How to design viral loops (and measure K-factor)

What a viral loop actually is

A viral loop is a repeating path in which using the product creates new users, who then use it and create more. Each turn of the loop produces the input for the next turn. A user sends a file, the recipient has to sign up to open it, the recipient now sends their own files — and the loop has gone around once. The defining feature is that the fuel is usage, not budget. A paid channel stops the moment you stop spending; a working loop keeps turning as long as people keep using the product.

That is the appeal and the trap. The appeal is self-sustaining growth that does not cost per new user. The trap is believing a loop can be added to any product, when in fact most products have no genuine reason for one user to bring another, and a loop without that reason does not turn.

How K-factor measures the loop

The health of a loop has a single number: the K-factor, or viral coefficient. It is the number of new users each existing user brings, on average, and it is the product of two things:

K = invites sent per user × share of invites that convert

An illustrative example makes the arithmetic concrete: if the average user sends four invites and one in four of those recipients signs up, K equals one. Send four and convert half, and K is two. These numbers are for illustration only, not benchmarks — your real K comes from your own data.

The threshold that matters is one:

• K above one — each user brings more than one new user, so the user base grows on its own. This is true self-sustaining virality, and it is rare.
• K below one — each user brings a fraction of another. The loop does not sustain growth alone, but it still lowers your effective acquisition cost on every other channel.

The other half of the loop: cycle time

K tells you whether the loop compounds. It does not tell you how fast. Two loops with the same K grow at wildly different rates if one turns in a day and the other in a month. Viral cycle time is the time it takes a new user to complete the loop and produce the next user — sign up, reach the sharing point, and bring someone in.

Cycle time is the exponent's clock. Take a hypothetical loop with K just above one: a version that turns in a week compounds far faster over a quarter than the same K turning monthly, because it goes around many more times in the same window. This is why shortening the loop often beats raising K. Moving the sharing point earlier in the first session — so a new user invites others before they have finished onboarding — can do more for growth than squeezing a few more points out of invite conversion.

So measure two things, not one. K says whether the loop is alive; cycle time says how quickly it breathes. A loop that compounds slowly is still worth speeding up, and the fastest wins usually come from removing delay between the steps, not from adding steps.

Why a sub-one loop is still worth building

A K below one reads like failure and is not. Suppose K is one half: every user you acquire, through any channel, brings half of another user for free. That halving-and-adding effect compounds across your paid and organic acquisition, so a channel that used to cost a certain amount per customer now delivers extra customers at no marginal cost. The loop is not sustaining growth by itself, but it is making every other channel go further.

This reframes the goal. Aiming only for K above one sets a bar most products never clear and dismisses loops that would meaningfully cut acquisition cost. The useful question is not "is K above one?" but "does the loop lower our cost to acquire a customer?" — and a loop with modest K almost always does.

K-factor	What it means	What it does for you
Above 1	Each user brings more than one	Growth compounds without spend
Below 1	Each user brings a fraction	Amplifies and cheapens every other channel
Near 0	The loop does not turn	The product has no real reason to be shared

Why most viral loops fail

Loops fail for one dominant reason: they are bolted onto products people have no genuine reason to share. A referral form added to the settings page does not create virality. Virality comes from the product being worth talking about, or from using it naturally exposing it to new people. When the underlying reason to share is absent, no incentive or invite prompt manufactures it — you are optimizing the conversion rate of a loop whose first step never happens.

This is the line between a viral loop and a referral program. A referral program is an incentive to invite — share for a reward. A viral loop is a path built into how the product works, where sharing is part of getting value: inviting a teammate to collaborate, sending a document someone must open, publishing a link others click. A program can boost a loop that already turns. It cannot create one where the product gives no reason to share.

So the first question is not how to design the loop but whether the product is shareable at all. If involving another person makes the product more valuable to the existing user — collaboration, communication, a network that improves as it grows — a loop is available. If not, the honest answer is to invest elsewhere rather than build a mechanism the product cannot support.

It helps to know which kind of loop your product can support, because they differ in durability. An inherent loop is built into core use — you cannot get the product's value without exposing it to others, as with anything collaborative or communicative. A word-of-mouth loop runs on the product being good enough that users tell peers unprompted. An incentivized loop pays users to invite. The three are not equal: inherent loops are the sturdiest because sharing is inseparable from value, word-of-mouth loops are strong but slower and harder to engineer, and incentivized loops are the weakest, because the moment the incentive stops the sharing stops. Reach for an incentive to accelerate a loop that already turns on its own — not to start one that otherwise would not move.

The fourth loop: collaboration and network effects

The three loops just named — inherent, word-of-mouth, incentivized — are the classic set, but there is a fourth that is the strongest of all, and it is the one you can most deliberately design for: the collaborative loop. Here inviting someone is not just how the product spreads but how it works better. One user starts; the value climbs with every colleague or contact who joins; so users recruit their own collaborators for their own benefit. Design tools, communication tools, and shared workspaces all live here. Figma is the standard example — a design file exists to be shared, and everyone who opens it has to enter Figma to view or edit, so ordinary use pulls new users in by design.

