Developer Tools & Infra SaaS unit economics, benchmarked for 2026.

Developers are the only enterprise buyers who can evaluate, adopt, and deploy software without permission — and the only ones who will actively punish you for trying to sell to them. DevTools economics are therefore built bottom-up: a CLI installed in seconds, an API key generated without a sales call, usage that spreads team to team until procurement gets involved. The motion shows in the numbers: a ~2-month cycle, CAC of $1,200–$9,000, ARPA spanning $150–$1,800, and gross margin at 86%.

What makes the cohort exceptional is what happens after adoption. At 1.7% monthly churn, developer tools are among the stickiest products in software — because once your API is embedded in a company's production codebase and your tool runs inside their CI/CD pipeline, replacing you means rewriting infrastructure. Nobody volunteers for that. Expansion is usage-shaped rather than seat-shaped: build minutes, API invocations, data throughput grow as the customer's engineering org grows, often without any commercial conversation. The signature is trust-based compounding — slow to earn, brutally hard to displace, and expanding on its own momentum.

First, developer relations salaries. DevRel advocates, documentation engineers, and sample-code authors are the demand-generation team — conference talks, tutorials, and Stack Overflow presence replace SDR sequences dollar for dollar. A senior DevRel hire costs what an AE costs; the channel is just different.

Second, open-source maintenance as marketing spend. Engineering time poured into the free core — issue triage, PR reviews, release management — is acquisition investment wearing an engineering costume. It belongs in CAC math even though it sits in the R&D budget.

Third, free-tier infrastructure. CI minutes, API calls, bandwidth, and storage for non-paying users are real, metered COGS that scale with adoption. A generous free tier that converts at 2% means 98% of your compute bill is unmonetized — the most literal form of CAC in the cohort.

Fourth, the enterprise sales-assist ambush. Bottom-up adoption eventually hits procurement: SSO/SAML requirements, SOC 2 questionnaires, security reviews, data-processing terms. The "self-serve" deal that needs six weeks of security review just acquired an enterprise CAC — price and staff for it, or watch payback double on your largest accounts.

The 4.8x LTV:CAC reflects genuine structural advantage: 1.7% monthly churn implies a ~59-month customer lifetime, and usage-based expansion means year-three ARPA routinely doubles year one. An 8-to-11-month median payback is achievable because self-serve CAC stays low while expansion does the heavy lifting. Top-quartile teams under 7 months combine viral CLI distribution (single-command install, zero credit card) with consumption pricing that expands automatically — the product sells the upgrade, not a human.

The bottom quartile past 14 months splits into two failure modes: free-tier abuse (unmetered compute consumed by non-converters, occasionally literal crypto miners) and enterprise deals where a 12-month security review turned a $3,000 self-serve CAC into a $30,000 assisted one. Read payback segmented — self-serve and enterprise-assisted are different businesses sharing a logo. And guard the 86% gross margin jealously: egress fees and dedicated infrastructure for large tenants can erode it fast, so usage-based tiers need margin floors built into the pricing math, not discovered in the COGS report.

The most expensive mistake in DevTools is optimizing for GitHub stars instead of product-qualified leads. It feels like traction: 50,000 stars, a front-page Hacker News launch, conference talks to packed rooms, a Discord with ten thousand members. Meanwhile the metrics that pay — workspaces created, API keys in production, seats on paid plans — stagnate, because stars measure admiration and PQLs measure intent. Engineering salary keeps flowing into the open-source core to feed the community's appetite for features, while the conversion path from free to paid stays an afterthought. The 98% of users who will never pay keep consuming the free-tier compute that shows up on your cloud bill every month. Founders raise money and hire against community metrics, then discover the unit economics quietly inverted a year ago: CAC (fully loaded with DevRel and OSS maintenance) rising, conversion flat, and a beloved project that the market applauds and nobody buys. Stars don't churn — but they don't renew either. Instrument the funnel from `npm install` to paid invoice, gate the free tier's expensive dimensions, and treat every roadmap decision as a CAC decision. In DevTools, community is the top of the funnel, not the business.