Show the listener the meter you are already running

Every audiobook subscription I know of runs a meter. The only question is whether the listener can see it. At Lisandre Audio we pay most publishers for each hour of their books that our subscribers hear, so a subscriber who finishes four long novels in a month costs us several times what a subscriber who listens to one memoir costs. Somewhere in every finance department in our industry there is a spreadsheet counting those hours by subscriber. My position is that the listener should be shown the same count the company is using, that plans should be sold as bands of listening with an honestly priced unlimited option at the top, and that the plan called unlimited but governed by a fair-use clause should be retired.

I hold this view partly because I once defended the opposite, and partly because the hidden version very nearly became our policy. It is not an argument that every subscription should be metered. Where content is paid for as a share of revenue rather than by use, or where a company owns its archive outright, an extra hour costs almost nothing and an unlimited plan can be exactly what it says. The argument applies where marginal cost follows usage and the company therefore has a reason to restrain it.

The unlimited plan that was never unlimited

When I joined Lisandre in 2021, our main plan was called Illimité and cost €14.99 a month. Its terms allowed us to restrict access in cases of exceptional use. Nobody had ever applied that clause. By 2022, about 61,000 of the 870,000 people on the plan listened to more than forty hours of books a month. They were 7% of its subscribers and 34% of its royalty hours, and each cost us around €23 a month in royalties alone. Our gross margin had slipped to 31%.

The proposal that reached our executive committee that autumn was reasonable on its own terms. We would apply the clause quietly: after thirty hours in a month, new releases would load behind a short queue. The paper argued that the heaviest listeners might notice the delay but were unlikely to connect it with their own usage. Several of us, me among them, had suggested milder versions of the same idea in earlier years.

What changed my mind was an afternoon with our complaint logs. Our most loyal subscribers were also our most observant. They noticed when a title they had waited for took longer than usual to appear, and they wrote to us in detail, sometimes with timings. A limit that works only while people fail to notice it would have been applied to exactly the people most likely to notice. Once discovered, it would have become the story they told about us in forums and reviews, and they would have been right to tell it.

Count in a unit the listener recognises

In March 2023 we replaced Illimité with three plans: fifteen hours a month for €11.99, forty hours for €17.99, and unlimited listening for €24.99, a price that covered what our heaviest listeners cost. Unused hours roll over for a month. A listener who reaches the limit partway through a chapter can finish it. Existing subscribers kept their old price for a year and were told, in the first message about the change, which plan their own listening suggested.

Our worst mistake was in what we counted. Royalties are calculated on hours of recording, the running length of the book, and we built the meter on the same basis. Many of our subscribers listen at one and a half times normal speed. For them, forty hours of recording took about twenty-seven hours of their evenings, and the meter seemed to race. In the first month around 1,900 subscribers wrote to say the counter was wrong. It was accurate, but it measured something they did not recognise. We changed the display to show hours of recording, with a sentence of explanation and a progress bar for each title, and the complaints fell away within two months. A visible meter has to count in a unit customers can check against their own experience. Otherwise it is only a more elaborate way of hiding the arithmetic.

There is a fair objection that meters make people anxious, and that anxious customers use less. Our data showed some of that effect: listening on the fifteen-hour plan dipped in the last week of each month. We softened it with a notice at eighty per cent of the allowance rather than at the limit itself, and with the roll-over. I would still rather a listener plan a month around a number they can see than lose access to a book on terms they were never shown.

What a visible meter costs

The change was not free, and anyone proposing it elsewhere should expect the bill. About 13,000 of the 61,000 heavy listeners cancelled within six months, many after writing to explain why. Overall monthly churn rose from 3.4% to 3.9% for two quarters. Our head of growth, Inès Kerbrat, had argued in writing against a spring launch, when renewals are weakest anyway, and the first two quarters proved her right. On the other hand, about 22,000 heavy listeners chose the €24.99 plan, which surprised our finance team more than anyone.

By the end of 2024 churn had fallen to 3.1%, and in the first half of 2026 it averaged 2.8%. Average revenue per direct subscription rose from €13.20 to €14.60, and gross margin reached 38% in 2024. The churn improvement did not come from the plans alone, because we changed onboarding and recommendations in the same period, and I cannot divide the credit precisely. The margin improvement came mainly from the plans.

I would suggest three tests to any product leader or board looking at a subscription whose costs follow usage. First, write down the meter the company already runs internally, customer by customer, and ask whether customers could reconstruct it from what they are shown. Second, if the terms contain a limit, ask what would happen if every customer read that clause tomorrow. If the honest answer is that many would leave, the limit is not a policy but a wager on inattention. Third, price unlimited at what unlimited costs. Some people want it and will pay for it, and offering it openly removes the temptation to ration them in secret.

None of this makes a metered plan pleasant. It makes it legible. In a market where several services describe slightly different limits in slightly different language, legibility is something a listener can compare, remember and, in time, come back to.