Renting the Future

Somewhere in the last fifteen years, without an announcement and without much resistance, the default relationship between a person and the things they use changed from ownership to access.

Software was first, then media, then storage, then tools, then vehicles, then — with a certain inevitability — the heated seats in the vehicles. The pattern is now sufficiently general that it is worth examining as a single phenomenon rather than as a series of industry-specific decisions, because the same mechanism is operating in all of them and it has consequences that follow from arithmetic rather than from anyone's intentions.

The mechanism

Under a transaction model, a firm sells a unit, recognises the revenue, and begins the following year at zero. The relationship with the customer resumes only when the customer decides to buy again, which may be never. Forecasting is difficult, revenue is lumpy, and the incentive at the moment of sale is to maximise the price of that sale.

Under a subscription model, revenue from an existing customer continues without further action by either party. A firm entering the year with substantial recurring revenue and modest attrition knows most of its income before it does anything at all. The valuation consequences are large: predictable revenue is worth a multiple of unpredictable revenue, and public markets price the difference explicitly.

This changes what a business optimises for, and the shift is genuinely favourable to customers in one specific respect.

Under the transaction model, the firm's interest ends at the moment of purchase. Under the subscription model, the firm's revenue depends on the customer not leaving, every single month, indefinitely. Deficiencies that once surfaced as bad reviews after the money had changed hands now surface as cancellations while the money is still arriving. The feedback loop shortens considerably, and products improved across several categories for exactly this reason and no other.

The arithmetic that governs behaviour

Three quantities determine how a subscription business acts, and an outside observer who knows them can predict a great deal.

The first is acquisition cost: everything spent on marketing and sales divided by the customers actually obtained. It is paid immediately and in full.

The second is lifetime value: gross margin per customer per period, multiplied by expected duration. It arrives slowly, in fragments, contingent on the customer remaining.

The third is churn: the proportion who leave each period. It is the most consequential of the three, because it appears in the denominator of duration. At two per cent monthly churn, the average customer lasts about four years. At five per cent, under two. The same product, the same price, the same acquisition cost — and a business worth double or half depending on a number many firms did not measure at all until they were forced to.

From this arithmetic one behaviour follows mechanically. Growth in a subscription business consumes cash, because acquisition is paid today and recovered over years. A firm growing quickly therefore looks unprofitable while being economically sound, and a firm that has stopped growing suddenly appears profitable while being in serious difficulty. Both illusions have been extensively exploited in both directions.

Where the interests diverge

So far the model is defensible. The divergence begins with a distinction that most commentary elides: there are two ways to keep a customer.

The first is to remain worth paying for. The second is to make leaving expensive.

These are entirely different strategies producing identical retention metrics, and no dashboard distinguishes them. Switching costs accumulate quietly — data held in a proprietary format, integrations built over years, a team trained on one interface, a photo library of fifteen thousand images sitting in one provider's storage. Once the cost of leaving exceeds the annoyance of staying, price increases become possible that the market would not have accepted at acquisition. The industry term for this sequence is polite, and the practice is old enough to have been studied in every network industry of the previous century.

The tell is cancellation. A firm confident in its value proposition makes leaving easy, because it costs nothing to release someone who will return. A firm relying on inertia makes leaving a process: a retention flow, a telephone queue, an offer, a second offer, a confirmation worded to imply foolishness. That asymmetry between the ease of starting and the difficulty of stopping is now a regulated matter in several jurisdictions, which tells you how reliably it recurs.

The second model: paying with attention

Alongside subscription, the other dominant structure charges nothing and sells access to the user's attention. The economics are inverted but the logic is parallel.

Here the optimised quantity is engagement — time, sessions, actions — because inventory for sale is a direct function of it. And here the divergence is sharper, because engagement and benefit are correlated only loosely and sometimes not at all. A person who intended to spend ten minutes and spent ninety has generated more revenue and may be worse off. No equivalent of churn corrects this, since the cost of continuing is not billed.

The structural point is worth stating precisely. Under subscription, the firm profits when you find the product worth paying for. Under attention funding, the firm profits when you use it, whether or not it served you. The first has a built-in, if imperfect, alignment. The second must import its alignment from outside — from professional norms, from regulation, or from competition for users who have noticed.

What is actually being rented

The last consequence is the one least discussed at the time of adoption, and it is not primarily financial.

A purchased object is inert and yours. It does not change overnight, it cannot be withdrawn, and it continues to function if the manufacturer ceases to exist. An accessed service is a continuing relationship with a counterparty, and its terms are theirs to alter. Features are removed as well as added. Interfaces change on a schedule you did not set. Prices are revised. A product can be discontinued, and everything that depended on it stops — not degrading gracefully like an old tool, but ceasing, on an announced date.

This is the trade that was made, largely without being articulated: continuous improvement and low entry cost, exchanged for permanence and control. For most software, most of the time, it is a trade worth making, and the improvement has been real.

It is worth being conscious that it was a trade, and worth asking of anything you now access rather than own: what exactly happens to this if the company changes its mind — and how much of what I have built on it comes apart when it does?