Worse Margins, Better Businesses
The New Bottom Line
CAC to LTV. CAGR. ARR and NRR. Day 7 retention. Logo churn.
These are the commandments of SaaS.
But here’s the thing nobody realizes: those metrics aren’t tech metrics. They’re PE metrics.
The Path to Exit
The vast majority of software exits are financial, not strategic. Most founders won’t sell to Google. They’ll sell to Vista, Thoma Bravo, or some growth equity fund that’s buying a revenue stream with predictable characteristics.
And smart founders know one thing: build what the buyer wants to buy. Well, PE wants to buy businesses with recurring revenue, high gross margins, low churn, and a customer acquisition engine that can be modeled in a spreadsheet.
So that’s what we built. An entire industry optimized its operating model around the preferences of its eventual acquirers (and growth investors too!). SaaS didn’t emerge because subscription software was some natural law of technology. SaaS emerged because PE found a way to project its ideal business model onto the tech industry, and founders — rationally — built to spec.
The 80% gross margin benchmark? That’s not a technology insight. That’s a financial buyer saying: “I need enough room above COGS to service debt, pay my operating partners, and still generate returns.” The whole SaaS gospel is, at root, a leveraged buyout thesis wearing a hoodie.
Above the Line
Here’s the thing about the SaaS model that made it so beautiful for financial buyers: the margins were almost guaranteed.
Software’s marginal cost was trivial. Once you built the product, serving the next customer cost nearly nothing — some server capacity, a few support tickets. Gross margins of 80%+ weren’t aspirational. They were structural.
Now? COGS matter. Every feature that touches a model carries real marginal cost — inference, tokens, compute. The days of assuming 80% gross margins as a baseline are ending. A company running heavy AI workloads might land at 40% or 45%, and that’s not a failure — that’s the physics of the product.
Below the Line
But SaaS had a well-known catch: headcount. Engineering teams, sales teams, customer success teams, marketing teams. A company could run 80% gross margins and still burn cash because it took 200 people to build, sell, and support the product.
But the same AI capabilities that increase COGS are simultaneously making teams radically more efficient. Engineering teams are shipping with fewer people. Support is increasingly automated.
One company’s headcount reduction is another’s newfound revenue.
The PE playbook assumed high gross margin, high operating expense, modest operating profit. This made for a simple playbook: wait until the business was mature enough, buy the recurring revenue stream, then squeeze headcount until operating margins materialized. It worked because the gross margin cushion was enormous and reliable.
The Bottom Line
But the new AI P&L assumes lower gross margin, lower operating expense, but hopefully better operating profit.
SaaS metrics were never really about technology. They were about financial engineering. And now the financial engineering is changing.
The companies that win the next decade will be the ones that figure out the new math: AI raises your cost of goods and lowers your cost of people, and if you manage both sides well, you end up with a better business than SaaS ever produced.
Worse margins. Better businesses.
Originally published on Griff’s Substack on 27 March 2026.