PLG Funnel Model

A self-serve product where signups are an output rather than an input. Most SaaS funnel models take a monthly signup number as an assumption and convert it downward through fixed stage rates; this one refuses to, because the largest single source of new signups is the product's own retained base. Habitual users send invites, a share are accepted, and every accepted invite re-enters the top of the same funnel it came out of. The whole archetype turns on one line: the inviting base is last month's closing habitual users, so the dependency runs forward in time only and a genuinely self-referencing growth engine is modelled with no circular reference anywhere in the workbook and no iterative calculation setting. Reaching back one column instead of sideways within the same column is the entire trick, and it is also how the business really works: a user who signs up on the fifteenth is not inviting anybody that month. Only habitual users invite, which is deliberate - sending the loop off the signup cohort would let a burst of paid traffic manufacture viral growth it has not earned. The coefficient is then built as a lifetime measure rather than a monthly one, in three auditable steps: a monthly loop rate of 3.4 per cent, being invites per user times acceptance times the signup-to-habitual rate, carried across an average lifetime of 26.3 months to give k of 0.88 and an amplification factor of 8.5 times. A stability check flags the point where k crosses one and the amplification formula would invert. Across the window k decays to 0.84 as invite rates mature, while the loop's share of signups rises from 36.9 to 40.5 per cent - not a contradiction, but the divergence a reader who conflates the coefficient with what the loop delivers will misread. Three paid channels are carried separately, because the cheapest visitor at $1.35 converts at less than half the rate of the most expensive at $3.10. The loop is not free: every accepted invite carries a $12 referral credit that reaches 165 a month, a third of the entire sales and marketing line, and the loop is valued net of it - 368 a month of signups that would otherwise be bought, less 165 to run, netting 203. Three acquisition costs sit side by side rather than blended: a paid signup at $26.76, the blend at $14.49, and the $12.27 gap the loop is worth per signup. On the shipped case signups run from 14,642 to 33,978 a month, the habitual base compounds from 46,000 to 154,715 with a dormant pool of 67,391 building behind it, and paid users reach 46,266 at 31.4 per cent penetration. Recurring revenue grows from 199 to 1,762 a month, or 21,149 of ARR, gross margin improves from 64.6 to 83.9 per cent, EBITDA crosses into profit in month 22 and reaches 196, and cash troughs at 11,923 rather than running out. The unit economics improve because the loop does and not because the paid channels do: paid cost per signup is flat at roughly $26.70 all window while the blend falls to $14.49, cost per paid user falls from $191 to $148, lifetime value - capped at a 36-month horizon rather than taken to perpetuity - rises from $534 to $852, and the ratio moves from 2.8 times to 5.7 times with payback compressing from 9.6 months to 4.6.