask the same person how the price was set and you will hear about a renewal. The two conversations describe the same product and share almost no facts, and the second one is where the money is.

This is not a British peculiarity, but Britain produced an unusually well-evidenced example of it. In the final decision of its cloud market investigation, the Competition and Markets Authority found that Microsoft's software licensing practices are adversely affecting the ability of AWS and Google to compete for customers who run that software, that certain products are not made available to those rivals through licensing agreements at all, and that customers holding existing licences cannot in most cases bring them across. The summary of that decision, dated 31 July 2025, is worth the twenty minutes it takes to read, because it is a regulator describing in plain language the mechanism most enterprise buyers experience and cannot name.

The public sector version is documented, and the shape is familiar

The State of Digital Government Review recorded roughly £26 billion of public sector technology and data spending in 2023. Less than 20% of it, about £5 billion, went on permanent public sector staff. Some 55%, about £14.5 billion, went to contractors, managed service providers and IT consultants.

That ratio is a buying decision rather than a fact of nature, and it has a consequence that shows up at every renewal. Capability you rent cannot negotiate on your behalf, because the people who know what the system does are employed by the people you are negotiating with. The organisations that hold their software costs flat tend to be the ones that kept a small number of expensive permanent engineers who understand the estate, which is an unfashionable and consistently effective position.

The competition regulator noticed the same concentration on the public side. In the decision cited above it observed that Microsoft and AWS appear to be the largest suppliers to the public sector, consistent with their overall market positions, and suggested that government keep monitoring the outcomes of public procurement and drive better practice in how the buying frameworks are used. A framework is a shortlist, and a shortlist compiled once becomes a default that outlives whoever wrote it. How that plays out in practice is the subject of the buying process nobody puts on the website.

Pricing is moving to consumption, and slower than the pitch deck says

Per-seat pricing has been declared dead every year for a decade. It is not dead. It is being supplemented, mostly by consumption or usage components bolted onto a seat-based floor, which is a structure that transfers forecasting risk to the buyer while retaining the vendor's revenue certainty.

Anyone claiming a precise figure for how far British contracts have shifted is inventing it. There is no public register of enterprise software contract terms, private sector pricing is covered by confidentiality clauses, and vendor disclosures aggregate at a level that hides the mix. What can be assembled is public sector award notices, listed vendors' filings and framework price lists, which is a partial view with a known bias toward large public buyers. That is the method used at per-seat pricing is dying more slowly than advertised, and the bias is stated in the piece rather than buried.

The ten-year cost is set by six clauses nobody reads

True-up terms, audit rights, uplift caps, the definition of a user, what happens to your data at termination and how long you have to extract it. Those six decide what a system costs across its life, they are written down in documents your organisation already holds, and they are almost never discussed because they are dull. Dull and consequential is the useful quadrant, and it is where this section spends its time.

The other half of the cost is organisational. Platform teams are usually blamed for their tooling when they fail, and the reasons are almost always reporting lines, funding models and the absence of anyone empowered to say no, which is the argument in platform teams fail for organisational reasons.

What consistently fails to predict any of this: feature comparison tables, category quadrants produced by firms paid by the vendors they rank, and the perennial promise that a new interface layer will finally make an enterprise resource planning migration painless. A vendor claim that can be checked against its own filings or against a regulator's findings gets checked against those here, and one that cannot be checked at all is not repeated. Related infrastructure coverage sits in cloud and infrastructure.