Growth marketer operator · San Jose, CA | ZuAI: 10K → 2M users at $0.02 CAC | $300k/mo ad spend managed
Ar.Bhavesh Pro
Product-led growth & viral loops

K-factor math that isn't fantasy

the real k-factor formula, why k above 1 is almost mythical, and how sub-1 k still discounts your paid CAC if you measure honestly.

last updated 17 November 2026

k-factor is invites or shares sent per user, multiplied by the rate at which each one becomes a new active user. k above 1 means the product grows itself without spend, and almost no consumer app holds that for more than a spike. the useful version is sub-1 k, which quietly discounts your paid CAC, provided you measure both inputs honestly.

The real formula

k = i times c, where i is the average number of invites or shares a user sends in a defined window, and c is the fraction of those that convert into a new activated user. both definitions carry weight. i counts sends that actually left the app, not taps on the share button. c counts activated users, not installs. and the window is part of the number: a k measured over a cohort’s first month is a different claim than a k measured over a launch week, and a k quoted with no window at all is fiction. when k is below 1, one seeded user eventually brings 1 divided by (1 minus k) total users, which is the whole reason sub-1 k is worth anything.

Why k above 1 is almost mythical

k above 1 means every user replaces themselves and then some, forever, which is an exponential, and consumer attention does not hand out exponentials for long. the products that touch it do so in bursts: a viral moment pushes shares and conversion up together, k crosses 1 for days or weeks, the novelty decays, and k settles back under. the trap is measuring during the burst and planning the year on it. launch-week k includes your most excited users sharing to their most receptive friends, a pool that depletes by definition. treat any k above 1 in your dashboard as a timestamp, not a property of the product, and re-measure on a boring cohort in a boring month before you believe anything.

Sub-1 k still compounds paid

the honest prize is arithmetic, not magic. if k is stable below 1, every paid user brings a tail of free ones, and your effective CAC is the paid CAC multiplied by (1 minus k). run the hypothetical with your own numbers: a $2 paid CAC with an honestly measured k of 0.3 is $1.40 per user once the tail arrives. that is a 30% discount on every channel at once, earned by product work, and it compounds with every improvement to the loop.

this is also the honest way to read blended numbers like the ZuAI case. ZuAI went 10K to 2M users at a $0.02 blended CAC, and share loops were a supporting mechanic in that mix while paid and UGC did the volume: the loops made every acquired user bring a little more than themselves, and the blend across everything is where the $0.02 lives. the arithmetic behind that number is its own page, and no single loop or channel produced it alone.

Measure both inputs honestly

the two dishonest habits are counting intent as sends and installs as users. for i, count shares that left the device, deep links created and delivered, and accept that screenshots and dark shares will undercount you; undercounting is survivable, inflating is not. for c, tie referred installs to activation, the friend reaching first real value, through referral codes and deep link attribution, and let unattributable installs stay unattributed rather than guessing them into the numerator. then hold the window fixed and re-measure monthly, because a k that only exists in your best month is a number you will spend real money believing.

What is a good k-factor for a consumer AI app?

there is no published number worth trusting, and anyone quoting one is selling something. any stable k above zero is a real discount on your CAC; measure yours, then work the inputs.

Is k-factor the same as viral coefficient?

yes, same formula, two names. the variation between sources is in the window and in what counts as a converted user, which is exactly why you define both when you quote yours.

Should I count shared outputs in k or only invites?

count both, tagged separately. an invite and a shared output are different loops with different conversion rates, and blending them hides which one deserves the next sprint.

k is the measurement layer of the product-led growth playbook: it does not create the loop, it tells you the truth about the one you have. want your k computed from real data instead of hope? the breakdown call is free. Let’s talk Growth.

this page is part of the Product-led growth & viral loops playbook.

next

More in this playbook

Product-led growth & viral loops

Why nobody used your referral program

the post-mortem of the bolted-on referral tab: missing motive, wrong timing, friction, rewards nobody wants, and what to fix before rebuilding.

Product-led growth & viral loops

Shareable outputs: the AI-native viral loop

how to design outputs people want to post, make export native per platform, and why share rate times conversion beats invite programs.

Product-led growth & viral loops

Watermarks: growth vs user experience

a watermark converts viewers of shared outputs, but too loud and sharing stops. the subtle-but-legible rules and when to sell removal.

Product-led growth & viral loops

When to add a referral program: pre or post PMF

add referral after retention flattens, not after launch. the readiness checklist, and why referral amplifies love it cannot create.

Product-led growth & viral loops

Referral incentives that work for freemium AI apps

credits, generation limits, early features and cosmetics: rewards a freemium AI app can afford, with accounting too simple to farm.

Product-led growth & viral loops

Apple's rules and rejected referral programs

the reward mechanics Apple rejects under guideline 3.2.2, why referral rewards trip review, and designing incentives that pass.

Product-led growth & viral loops

Onboarding as a growth surface

time to first output is the retention lever. how removing onboarding steps compounds every channel, and why activation beats signups.

Product-led growth & viral loops

Retention before referral: the order of operations

loops multiply retention, they never replace it. the honest sequence of activation, habit, then loops, and how to read your curve.

Want this run on your app?

the playbook is free and yours. if you would rather it was executed than read, tell me what you are working on.

answers go to a private google sheet and to my inbox, used only to reply and follow up. the newsletter is opt in and separate.

Want this run on your app?

the breakdown call is free. bring one real growth problem, leave with the fix.

Let's talk Growth