Subscriptions stopped being a growth story

One number in Sage’s results for the year ended 30 September 2025 explains more about the state of the subscription model than any analyst commentary published that month. The company reported a renewal rate by value of 101%, unchanged on the previous year.

Renewal by value counts what last year’s customers are worth this year. At 101%, the installed base contributes one percentage point of growth on its own, before a single new customer signs anything. That is a healthy figure and a sobering one. It means the growth a subscription business reports has to come from somewhere other than its existing customers behaving well, and in practice that somewhere is price, cross-sell into adjacent modules, and acquisition.

The rest of the FY25 disclosure fills in the picture. Underlying total revenue up 10% to £2,513m. Subscription revenue up 12% to £2,093m, from £1,876m. Subscription penetration up to 83%, from 82%. Annualised recurring revenue up 11% to £2,574m, and underlying operating margin up 150 basis points to 23.9%.

Note the relationship between two of those. Annualised recurring revenue, at £2,574m, exceeds reported revenue of £2,513m, because it is a run rate measured at the year end rather than a sum of what was invoiced across it. This is not sleight of hand and every subscription business discloses it, but a reader comparing an ARR headline against a competitor’s revenue line is comparing a photograph with a film.

The transition finished, and nobody rang a bell

For twenty years the interesting question about subscriptions was conversion. How fast could a licence business move its base onto a recurring contract, what would it do to reported revenue during the crossover, and would customers tolerate paying annually for something they used to own.

At 83% penetration, and after a decade of climbing a percentage point or two a year, that question has answered itself for a company like Sage. The migration is essentially complete. What remains is a business whose economics are set almost entirely by retention, pricing power and the cost of serving customers who never leave and never spend much more either.

This is a different management problem from the one the model was sold on. Growth-era subscription thinking optimised for acquisition, on the reasoning that a customer once acquired produces revenue indefinitely. Mature subscription thinking optimises for the second and third renewal, where the incremental cost of keeping a customer is small and the incremental revenue depends on persuading them to buy something more. Those two disciplines require different organisations, different metrics and different people, which is why so many software companies have spent the last three years reorganising their commercial functions without saying so publicly.

Then Parliament made the renewal an event

The second thing that changed is legal, and it applies to consumer subscriptions rather than enterprise ones, which is precisely why it deserves attention from people who sell to businesses.

Part 4 Chapter 2 of the Digital Markets, Competition and Consumers Act 2024 rebuilds the mechanics of a consumer subscription. Traders must give specified pre-contract information, and for online contracts the key information must be given in writing in such a way that the consumer is not required to take any steps to read it beyond the steps needed to enter the contract. A reminder notice must be given in respect of each renewal payment relating to the end of a relevant six-month period, and it must be more prominent than any other information given at the same time. Consumers get a fourteen-day initial cooling-off period and a further fourteen days after certain renewals. The right to cancel is exercisable in any circumstances, and no penalty or charge may be imposed for exercising it.

The Department for Business and Trade published its response to the implementation consultation on 2 April 2026, reporting 75 responses and stating that the regime is anticipated to commence in spring 2027. It also confirms the refund position: no supply before cancellation means a full refund, and supply already begun means a proportionate one. Where a trader fails to inform a consumer of cooling-off rights, the period extends by up to twelve months.

The same day’s announcement put a value on it, estimating £400 million of consumer benefit a year and around £14 a month, nearly £170 a year, for each unwanted subscription shed.

What that £400 million actually is

It is revenue. Specifically, it is revenue currently recognised by subscription businesses from customers who have stopped wanting the product and have not got round to leaving, and the state has now quantified it and set a date for removing it.

That is the part worth sitting with. The consumer subscription model has, for two decades, quietly capitalised on the renewal being invisible. Not fraudulently in most cases, simply by default: the payment recurs, the reminder is absent, the cancellation route is three screens deep, and inertia does the rest. A prominent reminder notice attached to every renewal period, plus an unconditional cancellation right, converts an invisible event into a decision point. Every decision point loses some customers.

Businesses that have run on that inertia will discover their true retention rate at the same moment their competitors discover theirs. Businesses that have not will find compliance to be an operational nuisance and a competitive gift.

The engineering consequence is larger than the legal one and it is being underestimated. Most billing platforms in use today were designed on an assumption that a renewal, once taken, is final. A proportionate refund obligation attached to a cancellation window means the system has to calculate the unused portion of a period, reverse part of a charge that has already been recognised as revenue, and do it within a deadline, for every product in the catalogue including the ones acquired with a company three years ago. Organisations discovering this in the final quarter before commencement will be buying consultancy at a premium, because everybody else will be discovering it at the same time.

There is a measurement point too. Monthly churn, the metric most consumer subscription businesses report internally, will become considerably less informative once cancellations cluster around two defined windows rather than spreading across the year. A business that manages to a monthly figure will see the number move and will not immediately know whether it is looking at a change in customer sentiment or at the arithmetic of a new calendar.

The enterprise version is unregulated and identical in structure

None of this reaches business-to-business contracts. An enterprise software agreement can still auto-renew on notice periods measured in months, with escalators written years earlier, and the counterparty is presumed capable of reading its own contracts.

The presumption is generous. The mechanism that traps a consumer, which is a renewal nobody is required to draw attention to, operates on a corporate buyer through a different route: the renewal lands with a procurement team that did not negotiate the original, on a system nobody has evaluated since implementation, with a switching cost that has grown every year since. The absence of a reminder notice is not the problem. The absence of anyone whose job it is to ask whether the thing is still worth the money is.

Which suggests the practical response for a corporate buyer, and it is to build the reminder Parliament has not required. A renewal calendar held by someone independent of the system owner, with a review triggered a full cycle ahead of the notice deadline, does most of what the consumer regime does, and it costs nothing but diary discipline. How that plays out against the vendors is covered in how enterprises actually buy the software they run, and the parallel shift in how the meter itself is set is taken up in why per-seat pricing is dying slowly. Both sit under SaaS and enterprise software.