$347,000 Gross. $211,500 Net.
One Breakdown. Five Years of Freelancing.

Every freelance income post shows an annual number. None show the month inside the year. My peak gross was $110,000 in 2022. My lowest month that same year was $6,200. The month before my highest ever invoice, I was calculating whether I could defer a credit card payment.

*Net = gross minus Upwork fees (5–20% sliding scale), self-employment tax (15.3%), estimated federal/state income tax (~15% effective rate). Excludes health insurance ($450/mo in 2023) and retirement savings ($0 most years). These exclusions make net look better than it was.

I tracked rates through direct conversations with 23 freelancers between 2020–2024. PROBABLE This is not survey data, not statistically valid, and unverifiable. These are numbers people told me. The table below shows what I heard — including where my own rates landed against those tiers. For a fuller picture of current freelance developer rates and what actually determines them, that breakdown covers market benchmarks I couldn’t access during my run.

My rate progression — actual numbers:

For context on why platform freelancing via Upwork has structural disadvantages versus direct client acquisition — particularly the race-to-rate-floor dynamic — that’s a pattern I watched accelerate between 2021 and 2023.

I started with structural advantages that I cannot separate from the income data. ESTABLISHED Ignoring these would make this post useless for anyone who doesn’t share them.

In 2019, I believed freelancing would be exposure therapy for anxiety. Facing uncertainty would build resilience.

What actually developed was not resilience. It was hypervigilance — a distinction my psychiatrist made explicit. ESTABLISHED

In July 2022, I received a Generalized Anxiety Disorder (GAD) diagnosis after three months of panic attacks. The psychiatrist asked: “Do you constantly monitor for threats?” The answer was yes on every channel simultaneously.

ESTABLISHED per Mayo Clinic GAD diagnostic criteria: I met 6 of 6 criteria. This is not “being careful.” The monitoring behavior was a symptom pattern, not a professional habit.

The financial consequence: I earned $66,000 net in 2022 and felt materially poorer than I do now earning $165,000 at Stripe. PROBABLE Predictability has monetary value I failed to calculate. For a realistic look at how financial planning differs between freelance and salaried developer income — particularly the hidden costs of income variance — that framing may help before making the jump.

What I genuinely don’t know: whether those who “broke through” had superior psychology, superior skills, or simply a higher tolerance for prolonged distress. I do not know if that distinction is meaningful or whether it is just different failure modes arriving at different times. SPECULATIVE

Hard limits on what this post can tell you:

What I suspect but cannot prove: SPECULATIVE

Platform freelancing (Upwork, Fiverr) gets structurally harder each year — more competition, AI commoditization of generalist tasks, rate pressure from global supply. Direct client relationships matter more now than when I started; roughly 60% of my income came from referrals, but I cannot verify this is representative. Most people who attempt freelancing quit within 18 months — based on watching approximately 50 people start in communities I joined, without systematic tracking. For analysis of what sustainable developer career paths actually look like beyond the freelance-vs-employment binary, the range of exit points is wider than most “quit your job” content acknowledges.

I cannot tell you whether to try freelancing. I can describe what I wish I had done differently — not as prescription, but as the specific failure pattern I ran.

The minimum viable test: One paying client secured within 90 days while still employed. If you cannot do this with zero financial pressure, the information is that you are not ready — not a moral judgment, just a data point. See also: what a developer portfolio needs to do before it generates referrals, which is the precondition for the 60% referral rate I eventually reached.

Red flags this isn’t the right moment: