Viral Coefficient Calculator

The maths behind it

The viral coefficient is invitations sent per customer × conversion rate of those invitations: k = invites × conversion.

A k above 1 means each customer produces more than one new customer, and the base grows without any acquisition spend. Below 1 it amplifies other channels without sustaining growth alone.

Cycle time, how long a referral loop takes, matters as much as k, because a k of 1.2 on a six-month cycle grows far more slowly than 1.05 on a weekly one.

Putting numbers to it

Each customer invites 3.2 people on average, and 14% of invitations convert: k = 3.2 × 0.14 = 0.45.

Below 1, so referral does not sustain growth by itself. What it does is multiply paid acquisition: 1,000 paid customers produce 450 referred, who produce 202, then 91: a total of roughly 1,818 from 1,000 paid.

Effective CAC becomes $27,000 ÷ 1,818 = $14.85 rather than $27, which is a 45% improvement in acquisition economics from a coefficient well under 1.

That amplification is where nearly all real-world value from referral sits, and it does not require anything approaching virality.

Where it is unreliable

Sustained k above 1 is very rare outside products where sharing is intrinsic to using them. Businesses planning growth on the assumption of achieving it are planning on an outlier.

The coefficient also decays as the addressable network saturates: early customers invite people who are easy to convert, and later ones invite people who have already declined.

How to act on this

Model the amplification rather than chasing k above 1. Improving k from 0.3 to 0.5 changes effective CAC by roughly 30%, which is a large and achievable result.

Then work on the two terms separately. Invitations per customer is a product and prompt problem; conversion of invitations is a landing and offer problem, and they respond to different work.

Why cycle time is the underrated term

Two products with the same coefficient can grow at wildly different rates depending on how long a loop takes. A referral that happens at the moment of purchase compounds far faster than one prompted in a quarterly email.

Shortening the loop usually means moving the referral prompt closer to the moment of satisfaction, immediately after delivery or first successful use rather than weeks later.

That timing change frequently does more for total referred customers than raising either component of k, and it is easier to implement than either. Businesses that treat referral as a campaign rather than a moment in the customer journey generally have a long cycle time without realising it is the constraint.

Measuring k by cohort shows whether it is decaying as the network saturates, which is the usual pattern and the reason early results overstate what a referral programme will sustain.

Where k is measured only in aggregate, a strong early period can mask a coefficient that has already fallen well below the level the growth model assumes.

Making the shared thing genuinely useful to the recipient, rather than an invitation to join something; is what separates referral mechanics that work from ones that are ignored.