Growth
Viral coefficient (k-factor)
The average number of new users each existing user successfully refers, measured per acquisition cycle. k > 1 means self-sustaining growth; k < 1 means the product needs paid or earned acquisition on top of referrals.
By Daniel Reyes · Last updated June 23, 2026
In plain English
Per 100 users you have, how many new users do they bring in through their use of the product (invites, shared links, viral mechanics)? Multiply how many people each user invites by how many of those invites convert. A k of 0.5 means every 100 users produces 50 new users via the loop; a k of 1.2 means 100 users produces 120 — and the next 120 produce 144, and so on. True virality is rare.
Example
Notion's free-tier-invite-someone-to-collaborate mechanic might produce: each active user invites ~0.8 collaborators per quarter, and ~30% of invitees convert to active users. k = 0.8 × 0.30 = 0.24. Below 1, but still meaningful — it lowers effective CAC by 24%. Dropbox's classic referral-for-storage drove k as high as ~0.6 at peak. Sustained k > 1 in mature products is extraordinarily rare.
Formula
k = (invites per user) × (conversion rate per invite) measured over a defined acquisition cycle (often per quarter).
Why it matters
Founders pitching 'we're going viral' typically have k between 0.05 and 0.3 — useful as a CAC reduction lever, not as a growth engine on its own. True viral products (k > 1) compound exponentially and look like the canonical case studies (Dropbox, Slack early days, WhatsApp). The realistic founder use of k-factor is: model it into your CAC math, identify which referral mechanics actually move it, and don't bet the company on virality unless you can demonstrate k > 0.5 with statistical confidence over multiple cohorts.
Common mistakes
- Averaging k across all users; the real signal is k for power users vs k for casual users — they often differ by 10x
- Counting invites sent rather than invites that convert — only the conversion side matters for the math
- Treating k as a constant; it almost always decays as the product saturates the natural-referral network
- Conflating k with NPS (sentiment) — high NPS doesn't automatically produce high k unless the product has a built-in referral mechanic