Why this matters
Viral growth is the loop everyone wants and few measure correctly. Teams say “we are going viral” when a post gets shared, or “virality does not work for us” when K is 0.2. Both statements skip the math.
The math is simple and it changes decisions. A viral coefficient of 0.3 sounds weak. But it means every user you buy brings 0.43 extra users over time, for free. That cuts your effective CAC by 30%. Few ad optimisations deliver that.
Dropbox is the classic example. Drew Houston’s 2010 slides show that paid search cost 233 to 388 dollars per customer for a 99 dollar product. A two-sided referral program then lifted signups by 60%, permanently. That is a change in unit economics, not a marketing campaign.
This lesson gives you the exact formulas, a worked example of the amplification math, and a playbook for incentives that raise K without poisoning your user base.
The framework
The viral loop
Viral referral loop
Each step feeds the next — step 4 (“Friend signs up and activates”) loops back into step 1 (“New user gets value”). Tap a step to trace it.
The core formulas
If 1,000 new users send 2,000 invites (i = 2) and 300 of those become users (c = 15%), K = 2 × 0.15 = 0.3.
| K | Amplification 1 ÷ (1 − K) | Effective CAC if paid CAC is $50 |
|---|---|---|
| 0.1 | 1.11 | $45.00 |
| 0.2 | 1.25 | $40.00 |
| 0.3 | 1.43 | $35.00 |
| 0.5 | 2.00 | $25.00 |
| 0.7 | 3.33 | $15.00 |
| 0.9 | 10.00 | $5.00 |
Notice the shape. Each step from 0.5 to 0.7 to 0.9 is worth far more than the step before it. Late gains in K are worth more than early ones.
Cycle time
K tells you how many users you eventually get. Cycle time tells you when. Cycle time ct is the median days from a user signing up to their invitees signing up.
When K is above 1, the series does not converge. Each generation is bigger than the last until you run out of reachable people. That is real exponential growth. It is rare, and it usually fades as the easiest networks saturate and novelty drops. Plan for K below 1 and treat K above 1 as a temporary bonus.
What sits inside i and c
Break each term down so you know what to work on.
- i, invites per user = share of users who invite at all × invites per inviter. The first term is usually the bigger lever. Most users never invite.
- c, conversion per invite = invite delivered × opened × clicked × signed up × activated (if you only count activated users, which you should).
Step-by-step playbook
-
Define the counted user. Count an invitee only when they reach your activation event (first project created, first order delivered). Raw signups inflate K and hide low-quality invites.
-
Fix the windows. Use invites sent within 30 days of the inviter’s signup, and invitee signups within 14 days of the invite. Report K per weekly cohort. Without windows, K drifts every time you look at it.
-
Place the ask where another person is needed. The highest-converting invites come from product use: sharing a file, assigning a task, splitting a bill, sending a gift. Zoom-style and Dropbox-style shared-folder loops beat “invite friends” pages because the invitee gets something immediately.
-
Ask after success, not at signup. Prompt right after the user’s first win (first report shared, first order delivered, first payout). Satisfaction is highest then, and the user has a story to tell.
-
Design the incentive. Use these rules:
- Double-sided. Give the invitee something too. It raises c because the invite feels like a gift, not a sales pitch. Dropbox gave both sides extra storage.
- Product-native. Storage, credits, free months, free shipping. They cost you less than face value and attract people who will use the product.
- Pay on value. Release the reward when the invitee activates or the first order ships, not on signup.
- Cap and screen. Cap rewards per referrer per month. Check device, payment card, address and email patterns for self-referral.
- Tiered for power users. Milestone rewards at 3, 5 and 10 referrals keep your best advocates going.
-
Shrink cycle time. Pre-fill invite messages. Support WhatsApp and SMS sharing, not just email (essential in India). Send one reminder to the invitee after 48 hours. Make the invitee landing page show who invited them and what they get.
-
Test one term at a time. Run A/B tests that target i or c separately so you know which moved. Check sample size before calling a winner.
-
Watch invitee quality. Compare 90-day retention and revenue of referred users with other channels. If referred users churn faster, the incentive is attracting deal-seekers.
Worked example
Say a B2B scheduling tool for clinics in India buys 10,000 new users a quarter through paid at a CAC of ₹1,200. Paid spend is ₹1.2 crore (₹12,000,000).
Current viral numbers from last quarter’s cohort:
- 25% of new users invite a colleague within 30 days.
- Inviters send 4 invites on average. So i = 0.25 × 4 = 1.0.
- 20% of invites lead to an activated user within 14 days. So c = 0.20.
- K = 1.0 × 0.20 = 0.20.
Amplification. 1 ÷ (1 − 0.2) = 1.25. Total users = 10,000 × 1.25 = 12,500. Effective CAC = ₹12,000,000 ÷ 12,500 = ₹960.
Generation by generation. 10,000 then 2,000 then 400 then 80. After three cycles, 12,480 users, which is 99.8% of the eventual total. With a median cycle of 12 days, nearly all the viral value lands within about five weeks.
Now the team ships two changes: an invite prompt right after the first appointment is booked, and a double-sided reward of one free month for both clinics. Results:
- Share inviting rises to 40%, invites per inviter stays 4. i = 1.6.
- Conversion rises to 25% because invitees also get a free month. c = 0.25.
- K = 1.6 × 0.25 = 0.40.
New amplification. 1 ÷ (1 − 0.4) = 1.67. Total users = 10,000 × 1.67, about 16,667. Effective CAC = ₹12,000,000 ÷ 16,667, about ₹720.
Cost of the incentive. About 6,667 users came from referrals. Each referral rewards two accounts with a free month. Say a month of service costs you ₹150 to serve at gross margin. That is roughly 6,667 × 2 × ₹150 = ₹2,000,100, about ₹20 lakh. Total cost becomes ₹1.4 crore for 16,667 users, about ₹840 per user. That is still 30% below the paid-only CAC of ₹1,200, and the extra users arrive inside a quarter.
The point: moving K from 0.2 to 0.4 is worth more than most paid-media optimisations, and you can price it exactly.
B2B SaaS vs D2C: what changes
| Topic | B2B SaaS | D2C / e-commerce |
|---|---|---|
| Natural viral action | Inviting teammates, sharing work with clients, external guests, “powered by” footers | Gifting, sharing a code, posting the product, group orders |
| Who receives the invite | Colleagues and clients in the same workflow, high intent | Friends and family, broader and lower intent |
| Best incentive | Free months, extra seats, usage credits, feature unlocks | Store credit, free product, free shipping, early access to drops |
| Qualifying action for reward | Invitee activates or becomes a paid seat | Friend’s first order ships, past the return window if possible |
| Typical cycle time | Days to weeks inside a team; longer across companies | Days, driven by social sharing and gifting moments |
| Main risk | Invites to people with no authority to buy | Coupon abuse, self-referral, discount-only buyers |
How to measure it
- Invites per user (i). Invites sent within 30 days of signup ÷ users in the signup cohort. Also report share of users who invite at least once.
- Invite conversion (c). Activated invitees within 14 days of invite ÷ invites sent. Also report each sub-step: delivered, opened, clicked, signed up.
- K. i × c for each weekly cohort. Track a 4-week rolling median for decisions.
- Cycle time. Median days from inviter signup to invitee activation.
- Amplification and effective CAC. 1 ÷ (1 − K), and paid spend ÷ (paid users × amplification). Report both in the monthly growth review.
- Referred-user quality. 30, 60 and 90-day retention and revenue of referred users compared to paid and organic.
- Segment. By platform, country, plan, and acquisition source of the inviter. Users from organic often invite more than users from paid.
- Fraud rate. Rewards reversed ÷ rewards issued, per month.
Templates
Referral incentive checklist:
- Reward for both sides, in product value or store credit
- Paid on activation or first shipped order, not signup
- Cap per referrer per month
- Device, card, address and email checks for self-referral
- Invite placed at a natural “need another person” moment
- WhatsApp, SMS and copy-link options, with pre-filled message
- Invitee landing page names the referrer and the reward
- One reminder to invitee after 48 hours
- Referral source tagged so K can be measured per cohort