North Canterbury · © John Stroh
Drafted with AI assistance, then checked and revised by the author. The judgements and the errors are the author’s. How this site is written sets out what is declared on every piece, who checks it, and where the per-piece record lives.
How the decision stopped being yours, one ordinary purchase at a time.
Series part A · Version 0.1 · August 2026
You have probably had this conversation, or a version of it.
The renewal quote arrives and it is higher than last year. Not outrageously — twelve per cent, perhaps eighteen. The account manager explains that the plan you are on is being retired, that the new one includes capabilities you did not ask for, and that the price reflects them. You look at the alternatives. Moving would mean re-entering four years of records, retraining six people, rebuilding the connection to your accounting system, and finding out in month three what else was quietly depending on it.
So you sign. Everyone signs. Signing is the rational decision, and that is precisely the problem.
Repricing at renewal has become the market default rather than the exception. The Zylo 2026 SaaS Management Index found 79% of IT leaders had encountered a price increase at renewal in the preceding twelve months. Gartner figures reported alongside it put subscription cost rises from several large vendors at 10 to 20 per cent during 2025 — against IT budget growth of under three per cent.
Named vendor actions are firmer ground still, because the vendors published them:
Across the largest five hundred software companies there were 339 pricing and packaging changes in 2024 and 2025 alone. The terms of the software your organisation runs on changed 339 times in two years.
If it were only price, you could budget for it. The mechanism is wider, and most of it does not appear as a number going up.
The plan you are on is retired. You are not offered a higher price for the same thing. The same thing stops existing, and the nearest replacement costs more. This is now the largest single driver: vendors that added AI tiers during 2024 and 2025 began sunsetting the plans without them, so customers move up whether or not they want the features. One dataset tracking ninety-plus tools over six months attributes an average 24 per cent increase to this mechanism alone.
The tier gets thinner and the price does not. Analysis of late-2025 procurement data found 28 per cent of renewals involved reduced support, tightened service levels, or fewer features at the same base price. You are not paying more. You are getting less, which is the same thing arriving through a door nobody watches.
The minimum changes underneath you. When Broadcom restructured VMware, the minimum licence for vSphere Standard went from sixteen cores to seventy-two. For a large customer that is nothing. For a small one it is a bill for capacity you do not have and cannot use. A threshold change is neutral at scale and severe below it, which is why small organisations experience these transitions as something aimed at them even though nothing was.
The upgrade arrives on their schedule. Your workflow does not quite fit the standard, so someone adapted it. Then an upgrade you did not request and did not schedule breaks the adaptation, and your team spends weeks rebuilding. The cost is not only the weeks. It is that the person who understood the adaptation has now watched it be destroyed twice, and has stopped making them.
And something new is inside it that you did not buy. The current justification for most of this is AI. Which means the AI now doing work inside your practice management system, your membership platform, your accounting suite, arrived as a line item you never evaluated, from a supplier you cannot name, under terms you were not shown. That is the subject of the next part of this series, and it is the sharpest edge of the whole problem.
This is worth saying plainly, because the argument fails if it becomes a complaint about greed.
Every action above is an ordinary commercial decision by a firm with pricing power, made by people with a duty to their shareholders and no duty to you. The account manager is not lying. The product genuinely does cost more to run. AI capability is genuinely expensive. None of it requires bad faith and pointing at bad faith will get you nowhere, because there is none to find.
What has happened is structural. Software moved from something you bought to something you rent, and the thing you rent accumulates your work inside it. Every month of use makes leaving harder, which makes the next price rise easier to impose, which is not a conspiracy but simply the shape of the arrangement. Anyone in that position would eventually do the same, and the ones who did not would be replaced by ones who did.
Not money. Money is measurable and you can argue about it.
You lost the ability to say no. Somewhere between year one and year four, declining a change stopped being an option you could exercise. Nothing marked the moment. There was no meeting, no decision, no notification. Each individual step was reasonable and the sum of them was a position from which refusal is not available.
That is the whole thing, and it is why “price rises” is the wrong frame. A price you can refuse is a negotiation. A price you cannot refuse is a tax levied by a party you did not elect.
The obvious responses do not work, and it is worth being precise about why.
Switch vendors. To another firm with the same commercial model, the same incentive, and the same knowledge that you cannot easily leave. You reset the clock and buy perhaps three years.
Negotiate harder. A four-person practice has no leverage. Enterprise customers get discounts because their loss would be noticed.
Run it yourself. Now more affordable than most people realise — that is the substance of parts C and D of this series. But it needs skills you probably do not have, someone to keep it running in year three, and a way to get help when it breaks at four in the afternoon.
Wait for regulation. Part H makes the case for what a government could do, and some of it would help. None of it arrives at the speed of your next renewal.
Each of these fails for the same reason: you are one organisation facing a party far larger than you, and the imbalance is the problem. Every remedy available to you individually leaves that imbalance intact.
An organisation the size of a plumbing firm cannot run its own infrastructure, cannot negotiate, and cannot wait. Forty of them together can do all three.
Not as customers of a better vendor — that just moves the same relationship. As owners of the thing they depend on, so that the terms cannot be changed against them by anyone whose interests differ from theirs, because there is nobody in that position.
That is not a new idea. It is what farmers did about processing, what retailers did about wholesale buying, and what rural communities did about insurance and electricity, in this country, more than once, when facing exactly this imbalance against parties far larger than themselves. The form is well understood here and the law for it already exists.
What is new is the thing being co-operatively owned: not a dairy factory but the software and the machines a modern organisation runs on, and the ability to use AI without handing over the records that make it useful.
The rest of this series is the blueprint for that — what it must do, what it costs, how it is governed, and what a government could do to make it easier. It is published for anyone to build from. Nothing in it asks you to buy anything.
Vendor actions above are as announced by the vendors. Survey figures are from the Zylo 2026 SaaS Management Index and Gartner and Forrester data reported in secondary sources. Percentages attributed to price-tracking datasets and procurement analyses come from software asset management vendors and licensing consultancies — firms that hold the data because they negotiate these renewals, and that also sell services predicated on the problem being severe. Directions are reliable; treat magnitudes as indicative.
Two gaps we did not fill. We found no New Zealand–specific data on small-organisation software costs or renewal experience, and nothing on how New Zealand public bodies have responded to these particular changes. Both would strengthen this part and neither should be inferred from international figures.
If you have New Zealand data on either, we would rather publish a corrected version than a defended one.
Published under CC BY 4.0.
What this publication does not claim, and what is outstanding against it in the register.
A question this rests on is parked: What do we do with an application that refuses to run inside the boundary?
We do not claim that a member's existing applications can be made to run inside the boundary. We claim only that the boundary reveals which ones cannot.
Alongside: questions and answers · glossary · sources and provenance · slides