Audited —·Last updated 29 Jul 2026·3 citations·Tier 3·0 uses

Viral Coefficient Calculator — K-Factor and Growth Regime

Calculate the viral coefficient (k-factor) from invites and conversion rate, and see what it means below, at and above the k = 1 threshold.

Viral Coefficient Calculator

What do you want to solve for?

Invites sent per userAverage invites, shares or referrals each user in the cohort sends. Count invites actually sent, not invites the product offered.

Invite conversion rateShare of invites that become new active users. Cannot exceed 100% — one invite can bring at most one new user.

Target kUsed only in the two inverse modes. k = 1 is the self-sustaining knife edge; most successful products with referral loops sit well below it.

Starting cohortThe users you bring in yourself — by advertising, launch coverage or sales — before any invites are sent.

Invite cycles to modelOne cycle is the time from a user joining to their invitees joining — the viral cycle time. Six monthly cycles is half a year; six hourly cycles is an afternoon. The model has no clock, only cycles.

Viral coefficient (k)

0.72

Invites per user × invite conversion rate — the average number of new users each user brings. Read it with the regime note below: everything about how to interpret k depends on which side of 1 it falls.

What this k means

k is below 1 — subcritical. Each cohort produces a smaller one, so viral growth dies out on its own. It still amplifies paid acquisition: every user you bring in eventually brings 1 ÷ (1 − k) users in total, which is a real and permanent discount on your acquisition cost. k is not a property of the product. It moves with the audience, the incentive, the channel and how much of the addressable market already uses the product — exactly as the epidemiological reproduction number it is modelled on "is not a biological constant for a pathogen". Re-measure it per cohort rather than treating one number as fixed.

Invites per user used

Invite conversion rate used

9.00%

New users in the first cycle

720

Total users after the modelled cycles

3,213.1924

Of which came from invites

2,213.1924

Total users per user acquired

3.5714

Users at the converged total

3,571.4286

Background.

The viral coefficient — usually written k, and often called the k-factor — is the average number of new users each existing user brings in. It is the product of two things you can measure: how many invites, shares or referrals the average user sends, and what share of those invites turn into new users. Eight invites at a 9% conversion rate is a k of 0.72.

Be clear about what kind of number this is. "Viral coefficient" and "k-factor" have no standards-body definition. MASB's Universal Marketing Dictionary carries an entry for viral marketing — "a marketing phenomenon that facilitates and encourages people to pass along a marketing message, which can build until thousands of people react," from the AMA Dictionary — and gives no formula and no coefficient at all. The arithmetic on this page is practitioner convention, and the page says so rather than dressing it up.

What is properly sourced is the threshold, because k is borrowed wholesale from the epidemiological reproduction number. The CDC's Emerging Infectious Diseases states the rule plainly: "an outbreak is expected to continue if R0 has a value >1 and to end if R0 is <1." Heffernan, Smith and Wahl put the same threshold in the Journal of the Royal Society Interface: "When R0<1, each infected individual produces, on average, less than one new infected individual, and we therefore predict that the infection will be cleared from the population... If R0>1, the pathogen is able to invade the susceptible population." That is why this calculator treats k below 1, exactly 1, and above 1 as three different answers rather than three points on a scale.

Below 1 the loop dies out on its own — but it is still worth a great deal. Every user you acquire brings 1 ÷ (1 − k) users in total once their invitees, and their invitees' invitees, are counted. At k = 0.72 that multiple is 3.5714285714, which divides your effective acquisition cost by more than three. Most successful referral programmes live here, and describing that as "not viral" misses the point entirely.

Above 1 the model says growth compounds without limit, and no product sustains that. The reason is in the same source that gives the threshold: R0 "is not a biological constant for a pathogen" — it depends on the population and the setting. So does k. It falls as the addressable audience saturates, as the incentive fatigues, and as the invitable segment of your users is used up. Read a k above 1 as a description of one cohort at one moment, never as a forecast.

What is viral coefficient calculator?