Laid side by side, the loops differ mostly in how durable they are:

Loop type	What drives it	Durability
Collaborative / network	The product is worth more as the user's contacts join, so users pull them in	Highest — each new user makes it better for the others
Inherent / product	Using the product exposes it — a shared file, a sent message, a published link	High — sharing is inseparable from value
Word-of-mouth	The product is good enough that users tell peers unprompted	Medium — real, but slow and hard to engineer
Incentivized / referral	Users are paid or rewarded to invite	Low — sharing stops when the reward stops

The collaborative loop is strongest because it coincides with a network effect: a loop brings new users, and a network effect makes the product more valuable as they arrive. When the same motion does both — the act that acquires also deepens the value — you get the sturdiest growth engine there is, because growing the base also raises retention and switching cost.

Why most "viral features" never go viral

Most features labeled "viral" — a share button, a "tell a friend" link, a badge on an email footer — never produce meaningful growth, and the reason is structural, not cosmetic. A share button is a single optional step, not a loop. For it to compound, the person who shares has to have a reason to, the recipient has to have a reason to sign up, and the new user has to reach their own sharing point — and each of those steps loses most of the people who reach it. Multiply three lossy steps together and K lands far below one; the feature "works" and moves nothing.

The features that do go viral are the ones where the loop is not optional. You cannot use the product without exposing it, or the product is worthless until you bring others in. That is why a share button bolted onto a solo tool stays quiet while a collaborative product spreads without one: in the first, sharing is a favor the user has no reason to do; in the second, sharing is how the product works. If you are adding a viral feature and can imagine the product being just as useful without it, it will almost certainly stay below the threshold.

The loop inside the organization: bottom-up and land-and-expand

In B2B, the most reliable viral loop runs inside a single company. One person adopts the product to solve their own problem, uses it in a way a colleague can see, and the colleague signs up — often on the same corporate email domain. The loop turns seat by seat, team by team, until the product is embedded across the organization. This is the bottom-up motion, and the expansion it produces is called land-and-expand: land one user or team, then grow the account from the inside.

It is a viral loop with a different unit. Instead of one user bringing a stranger, one user brings a coworker, and the "invite" is often just the product being visible in shared work. Trello grew this way — a freemium product that individuals adopted first and then spread across their teams, reaching millions of users before Atlassian acquired it. The companies that ran the same play, from Asana to Zoom, replaced the lead form with a single "sign up free" and let the product do the spreading.

Two roles make this loop turn, and both are worth naming:

• Champions — power users who expand usage inside their own organization. They invite teammates and pull the product deeper into the account.
• Advocates — users who recommend the product outside their organization, to peers at other companies. They start the loop somewhere new.

Design for both. A champion drives land-and-expand within one logo; an advocate starts a fresh loop in the next one. A product that produces neither has adoption but no engine.

The loop has to fit the product, not the other way around

Not every product can grow through a loop, and forcing one onto a product that cannot support it wastes the effort. This is product-channel fit: your growth model has to match what the product actually is. Viral and word-of-mouth growth fits products that are collaborative, shareable, or improve with a network — the same properties that make a loop possible in the first place. A product without them grows through other channels: content and search for products that solve searchable problems, sales-assisted motions for high-value products that need configuration, community for products with an ecosystem.

So the honest sequence is: first ask what kind of product you have, then pick the growth model that fits it, and only then design the loop — if a loop is even the right model. A team that decides "we need a viral loop" before asking whether the product can carry one is choosing the channel before the product, which is backwards. The loop follows from the product's shape; it cannot be imposed on it.

How to design and tune the loop

When the reason to share is real, design the loop as an explicit sequence of steps, then find its weakest link. Map it: the user acts, a new person is exposed, the new person signs up, the new person reaches the sharing point themselves. Estimate the conversion rate at each step from real data, and calculate K.

A loop is only as strong as its weakest step, so — as with any funnel — you fix the single worst leak first, re-measure, and repeat rather than scattering effort, exactly as inbound conversion lays out. Where the leak sits tells you what to fix: if exposure is high but sign-up is low, the problem is the landing experience for invited people, not the invite volume; if sign-up is high but new users never reach the sharing point, the loop draws people in and fails to bring them back around.

A loop multiplies retention — it never replaces it

A viral loop compounds the users you keep. It does nothing for the users you lose. If new users arrive through the loop and then churn before they reach the sharing point, you are pouring water into a leaking bucket faster — the bucket still leaks. Virality is a multiplier on retention, not a substitute for it, and a loop bolted onto a product with weak retention will flatter your signup chart while your active-user count stays flat.

This is why the loop and the habit have to be built together. The step that matters most is often not the invite — it is whether the new user comes back long enough to reach the point where they invite the next one. A loop with a leak at "new user returns and reaches the sharing point" fills the top and drains the middle.

Retention also feeds the loop through a second door: switching cost. In the Hook model, the Investment step is the moment a user puts something into the product — data, configuration, content, connections, invited teammates — that makes it more valuable to them and harder to leave. Every invited teammate is both a turn of the loop and a deposit that raises the cost of switching. The strongest loops are the ones where bringing others in is also what makes leaving hard: the same act grows the base, deepens the value, and locks in the user who performed it.

What the design produces

The deliverable is a loop design + K-factor: the loop drawn as steps, the conversion rate at each step, the resulting K, and the weakest step named as the next thing to fix. It turns "we should be more viral" into a specific mechanism with a measured number and an identified constraint.

The one part no calculation supplies is the reason to share, and it has to be real. AI can model every step, find the leak, and draft the invite copy — and that tuning is worth doing. But a loop bolted onto a product nobody wants to share does not go viral, however well tuned. Design the loop into the product, measure it honestly, and fix it one weakest step at a time.