Retention Is the Business: Reading the Cohort Curve

Every other lever eventually drains into this one. Demand, acquisition and referral all assume that a user who arrives will stay long enough to matter, and a consumer business is, in the end, the shape of its retention curve. If that curve flattens, you have a business that compounds. If it decays toward zero, you are refilling a leaky bucket forever, and no amount of acquisition will outrun the drain.

Why retention sits above everything

Retention is the input that quietly sets the other numbers. It is the bulk of lifetime value, since a customer who stays twice as long is worth roughly twice as much; it is what makes acquisition affordable, because a higher LTV justifies a higher cost to win; and it is the fuel for referral, since people only recommend what they kept using. The economics have been measured for decades: Bain & Company's classic finding in the Harvard Business Review is that lifting retention by just 5% can raise profits by 25% to 95%. Improving retention does not add to growth; it multiplies it.

How to read a cohort curve

A cohort is a group that started together, say everyone who signed up in one week, tracked over time. The retention curve plots what fraction of that cohort is still active week by week. It always falls at first, as the merely curious drop away. What matters is what happens next: does it keep falling to zero, or does it bend and settle on a plateau. That plateau is the share of people for whom the product became a habit, and the height of it, not the steepness of the early drop, decides whether you have a business.

The flattening curve is the real signal

Among growth practitioners, the flattening of the retention curve is treated as the clearest evidence of product-market fit, a point Andrew Chen and the Reforge community have made for years. The instructive part is that the plateau does not need to be high, it needs to exist. A product that settles at forty-something percent of users active months later has found a core that genuinely needs it, and that core is what every other lever then compounds on. What a healthy floor looks like varies enormously by category, daily for a messaging app, monthly for a travel product, so the comparison that matters is against your own earlier cohorts, not a universal benchmark. In a smaller home market such as Taiwan, where the addressable user base is finite and paid acquisition saturates quickly, the plateau matters even more: there is less new demand to refill a leaky bucket with, so a curve that flattens is what lets a business compound without an endless acquisition spend.

How retention gets faked

The most common trick is not a lie but a flattering definition. Counting a login, an app open or any trivial action as 'active' inflates retention with motions that carry no value, so a curve can look healthy while the behaviour that pays the bills quietly erodes. The other classic is hiding behind aggregate monthly active users: a total that grows because acquisition outruns churn can mask a leaky curve underneath, since new arrivals refill what departing users empty. A rising MAU with a falling cohort curve is a company spending more and more to stand still. The honest read is always per cohort, on an action that means the product did its job.

How we read it

Reading a consumer deal, the retention curve is the first thing we ask for, by cohort, on a meaningful action, over as long a window as the company has data. We look for the bend: does the curve flatten, and where. We check whether recent cohorts retain better than older ones, which says the product is improving, or worse, which is a quiet alarm no headline growth number will sound. And we treat a flattening curve as the strongest evidence that the demand in what consumers actually buy is real, because retention is desire, proven over time. It is lever three of the four levers, and the one we trust most.

The takeaway

Before you spend the next dollar on growth, find out whether your curve flattens. If it does, acquisition is an investment and you should make it boldly. If it does not, growth is just a faster way to spend money teaching strangers that they do not need you yet. Fix the curve first; everything else in a consumer business is built on its plateau.