Why the model you pick changes the maths

The economic case for taking retention seriously is not new, and it is not subtle. In their 1990 Harvard Business Review study "Zero Defections", Reichheld and Sasser showed that cutting the customer defection rate by just five per cent raised profits by between twenty-five and eighty-five per cent depending on the industry. When MBNA halved its ten per cent defection rate, profits rose well over a hundred per cent. Sit with that: the same five-point move in churn that an acquisition-obsessed founder would not even notice on the dashboard is, at equal effort, a larger profit lever than almost anything they could do at the top of the funnel. A retention-model bet pays out bigger than an acquisition-model bet at the same effort, and it has done for decades.

The number a Revenue-terminated model buries. Fast-forward to today and the proof has only sharpened. SaaS companies with high net revenue retention grow markedly faster than their low-NRR peers, with the better datasets putting that gap at roughly 2.5 times the growth rate. The median venture-backed firm sits near 106 per cent NRR. Above that line you compound revenue from customers you already have; below it you lose ARR every period and then have to win new business just to draw level. That gap is decided entirely by a stage that classic AARRR pushes off the end of its diagram.

The segment trap. The data makes the trap concrete. Enterprise accounts hold net revenue retention well above SMB, and SMB churn runs many times higher than enterprise. The same product can look like a growth machine or a leaky bucket depending purely on which segment, and therefore which model, you built for. Choose the wrong lens and you will read your own numbers as success while the business quietly leaks. If you sell to SMB, expansion is not a nice-to-have; it is the only thing that keeps the bucket from emptying.

Expansion is now structural, not optional. The final blow to the Revenue-terminated model is where the growth has physically moved. Median expansion revenue rose from roughly 25 per cent of total new ARR in 2022 to about 40 per cent in 2024, and for companies above fifty million ARR it now generates the majority of new ARR, in the region of 58 to 67 per cent, surpassing new sales outright. A model that ends at "Revenue" systematically under-weights the very place the growth now comes from. This is exactly why Winning by Design re-drew the funnel as a bowtie, and exactly why, for a recurring-revenue business, choosing the bowtie over the pirate funnel is not a matter of taste. It is a matter of pointing your attention at the larger half of the picture instead of the smaller one.