Subscription economics, honestly
Consumer subscription businesses are usually described in terms of acquisition. Channels, creative, cost per install, conversion to trial, conversion to paid. It is the part of the business that is visible, controllable and satisfying to optimise.
It is also the part that matters least to the outcome. Acquisition determines how many people enter. Retention determines what they are worth. One of those numbers compounds and the other does not.
Lifetime value is an assumption, not a measurement
Lifetime value is derived from a retention curve extended into the future. That extension is where most of the error lives.
Extrapolating from early data almost always flatters the business, because early cohorts are self-selected — the most motivated users arrive first. Assuming a constant monthly churn rate flatters it too, since churn is rarely constant; it is high early and lower later among survivors. Any model that treats lifetime value as a measured fact rather than a forecast with error bars will overstate what a customer is worth, and every downstream decision inherits that overstatement.
The practical discipline is to model cohorts separately, hold early ones to the same standard as mature ones, and be explicit about where the curve stops being data and starts being assumption.
The curve either flattens or it does not
The most informative question about a consumer subscription product is whether its retention curve flattens.
A curve that flattens means a group of users has found durable value and will keep paying. That group is an asset, and every acquisition cohort adds to it. A curve that decays steadily, however slowly, means the business is a leaking bucket and growth is purely a function of how fast the tap runs.
Both can look identical for two or three months, and both can post the same headline growth. Only one of them survives a rise in acquisition costs. This is the single number most worth being honest about internally, and the one most often reported selectively.
The first week decides the first year
In consumer software, most of the retention outcome is set before the user has any real history with the product. Whether they reached a first meaningful result, whether the product was configured well enough to be useful, whether the value was demonstrated before the first charge.
This makes onboarding a retention lever rather than a conversion lever, and the distinction matters. Onboarding optimised for conversion maximises the number of people who start paying. Onboarding optimised for retention maximises the number of people who are still paying six months later. The two frequently pull in opposite directions — friction removed to lift conversion is sometimes the same friction that produced an activated, committed user.
Involuntary churn is a product decision
A meaningful share of cancelled subscriptions are not decisions at all. Cards expire. Issuers decline renewals. Balances are temporarily insufficient. Users who intended to keep paying stop paying, and in many products nothing adequate happens next.
Recovering these is unglamorous work: intelligent retry schedules that respect issuer behaviour, card updater services, advance notice before expiry, clear and non-alarming recovery messaging, and a straightforward way to update payment details without navigating a support queue. The returns are immediate and rarely require touching the product itself.
Any business that has not separated voluntary from involuntary churn in its reporting does not yet know which problem it has.
Cancellation is a source of truth
Making cancellation difficult reliably improves this quarter's numbers. It also converts departing customers into disputes, negative reviews and payment-processor scrutiny, and it destroys the single most honest feedback channel in the business.
An easy cancellation flow that asks one optional question produces something valuable: a continuous, unfiltered account of why the product failed people. That data is more useful than most research, and the customers who leave cleanly are the ones who can return later.
The inverse is also true. Retention achieved through obstruction is not retention. It is deferred churn with added reputational and payment risk.
What is actually worth instrumenting
Four things, reported together and consistently.
Cohort retention by acquisition source, because channels do not deliver equivalent users and blended figures conceal that entirely. Time to first meaningful outcome, as the leading indicator of everything downstream. Voluntary and involuntary churn as separate series, since they have different causes and different fixes. And payback period measured against realised revenue rather than projected lifetime value — the version of the number that cannot be argued with.
The summary
Acquisition buys attention; the product decides what that attention is worth. Businesses that treat retention as a growth channel rather than a reporting metric end up with lower acquisition costs, more resilient economics and considerably fewer conversations with their payment providers.