The viral coefficient measures how many new users the average existing user produces through invitations, referrals or shares. It is calculated as invites sent per user multiplied by the share of those invites that convert, so a user sending 12 invites at an 11% conversion rate has a k of 1.32. The number matters mainly through its relationship with 1. Below 1 the process is subcritical: each generation of users is smaller than the one before it and the cumulative total converges on a finite multiple of the cohort you started with, 1 ÷ (1 − k). At exactly 1 it is critical: each generation reproduces itself, the total grows by the same amount every cycle, and growth is linear rather than exponential. Above 1 it is supercritical and the geometric series diverges — the model has no finite total, which is a statement about the model rather than a promise about the product. The threshold behaviour is inherited from the basic reproduction number in epidemiology, where the same condition determines whether an outbreak grows or dies out. The term itself is practitioner vocabulary, not a standardised metric: there is no governing body, no audited definition, and no agreement on the cycle length, which is why this page asks for cycles rather than months and states the assumption on the page. Note also that k says nothing about retention — a loop that brings in users who leave immediately can post a high k while the user base shrinks.

How to use this calculator.

Measure invites per user on a defined cohort over a defined window, counting invites actually sent rather than invites the product offered.
Measure the conversion rate of those invites into new active users, using your own definition of active and stating it.
Enter your starting cohort — the users you brought in yourself before any invites went out.
Set the number of invite cycles to model. A cycle is the time between a user joining and their invitees joining; the model has no clock, so a cycle is whatever your viral cycle time is.
Read the regime note before the numbers. It tells you whether the total below is a converged figure or just the point the model happened to reach.
If k is below 1, use the total-users-per-user-acquired output as a multiplier on your paid acquisition: it is the permanent discount the loop gives you on effective cost per user.
Use the inverse modes to test a plan. Asking what conversion rate you would need for k = 1 at your current invite rate is usually the fastest way to find out that the target is unreachable — the calculator refuses any target that would require a conversion rate above 100%.
Re-measure k per cohort. A single k applied to the whole user base averages together early adopters and a saturating audience, and it will overstate the future every time.

The formula.

k = i × c ; Total(n) = U₀ × (1 − k^(n+1)) ⁄ (1 − k) for k ≠ 1

i · invites sent per userc · invite conversion rate, as a fractionk · viral coefficientU₀ · starting cohortn · invite cycles modelledTotal(n) at k = 1 · U₀ × (n + 1)Converged multiple (k < 1) · 1 ⁄ (1 − k)

The coefficient itself is one multiplication: k equals invites per user times the invite conversion rate. Twelve invites at 11% is k = 1.32; eight invites at 9% is k = 0.72. Everything else follows from summing a geometric series. A starting cohort of U₀ produces U₀ × k users in the first cycle, U₀ × k² in the second, and so on, so the cumulative total after n cycles is U₀ × (1 − k^(n+1)) ÷ (1 − k). At the default figures — 1,000 users, k = 0.72, six cycles — 0.72⁷ = 0.10030613004288 and the total comes to exactly 3,213.192392704 users. Three cases have to be handled separately and the calculator does. Below 1 the series converges as the cycles increase, to U₀ ÷ (1 − k): at k = 0.72 that is a multiple of 3.5714285714, or 3,571.4285714286 users from the same 1,000. Exactly at 1 the formula above divides by zero and is replaced by U₀ × (n + 1) — pure linear growth, 7,000 users from 1,000 over six cycles. Above 1 the series diverges: there is no finite total to report, so rather than printing an infinity the calculator falls back to the multiple actually reached over the cycles you set and the regime note says explicitly that this is not a converged figure. On rounding stage: every intermediate is carried at full arbitrary-precision decimal width and rounding happens once, at the return boundary, to ten decimal places. k in particular is never rounded before being raised to the (n+1)th power, which at six cycles would move the total by several users per thousand. The k = 1 threshold is tested immediately below, at, and immediately above, including a continuity check that the linear branch does not introduce a jump: at k = 0.999999992 the six-cycle total is 6,999.999832 and at k = 1.000000008 it is 7,000.000168, either side of the exact 7,000.

A worked example.

