Why revenue multiples alone don't tell you whether a company is on track

A SaaS company with $5M ARR growing at 3x year-on-year is a fundamentally different investment proposition than one with $20M ARR growing at 30%. But most VC deal models treat both with the same blunt instrument — a revenue multiple applied to a projected Year 5 number. The multiple is borrowed from comps, the projection is optimistic, and the result is a "base case" that nobody really believes.

What's missing from the revenue multiple approach is any sense of whether the company's growth trajectory is fast, slow, or average relative to companies that have actually exited at scale. The CGI framework answers that question systematically — by indexing a company's revenue and user growth against a calibrated benchmark set across five cohorts, producing a growth percentile that tells you exactly where this company sits in the distribution of historical outcomes.

The insight behind the framework

In oil and gas, engineers don't estimate a well's value by applying a price multiple to its current output. They fit its production trajectory to a decline curve, compare it against a type curve for the formation, and read off a percentile. That percentile — not the current output level — is what drives the reserve estimate and the investment decision. CGI applies the same logic to company growth: trajectory relative to cohort, not revenue multiplied by a guess.

Why combining revenue and user growth matters more than either alone

CGI normalises both revenue and user growth to a common index scale, then combines them using a weighted composite formula. The revenue milestone of $100M is the anchor point — a CGI of 100 means the company has hit that threshold in the index.

The formula is specifically designed to penalise imbalance — a company cannot score well on CGI by excelling on only one dimension. Both revenue and user signals must be present and proportionate. The mathematical mechanics behind this property are taught in Module 1.

For B2B companies where consumer MAU data doesn't exist, a proxy approach is used. This means B2B CGI scores are comparable to each other but carry an important caveat when compared against consumer platforms — one that significantly affects interpretation. The proxy methodology and when to apply it is covered in Module 1.

Benchmarking against Stripe, Slack, Uber, and Salesforce — at the same point in their growth

The benchmark set spans five distinct growth cohorts, each with a measurably different S-curve shape reflecting different go-to-market models, founding eras, and capital efficiency profiles. The question "is this company growing fast enough?" only has meaning relative to the right peer group:

The most consequential decision in the framework

A company growing at 3x year-on-year looks exceptional against a Classic SaaS benchmark. Against an AI Ultra-Fast benchmark, it might be below median. Cohort choice determines whether a company looks like an outlier or a laggard — which is why the assignment methodology matters as much as the index itself. When uncertain, run against two cohorts and compare: the one where the company scores P40–P60 is almost certainly the correct peer group.

Reading the growth curve: when is a company about to inflect?

Every company's growth follows an S-curve: slow start, rapid acceleration through an inflection point, then gradual deceleration as the market matures. Three parameters capture the shape. The ceiling (L) — how large the company ultimately gets. The steepness (k) — how fast it moves through the S-curve. AI Ultra-Fast companies have very high k; Classic SaaS very low. The inflection year (x0) — the year when growth is fastest. For AI Ultra-Fast companies this arrives at Year 2–3. For Classic SaaS it may not arrive until Year 8–12.

The inflection point is the most investment-relevant parameter for a GP. It tells you whether a company is still pre-inflection (early, high-risk, high-potential), currently inflecting (the highest-velocity moment), or post-inflection (scaling but decelerating). That distinction drives very different underwriting decisions — and it is invisible in a revenue multiple model.

Is this company tracking ahead of or behind its peer group?

A single fitted curve gives you the median trajectory for the cohort — the P50. Monte Carlo simulation generates 500 plausible curves from the cohort's historical parameter distributions, producing a P10–P90 envelope that shows the full range of outcomes consistent with this peer group's growth patterns. A company tracking above P75 is outperforming three-quarters of simulated cohort-consistent trajectories. A company below P25 is underperforming — and the framework tells you by how much, not just directionally. The bands are not forecasts. They are the range of outcomes the historical cohort data supports. Treat them as a structured prior, not a prediction.

From growth percentile to exit valuation: replacing the revenue multiple

This is where the framework directly replaces the revenue multiple. Instead of multiplying projected Year 5 revenue by a comp-derived multiple, you take the company's CGI percentile — say P72 in the SaaS/B2B cohort — and read the corresponding exit value from the Probit exit distribution for that cohort. The P72 exit value for SaaS/B2B companies, calibrated against real exit data, becomes your base case. The P10 and P90 become your downside and upside. The result is an exit range grounded in what companies at this growth percentile have actually achieved — not what a revenue multiple suggests they might.