The Pain: Open any one-person business community and the same line shows up: one person plus AI, $10K a month from the start. Then you look at the public tracking of 8,000+ micro-SaaS projects — average MRR $4,298, median MRR $145. The screenshots you keep seeing were taken by the 6.1%.
What You'll Learn: A checkable set of ledgers for the one-person business — the real revenue distribution of micro-SaaS, the concrete numbers behind three one-person cases, the dividing line between top-10% operators and the median, and a framework that stops you from judging a whole category by its average.
⚡ 10-minute speed read: read the numbers in sections 1–3, then jump to section 7 ("Where you should start") and the closing line.
🎯 Read what you need: sections 4 and 5 for the real ledgers; section 6 for the trend judgement.
📖 Full read: about 12 minutes to get the real revenue structure of a one-person business in the AI era, plus the framework for reading it.
1. Put the two numbers on the table: $4,298 and $145
The core claim: the average is not the income level — the median is. Across 8,000+ micro-SaaS projects, average MRR is $4,298 and median MRR is $145.
The numbers come from a tracking platform that follows 8,000+ micro-SaaS products, where revenue is verifiable in the billing backend rather than self-reported in a survey. Add Forbes coverage of one-person marketing firms and independent-operator income benchmarks, and that is the numerical base for this piece.
| Metric | Value |
|---|---|
| Average MRR across projects with revenue | $4,298 |
| Median MRR | $145 |
| Share passing $10K MRR | only 6.1% |
| Projects in the $1K–$50K band | about 850 |
Two numbers side by side, nearly 30x apart. I have a name for this: the median trap — using an average to represent the whole field, and treating the height of a few outliers as the normal state of everyone else.
The $4,298 average is real, and the $145 median is real. They do not contradict each other; they answer different questions. The average answers "how high can this category go". The median answers "where does a randomly arriving project most likely sit".
Pitfalls in this section:
- Any claim of the form "average income of XX per month" needs a median before you can judge it
- When the average sits dozens of times above the median, there are extreme values in the sample — do not picture the category as the average
2. $145 is not a product problem, it is a distribution problem
The core claim: the low median in micro-SaaS is not because the product cannot be built. It is because nobody finds it once it exists.
The $1K–$50K band holds about 850 projects — that is the tier that is genuinely alive. So why do far more projects stall at $145? Because building the product completes only half of the business.
Look at the order of starting speed: digital products 1–3 weeks, services 2–8 weeks, SaaS 8–16 weeks. People who build SaaS invest the longest stretch into the hardest single thing, and the hardest part of that thing is not writing the code — it is making people know it exists.
Distribution cost is trending to zero now that platform recommendation algorithms have replaced ad buying. But "trending to zero" is not "automatic": it requires you to keep producing content and keep being seen. The product is the pass; distribution is the door.
Pitfalls in this section:
- Before building, answer one question: what is my distribution channel? Without that answer, whatever you build most likely lands in the $145 tier
- Do not treat "the product is finished" as "the business is finished"
3. What the ceiling looks like: $200K, $150K, $113K
The core claim: a ceiling at the top does exist, but it is a survivor sample — projects past $10K MRR are 6.1% of the field.
Checkable ceiling samples: Rezi (an AI resume tool) around $200K MRR, Tally (forms) around $150K, Typefully around $113K, then Pallyy around $85K and Submagic around $83K.
That brings the second term I want to introduce: survivorship bias — every success story you see has been selected out of a silent majority. Out of 8,000 projects, the ones that get turned into a story are that 6.1%, or fewer. The value of this particular statistics report is exactly here: it writes down the denominator as well.
Pitfalls in this section:
- Before benchmarking against the ceiling, confirm which segment of the distribution you actually sit in
- A small number of cases does not make a path repeatable; find a reference at your own tier first
4. Three real one-person ledgers
The core claim: a one-person business is not "one person carrying everything" — it is "one person plus a set of Agents dividing the work". The difference is systems engineering, not how hard you work.
Three checkable cases:
| Case | Model | Real numbers |
|---|---|---|
| Ravenopus (Linara Bozieva, former senior analyst at a large company) | One-person marketing agency, 35 specialised AI Agents splitting the work | $20,000–$30,000 per month, founded May 2024, profitable on day one |
| Medvi (Matthew Gallagher) | GLP-1 telehealth | Started with $20,000, no employees, no office |
| 500k.io (Maxime Le Morillon) | Independent operator publishing MRR openly | $9,500 MRR |
The Ravenopus case is the one worth studying: 35 Agents are not "a pile of automation scripts". They are the job functions of a marketing agency broken into modules, with one dedicated Agent per module. The part the human keeps is decomposition and judgement.
That is also why, on the same tool stack, one operator reaches $20,000–$30,000 a month while another stops at $145. The missing stretch is not tool proficiency — it is the ability to break a business into deliverable modules.
Pitfalls in this section:
- Do not copy "one person carries all the work"; copy "split the work into modules, then hand modules to Agents"
- The scale of a case is not repeatable, but the decomposition method is
5. The dividing line between the top 10% and the median
The core claim: the difference between top operators and median operators is not tools. It comes down to two things — publishing your metrics to build trust, and delivering the same revenue in fewer hours.
