Why Retention, Not Acquisition, Is the Real Growth Metric
Emrah · Facebook's former VP of Growth argues that most startups optimize the wrong side of the growth equation first. Here's what founders and investors should each take from his framework.
The Story
In 2007, Alex Schultz joined Facebook's growth team, then still a small function inside a company with a fraction of its eventual user base. Over the following years, that team helped take Facebook from around 10 million users to more than 2 billion, and Schultz became one of the people most associated with turning growth into a rigorous, data-driven discipline rather than a set of marketing tricks.
In 2014, Schultz taught a lecture on growth as part of Stanford's CS183B, "How to Start a Startup," a course run jointly with Y Combinator. The lecture has since become one of the more widely referenced primary sources on startup growth, not because it introduces flashy tactics, but because it does the opposite: it argues that most of what founders think of as growth work comes far too early.
His central claim is simple to state and harder to accept. Before a team touches acquisition channels, paid ads, virality, or SEO, it needs to answer one question honestly: does the retention curve flatten, or does it keep sliding toward zero.
The Idea
Schultz defines growth in a deliberately unglamorous way: using whatever channel is available to get whatever outcome you want. There's nothing mystical in that definition, and that's the point. The mystique startups often attach to growth, the idea that some clever viral loop or growth hack will unlock a hockey stick, is exactly what Schultz spends the lecture pushing back against.
Instead, he puts retention at the center. A retention curve tracks the percentage of users still active some number of days after they first signed up. Plot it for a healthy product and, after some initial drop-off, it flattens into something close to a horizontal line. That flat line represents a durable core of users who keep coming back. Plot it for an unhealthy product and it keeps declining toward zero, meaning every cohort eventually churns out entirely.
The retention curve matters because it determines whether growth compounds or leaks. If the curve flattens, every new cohort adds to a growing, durable base, and acquisition spending pays for itself over time. If it doesn't flatten, a company is effectively refilling a bucket with a hole in the bottom. It can spend more on acquisition and post encouraging total-user charts for a while, but the underlying business is not actually getting stronger.
To make retention actionable, Schultz introduces the idea of a "magic moment," a specific point early in a user's experience where they first feel the product's core value. For Facebook, that moment was seeing your friends on the platform. For eBay, it was finding the specific item you'd been looking for. For Airbnb, it was discovering a place you actually wanted to stay. For WhatsApp, it was sending that first message. In each case, the magic moment isn't a vague feeling, it's something specific enough to be observed in user behavior data.
Facebook's growth team went further and quantified it: a new user who added 10 friends within 14 days was dramatically more likely to stay active long-term. That number wasn't guessed at or borrowed from another company. It came from studying which early behaviors correlated most strongly with long-term retention across cohorts. Once found, it became the organizing target for the entire growth function, long before the team optimized a single acquisition channel.
Schultz also argues for a single North Star Metric that cascades through an organization: Facebook used Monthly Active Users, WhatsApp used messages sent, Airbnb used nights booked, eBay used gross merchandise volume. The metric has to be simple enough that every team, from product to marketing, can see how their work moves it.
How Founders Should Read This
For a founder, the practical takeaway is sequencing. Before hiring a growth team or running acquisition experiments, define and measure your product's magic moment. This isn't a branding exercise or a slogan for the pitch deck. It's a specific, observable user action, found by looking at cohort data and asking which early behaviors separate users who stay from users who churn.
The trap most founders fall into is treating the magic moment as something they intuit rather than something they measure. A founder might say their product's magic moment is "when the user sees the value," which sounds right but isn't specific enough to act on. Facebook's team didn't stop at "seeing friends," they pushed until they had a number: 10 friends, 14 days. That specificity is what made the moment usable as an operating target rather than a talking point.
There's a second, related trap: building a growth team before there's anything durable to grow. Schultz's framework implies that a growth team hired to fix a leaking bucket will mostly generate expensive noise, better-looking top-line charts that mask a business that isn't actually getting healthier underneath. The better sequence is to nail retention first, even if that means growth headcount waits.
A concrete exercise worth running today: pull up your own product's cohort data, or a portfolio company's, and try to name the specific early action, not a vague moment, but an action with a number attached, that separates users who stick around from users who don't. If you can't name one, that's the actual work, before any acquisition channel gets touched.
How Investors Should Read This
For an investor, Schultz's framework converts naturally into a diligence lens. The question to ask isn't "how many users do you have," it's "show me your cohort retention curve, broken out by signup month, and tell me where it flattens."
Total user counts are a weak signal on their own because they almost always trend upward, even for products with serious underlying leakage, as long as acquisition spending continues. A rising top-line number can mask a retention curve that never flattens, meaning the company is running hard just to stay in place. The cohort view strips that illusion away: each cohort's curve either flattens into a durable base or doesn't, and a founder who can walk through that chart from memory, including where and why it flattens, is showing real command of their own business.
This matters directly for fund economics. Every dollar of acquisition spend only compounds into fund returns if the users it brings in stick around long enough to be monetized well past the payback period. A company with weak retention that raises a large round to "invest in growth" is often just accelerating the rate at which capital leaks out, not fixing the underlying business. Schultz's North Star Metric concept gives investors a second useful check: does the company have one metric the whole organization rallies around, and can every function articulate how their work moves it? A scattered set of vanity metrics, with no single number the team agrees matters most, is its own kind of red flag.
The practical diligence question, then, is less about market size or team pedigree in this specific dimension, and more mechanical: ask for the cohort retention curve before the pitch deck's growth slide, and read the flattening point as closely as the total-user chart.
Where the Two Views Collide
The tension between these two lenses is sharper than it first appears. A founder's instinct is often to build a growth team early, because growth work is visible, produces fast feedback, and is easy to narrate to a board or an investor as "traction." Waiting to nail retention before touching acquisition can feel, from inside a company, like moving too slowly while competitors are visibly running ads and posting user-growth numbers.
An investor, applying Schultz's lens, sees the same instinct differently. Acquisition spend against a retention curve that hasn't flattened isn't traction, it's noise that will eventually need to be walked back, usually at a worse valuation and with less founder credibility than if the sequencing had been right from the start.
Both readings can be correct depending on the specific situation. A founder is right to push for early acquisition investment if the retention curve has genuinely flattened but the team simply hasn't scaled distribution yet, since every day of delay there is a day of compounding growth left on the table. An investor is right to push back hard when a founder says something close to "let's grow first, we'll fix retention later," because in Schultz's experience that sequencing rarely works: it's much harder to fix a broken product after scaling acquisition around it than to fix it while the user base is still small.
The person who understands this distinction has a real edge at the table. Instead of accepting a rising total-user chart at face value, they ask for the cohort breakdown. Instead of accepting "we know our magic moment" as a claim, they ask for the number behind it. The chart everyone shows is the one that always goes up. The chart that actually matters is the one underneath it.
The Takeaway
Schultz's framework reduces to an uncomfortable but useful discipline: find the specific, measurable action that predicts whether a user sticks around, get the retention curve to flatten around it, and only then turn on the acquisition machine in earnest. Total user growth is easy to produce and easy to fake the health of. A flattening cohort retention curve is much harder to fake, and it's the one number, more than any other, that tells you whether a business is actually compounding or just running in place.
The applicable framework for your own week: pick one product you're close to, whether it's your own or one you're evaluating, and ask for its cohort retention curve broken out by signup date. If you can't get an answer more specific than "users seem happy," that gap is the most useful thing you'll learn all week.
Source: Alex Schultz, "Growth," How to Start a Startup (Stanford CS183B / Y Combinator), 2014 — watch the lecture