Example

A product launches to 5,000 users acquired through advertising. Each of them sends an average of 12 invites, and 11% of those invites become new active users. The viral coefficient is 12 × 11% = 1.32. Above 1, so in the model this loop is supercritical and self-sustaining. The first cycle brings 5,000 × 1.32 = 6,600 new users — more than the cohort that produced them, which is exactly what k above 1 means. Run it five cycles and the cumulative total is 67,028.950016 users, of which 62,028.950016 came from invites rather than from the advertising budget. The multiple reached is 13.4057900032 times the starting cohort. That multiple is where the honesty has to come in. Below k = 1 the equivalent number is a converged total — a real ceiling the cohort approaches. Above 1 there is no ceiling in the model at all, so 13.4057900032 is not a steady state; it is simply where five cycles happen to land, and a sixth cycle would take it to about 18.7. The calculator labels it that way rather than printing an infinite total. The reason no product actually delivers this is the same reason the epidemiological reproduction number the metric borrows from is not a fixed quantity. As the CDC's Emerging Infectious Diseases puts it, R0 "is not a biological constant" and "will fluctuate if the rate of human–human or human–vector interactions varies over time or space." k behaves identically: the pool of uninvited people shrinks, the enthusiastic early cohort is not representative, and the incentive stops being novel. A k of 1.32 measured on a launch cohort is a genuine and impressive fact about that cohort. Treated as a growth forecast, it is the most reliably wrong number in a startup deck.

conversion Rate Percent11

invites Per User12

target K1

cycles5

initial Users5,000

solve Fork

Frequently asked questions.

What does a viral coefficient below 1 actually buy me?

A permanent discount on acquisition cost, which is worth far more than the word "not viral" suggests. If k is below 1 the loop dies out on its own, but every user you acquire still brings a total of 1 ÷ (1 − k) users once their invitees and their invitees' invitees are counted. At k = 0.72 that multiple is 3.5714285714, so a cohort of 1,000 paid users eventually becomes 3,571.4285714286 users and your effective cost per user is divided by more than three. Nearly every successful referral programme sits below 1. Chasing k above 1 as a goal usually means degrading the invite experience for a threshold that cannot be sustained anyway.

Why is k = 1 the threshold?

Because k is the average number of new users each user produces, so k = 1 is the point at which each generation exactly replaces itself. The metric borrows this directly from epidemiology, where the same condition governs whether an outbreak grows or fades. The CDC's Emerging Infectious Diseases states that "an outbreak is expected to continue if R0 has a value >1 and to end if R0 is <1," and Heffernan, Smith and Wahl put it in the Journal of the Royal Society Interface: "When R0<1, each infected individual produces, on average, less than one new infected individual, and we therefore predict that the infection will be cleared from the population." Mathematically it is the point at which a geometric series stops converging: below 1 the totals approach a finite ceiling, at 1 they grow linearly forever, above 1 they have no finite sum.

Is a k above 1 sustainable?

No, and any model that assumes it is will be wrong. The same source that supplies the threshold also says why: R0 "is not a biological constant for a pathogen" and "will fluctuate if the rate of human–human or human–vector interactions varies over time or space." Everything in that sentence applies to k. The addressable audience is finite, so as the loop runs, an ever-larger share of invitations goes to people who already use the product or will never use it, and the conversion rate falls. Early cohorts are also unrepresentative — enthusiasts invite more and convert better than the mainstream that follows. This calculator will happily model a k of 1.32 for you, and it labels the result as the multiple reached over the cycles you set rather than as a converged total, precisely because there is no converged total to report.

What counts as one cycle?

The time between a user joining and their invitees joining — the viral cycle time. The model has no clock in it: it counts generations, not days, so six cycles is half a year for a product whose loop turns monthly and an afternoon for one whose loop turns hourly. Cycle time matters enormously in practice because it sets how fast the same k compounds, and it is the variable most within a product team's control: halving the time to first invite doubles the number of generations in a quarter without touching k at all. Because there is no standard definition here, state your cycle length whenever you quote a modelled total.

Does a high viral coefficient mean the product is growing?

Not on its own, because k says nothing about retention. It counts users who arrive, not users who stay. A loop that invites aggressively and converts well can post a healthy k while the total user base shrinks, if the users it brings in churn faster than they are replaced. The two metrics have to be read together: k tells you how efficiently the base reproduces itself, and churn tells you how fast it decays. The other thing k does not capture is quality — invited users are frequently worse-retaining and lower-spending than users who arrive under their own steam, so a rising k that comes from a stronger incentive can coincide with falling revenue per user.

References& sources.

Related calculators

Engagement Rate Calculatormar Share of Voice Calculatormar Customer Lifetime Value Calculatorsaa ROI Calculatorinv CAGR Calculatorinv

In this category

Embed

Quanta Pro

Paid features are coming later.

All 762 calculators remain free
No billing is enabled

Coming soon