The 500k.io tier study gives two dimensions.
First, metric transparency. Top operators publish MRR and key metrics; median operators do not. I have a name for that too: the compounding of transparency — publish real numbers, build trust, earn conversion, and let it roll into compound interest. Once data is verified, a layer of trust settles, and that layer is an asset you can call on repeatedly.
Second, hours invested. The tiering dimension is not income alone: the same $100K ARR delivered in 30 hours a week and in 60 hours a week are two different tiers. The revenue figure matches; the quality of the business does not.
Pitfalls in this section:
- Publishing data is not showing off; it lets other people verify you cheaply. Publishing process metrics is a valid start
- Looking at revenue without looking at hours invested leaves the time cost out of the account
6. 2.86 million registered one-person companies: the cost structure changed
The core claim: the rise of the one-person company is not sentiment, it is a change in the cost structure — AI broke company capability into modules that can be outsourced to Agents.
China recorded 2.86 million registered one-person companies in 2025. Behind that number, four things happened at the same time:
- AI broke "company capability" into modules that can be handed to Agents — marketing, support, content, data analysis
- Startup cost dropped from hundreds of thousands to a few thousand yuan — cloud tools plus subscriptions
- Distribution cost is trending to zero as recommendation algorithms replace ad buying
- The company as a form did not disappear, but "one person coordinating with AI" became the new default shape for starting up
Read that together with the $145 from section 1 and the logic closes: lower barriers let more people enter, which pushes the median down, and the real dividing point moves from "can you build it" to "can you be found, and can you be trusted".
Pitfalls in this section:
- Registering is easy; surviving is not — lower barriers accelerate homogenisation at the same time
- The cost structure changed, but the revenue structure did not: it still runs on verifiable distribution and delivery
7. Where you should start
The core claim: do not benchmark against $200K directly. The first target should be the $1K–$5K MRR tier — about 850 projects already sit there.
Step one: write down your distribution channel. If you cannot name a specific channel, do not start building the product yet. Services (2–8 weeks to start) get you feedback faster than SaaS (8–16 weeks), and they get you distribution experience faster too.
Step two: sell a service first, then productise it. The service stage gives you paid validation and delivery instincts; the product stage gives you scale. Skipping services and going straight to product means betting on both at once.
Step three: start publishing your process metrics. That is the starting point of the compounding of transparency, and the cheapest way to turn accidental revenue into reusable trust.
Pitfalls in this section:
- Do not treat incorporation as the start; the first paying customer is the start
- Do not make a short-path choice with a long-path mindset, and the reverse holds just as well
🔔 Where you stand right now
The one-line version: the truth about a one-person business is not in the phrase "one person" — it is in the revenue distribution. The average answers the ceiling, the median answers your position, and what decides which end you are on is distribution and trust.
Three things to hold on to:
- The average is the promotional figure; the median is the survival figure. A $4,298 average corresponds to a $145 median — so for any "average income per month" claim, ask for a median first
- A finished product is only half the work: the other half is being found. The distribution gap is more common than the technical gap
- The leverage in a one-person business is decomposition. The 35 Agents behind Ravenopus are the result of splitting a business into modules, not the result of mastering tools
💎 What is actually worth taking away
Value one: an income map you can check yourself against. Scenario: you see any "one-person business earning XX per month" claim. Solution: place the number inside the distribution (median $145, only 6.1% past $10K, about 850 projects in the $1K–$50K band). Reusable value: you can tell at a glance which tier a case belongs to and stop being carried away by a survivor sample.
Value two: a method for splitting a business into modules. Scenario: one person has to cover product, content, delivery and support at the same time. Solution: split by function, give each module one dedicated Agent, and keep only decomposition and judgement for yourself. Reusable value: the method does not depend on a specific tool, so it can be rebuilt in any industry.
Value three: a way to treat trust as an asset. Scenario: deciding whether to publish your own data and process. Solution: publish real metrics, build verifiable trust, and let it compound. Reusable value: once trust settles, it lowers the cost of every future acquisition.
Three steps to act on
| Step | Action | Check |
|---|---|---|
| 1 | Write down your distribution channel; if you cannot name one, do not build the product yet | You can list at least one channel that reaches your target customers repeatedly |
| 2 | Sell a service first for paid validation, then consider productising | You receive your first real payment within 30 days |
| 3 | Publish one process metric (delivery cycle, customer questions, cost) | Four weeks of continuous publishing, and someone approaches you because of it |
The line to keep: the average tells you where the ceiling is, the median tells you where you are standing — and the moment you see the second one clearly is the moment the business starts.
📖 Further reading from the Practitioner's series
- Why the One-Person Company Is Inevitable in the AI Era: From Mass Advertising to Precision Matching
- Selling the System: From a One-Person Company to a Replicable Business System
- AI Agent Infrastructure: The $350/Year Stack Behind an OPC
About the author: Guanlan (观澜) — AI / Agent / digital transformation practitioner. Practical, hands-on writing — follow along and it just works.
Top comments (0)