The quiet headline of the indie SaaS movement isn't the revenue screenshots; it's the margins behind them. A no-employee micro-SaaS in 2026 routinely runs 70–90% net margins, a profitability profile better than software companies of any previous era and better than almost any small business ever. At $10K MRR and 80% margins, a solo founder takes home roughly $96K/year from a laptop.
Those margins aren't automatic, though. They're a structure, and structures have load-bearing walls and known failure points. Here's the full playbook: where the margin comes from, where it leaks, and the decisions that protect it as you grow.
Where the margin comes from
The anatomy of a typical solo SaaS at $10K MRR:
| Line | Monthly | % of revenue |
|---|---|---|
| Infrastructure (hosting, DB, email, the standard stack) | $100–$300 | 1–3% |
| LLM/API costs (if AI features) | $100–$500 | 1–5% |
| Payment processing (2.9% + 30¢) | ~$320 | ~3.5% |
| Tools & subscriptions (agent, analytics, misc) | $100–$200 | 1–2% |
| Total costs | $600–$1,300 | 6–13% |
Three structural facts produce the famous margins. Software's marginal cost stays near zero: customer #500 costs you cents. The 2026 toolchain replaced headcount: the boilerplate replaced the contractor, the coding agent replaced the second engineer, and support automation replaced the first hire; payroll, the line that eats every other business, reads $0. And distribution can be owned rather than rented: compounding channels like search, comparisons, and community cost time rather than taking a percentage.
Notice payment processing is typically the largest cost line. That's what a healthy structure looks like: when your biggest expense is the fee for collecting money, everything else is working.
Where margins leak
Every founder who reports margin erosion names the same five leaks:
Paid acquisition. The fastest way to turn 85% margins into 30%: renting distribution at $2–10 a click against a $50/month product. Ads have legitimate uses post-validation, but a margin structure this good exists because indie SaaS runs on owned channels; the playbook is slower and keeps the 80 points.
Unmanaged AI costs. The newest leak: inference that scales per-action against pricing that doesn't. The defenses (model tiering, caching, quotas) are each a day's work and protect 5–15 points of margin at scale.
Tool sprawl. $29 here, $49 there: solo founders accumulate subscriptions like ballast. Quarterly audit rule: every tool re-justifies itself or dies. The boring discipline is worth 2–4 points forever.
Support gravity. The first cost that scales with customers is your time. Past ~200 customers, unstructured support quietly becomes a part-time job, the margin leak that doesn't appear on any invoice. The fixes compound: onboarding that prevents tickets (activation work doubles as retention work), docs written as answers, and an AI support layer over those docs, which in 2026 deflects most tier-1 volume credibly.
Discount drift. Lifetime deals, perpetual launch coupons, unreviewed legacy pricing. Each feels small; together they routinely shave 10–20% off effective revenue. Audit your actual average revenue per customer against your list price; the gap is the drift. (And never LTD a product with per-use costs.)
Protecting margin while growing
The strategic version of the playbook in three decisions:
Price for margin, not just for growth. A price increase is the only growth lever with ~100% margin flow-through: every extra dollar is profit at this cost structure. Most indie products are underpriced by 2–3x against the value delivered (run the math); a 30% price increase typically outearns a quarter of feature work, with zero new costs.
Hire software before people. Each potential hire in a solo SaaS is a 10–30 point margin decision. The 2026 sequence that preserves the structure: automate (agents, workflows), then productize (docs, self-serve), then contract narrowly (a freelancer-day for the rare specialist task), and only then hire, when revenue per founder-hour genuinely caps out. Many founders never reach step four; that's the one-person-business thesis working as designed.
Let the margin buy your runway, not your lifestyle, early. The deep strategic value of 80% margins pre-$10K: each month of revenue funds many months of costs, which means you can survive the 12–18 month median timeline indefinitely. Margin is time, and time is what bootstrapped products actually compete on.
Frequently Asked Questions
What net margin should a solo SaaS expect?
70–90% net is the normal range in 2026 for a no-employee SaaS: infrastructure runs $100–$1,300/month against four-to-five-figure revenue, payment processing takes ~3.5%, and there's no payroll. AI-heavy products land at the lower end unless inference costs are actively managed with model tiering, caching, and per-user quotas.
What's the biggest cost for a one-person SaaS?
On the invoice: usually payment processing (2.9% + 30¢, about $350/month at $10K MRR), with infrastructure surprisingly far behind. Off the invoice: founder time consumed by support and ops, which is the first cost that genuinely scales with customers and the leak most founders notice last. Onboarding quality, answer-shaped docs, and an AI support layer are the standard 2026 containment.
How do AI features affect SaaS margins?
They add the only cost line that scales per-action rather than per-month. Unmanaged, inference can take a product from 85% to 50% margins; managed (small-model routing for routine calls, response caching, hard quotas, usage-aware pricing tiers), AI features typically cost 3–8% of revenue. The management is a few days of engineering that pays permanently.
When should a solo founder make their first hire?
Later than instinct says: each hire is a 10–30 point margin decision, so exhaust the cheaper sequence first: automate the task with agents and workflows, productize it away with docs and self-serve design, contract it narrowly to a specialist by the day. Hire when revenue per founder-hour is genuinely capped and the role generates more than it costs including the management overhead. At 2026 tooling levels, many seven-figure solo products never get there, by design.