You can't judge a SaaS engine by its roar — you read the gauges

An amateur thinks the louder the engine, the better the car. Any real engineer knows you can’t judge an engine by its sound — you read the gauges: RPM, fuel burn, temperature, torque, a whole set of interlocking readings that together tell you whether it’s healthy and how far it will run. Researching a SaaS company is the same, and this chapter assembles the full dashboard, using one fictional company (CloudCo) so every number connects to the next.

Start with the words for “revenue,” because it turns out to have many, and they are not synonyms. ARR/MRR is the stock — the annualized run-rate of recurring revenue at a point in time, and a company-defined metric, not GAAP, so it’s never safely comparable across companies. ACV/TCV is the size of a contract (and a fat multi-year TCV can just be several years bundled to look strong today). Bookings is a leading demand signal; billings is an invoicing proxy easily distorted by whether customers switched from quarterly to annual pay. Beneath all of them sits one structural fact the rest of SaaS finance flows from: cash, invoicing, and revenue recognition are separated in time — which is exactly why deferred revenue, RPO, and cRPO exist to show the contracted business sitting off the income statement.

One more warning before the heart of it: none of these metrics has a standard, audited definition. ARR, NRR, “large customer,” even “billings” — each company draws the lines differently, and a metric that quietly changes its basis between quarters is one of the loudest warning signs there is. Read the definitions before you read the numbers.

But the heart of the dashboard is retention, and the single most useful chart in all of SaaS analysis is the revenue bridge (the “ARR walk”): start-of-year ARR + new + expansion − contraction − churn = end-of-year ARR. Lay it out and almost every other retention and growth-quality metric falls out of it — which is why it’s worth rebuilding for any company you’re serious about.

Crucially, the bridge forces you to read gross and net retention together. CloudCo’s NRR of 105% looks fine — until you see its gross retention is only 90%. The bucket is leaking faster than it looks; the 105% leans entirely on expansion holding up. Let one recession stall expansion and NRR flips below 100% overnight. That’s the difference between a number’s level (which tells you the present) and its trend and composition (which tell you the future).

The unit-economics gauges finish the picture. A Magic Number of 0.5 and a ~32-month CAC payback say CloudCo’s growth is bought, not earned. And a shiny 7.5x LTV/CAC? It rests entirely on assuming customers stay 20 years — halve the retention and it halves too. Look at LTV; never let it drive the decision. The gauges don’t lie. You just have to read all of them, together.

This is Chapter 3 of “Compounding in the Cloud.”