Glossary
What is K-factor?
K-factor is a viral growth metric measuring how many additional users each existing user brings in — calculated as invitations sent per user multiplied by the conversion rate of those invitations.
The calculation
K = invitations sent per user × conversion rate per invitation. A user who sends 5 invitations that convert at 20% has a K-factor of 1.
The threshold everyone quotes is K > 1: each user brings more than one new user, so growth compounds without acquisition spend. Below 1, virality amplifies other channels rather than replacing them — which is still valuable and is a very different business.
Why sustained K > 1 is rare
Because K is not a constant. It decays as a product saturates its natural network: early users invite the people most likely to want the product, and later users have fewer good candidates left to invite.
A product measuring K > 1 in its first months is usually measuring the enthusiasm of early adopters rather than a durable property. Planning a growth model on an early K figure is one of the more reliable ways to miss a forecast.
The other reason is time. K says nothing about cycle length, and a K of 1.2 with a 30-day cycle behaves completely differently from the same K with a one-day cycle.
Reading it honestly
- Measure by cohort, since K decays with saturation and a blended figure hides it.
- Include the cycle time. K without a period is not a growth rate.
- Separate prompted from spontaneous sharing. A referral programme buying invitations is not the same as organic virality.
- Discount for attribution over-claim, since some referred users would have arrived anyway.
Common questions
Is K-factor still a useful metric?
As a diagnostic of whether sharing is working, yes. As a growth forecast, rarely — it decays with saturation and ignores cycle time, so models built on it tend to be optimistic.
Can paid acquisition raise K-factor?
Indirectly, by enlarging the base that does the inviting. It does not change the underlying rate, and buying users to raise a viral coefficient is a fairly reliable sign the coefficient is not real.
More in growth and engagement
Acquiring users, and the behaviour that says whether they stayed.