Track 4 · The House · lesson 9

Designed to be hard to leave

13 min

You run the subscription now. Fifty a month, 240,000 subscribers, and a dashboard with one number on it that the whole company is measured against.

Here is a figure from that dashboard that nobody puts in the annual report: about one subscriber in seven has not opened the product in ninety days. They are still paying. They are, by a distance, your most profitable customers, because they cost nothing to serve and they never contact support.

Revenue from someone who is not using the product is the highest-margin revenue a business can have. There is no cost of delivery at all.

Every subscription business discovers this. What separates them afterwards is what they do about it, and the range of legitimate responses is wider than it looks — from emailing dormant users to ask if they still want it, through saying nothing, to designing the cancel path so that saying nothing is easier.

All three are legal. Only one of them is a strategy.

Your levers, priced

Two teams want budget. Growth can bring in thirty percent more subscribers for the same spend. Product thinks it can cut monthly churn from five percent to three and a half. Both are credible. You can fund one.

Subscription lifetime value

Hypothetical model, not a forecast

Turns a monthly churn rate into an expected lifetime and a lifetime value, discounted and not.

It assumes

Every subscriber pays fifty a month at an eighty percent gross margin.
The same share of remaining subscribers leaves at the end of each month.
Dormant subscribers churn at the same rate as active ones, which flatters nobody.
The discounted figure charges one percent a month for waiting.

It ignores

Cohort effects, where churn falls as a cohort ages
Upsell, downgrade, pauses and reactivation
The cost of acquiring the customer in the first place
Any risk that the business itself does not last the lifetime shown

A subscriber is worth 800 at five percent churn. Growth's thirty percent would make each of them worth the same and give you more of them. Find the churn rate at which one subscriber is instead worth more than 1,500 — and compare the two offers.

Customers lost each month500%

Gross profit a month: 40
Expected months as a customer: 20
Lifetime value, undiscounted: 800
Lifetime value, discounted: 673
Collected by the horizon: 763
Share of lifetime value collected by the horizon: 0.954

Collected per subscriber
Share still subscribing

A rising collection curve flattening as the surviving-share curve decays towards zero.
Month	Collected per subscriber	Share still subscribing
0	0	1
1	40	0.95
2	78	0.902
3	114	0.857
4	148	0.815
5	181	0.774
6	212	0.735
7	241	0.698
8	269	0.663
9	296	0.63
10	321	0.599
11	345	0.569
12	368	0.54
13	389	0.513
14	410	0.488
15	429	0.463
16	448	0.44
17	466	0.418
18	482	0.397
19	498	0.377
20	513	0.358
21	528	0.341
22	541	0.324
23	554	0.307
24	566	0.292
25	578	0.277
26	589	0.264
27	600	0.25
28	610	0.238
29	619	0.226
30	628	0.215
31	637	0.204
32	645	0.194
33	653	0.184
34	660	0.175
35	667	0.166
36	674	0.158
37	680	0.15
38	686	0.142
39	692	0.135
40	697	0.129
41	702	0.122
42	707	0.116
43	712	0.11
44	716	0.105
45	720	0.099
46	724	0.094
47	728	0.09
48	732	0.085
49	735	0.081
50	738	0.077
51	742	0.073
52	744	0.069
53	747	0.066
54	750	0.063
55	752	0.06
56	755	0.057
57	757	0.054
58	759	0.051
59	761	0.048
60	763	0.046

Not there yet — keep moving the controls.

Product's offer wins, and it is not close. Cutting churn from five to three and a half raises the value of every subscriber by more than forty percent — including the 240,000 you already have, which growth's thirty percent does nothing for.

Retention beats acquisition in almost every subscription business, and this is the arithmetic reason. It is also why the retention budget ends up being spent on whichever retention lever is cheapest, which brings us to the next decision.

The cancel flow

What would you do

Around 4,200 subscribers a month reach the cancel page. Currently the page has one button and they are gone in nine seconds.

Your retention lead has four proposals, each with a modelled effect. You have to pick one, and you will be asked to defend it at a board meeting and possibly, eventually, elsewhere.

Optimising for: Maximise subscription revenue over the next twelve months without triggering a regulatory review or a measurable fall in new signups.
You cannot know: How many of the saved subscribers would have come back on their own within a year
What friction on the cancel page does to signups from people who noticed the reviews
Whether your jurisdiction is one of the several currently legislating on this
How much of the dormant fourteen percent would cancel if you emailed them

Leave it: one button, nine secondsOffer two months at half price on the cancel pageAdd a six-step flow: reason, survey, offer, confirm, reconfirm, email linkEmail the dormant fourteen percent and ask whether they still want it

Being fair to the mechanism

Automatic renewal is not a trick. Most subscribers want continuity, and a service that stopped every twelve months until you re-entered a card would be worse for almost everyone. The convenience is real, and it is the reason the model exists.

The problem is narrower and more specific: the same default that serves the person who wants continuity also collects from the person who has forgotten, and no business can distinguish them from the outside. What separates the honest operator from the other kind is not whether they use default renewal. It is whether they make the renewal visible and the exit symmetrical with the entrance.

Now turn it around

You have just spent ten minutes designing this from the other side. Here is what that is worth.

Check

Which of these actually reduces the amount collected from people who have stopped wanting a subscription?

Reading the terms carefully before subscribingA calendar entry on the renewal date, made at the moment of subscribingPaying annually rather than monthly, for the discountChoosing services with good reviews and a strong reputation

One thing worth doing with this lesson, away from the screen: list the recurring payments leaving your account, then mark the ones you have not used in sixty days. Almost everyone finds at least one. The interesting part is not the money. It is that you will remember signing up for it and will not remember any month since.

Who pays, who benefits, who carries the risk

Who pays. Every subscriber, and disproportionately the ones who stopped using the service — who pay full price for nothing and subsidise the price paid by everyone who uses it heavily.

Who benefits. The business, on the highest-margin revenue it has. And active subscribers, genuinely, because a service priced without the dormant revenue would cost more.

Who carries the risk. The subscriber, and the risk is not financial ruin — it is a small, recurring, invisible cost that no single month makes worth investigating. That is a design property, not an accident: the amount is set below the threshold at which anyone would take an afternoon to deal with it.

Worth remembering

Revenue from a subscriber who has stopped using the product is the highest-margin revenue a business has, because it costs nothing to deliver.
Cutting churn raises the value of every existing subscriber as well as every future one, which is why retention usually beats acquisition on the arithmetic.
Making a cancel flow harder improves the numbers for about a quarter and imports chargebacks, complaints and a delayed fall in signups.
Automatic renewal is genuinely convenient; the honest test is whether the exit is as easy as the entrance and whether the renewal is visible.
The defence against a monetised default is a date, not diligence: the mechanism relies on being busy rather than on being careless.