What going freelance actually does to your career

"Going freelance is the best thing you can do for your career" is asserted far more often than it is measured. Here is a measurement that runs against it. In the Federal Reserve's 2025 household survey, 58% of self-employed adults said their income varied from month to month, against 28% of people working for someone else, and 22% of the self-employed said the variation left them struggling to pay bills, against 10% of employees (Federal Reserve, Economic Well-Being of U.S. Households in 2025, published May 2026).

That is not an argument against independence. It is an argument against the framing. Independence is a trade with a real credit column and a real debit column, and both sides have grown since 2020 — the upside is larger and the downside is now more expensive to cover, particularly in the US. What follows is the ledger. Whether the numbers work for you specifically is a set of three financial tests, and this post deliberately doesn't repeat them.

The one durable advantage is that you can reprice

Your price stops being set once a year by someone else. An employee's compensation moves at the cadence of a review cycle and within a band designed for a role. An independent's price moves per engagement, and if the market repositions, so can you — within a quarter rather than within a promotion cycle.

That mechanism is currently visible in the data. Upwork's Future Workforce Index 2026, fielded March to April 2026 with 2,400 US skilled knowledge workers, found freelancers doing AI-related work earning 34% more per hour, with earnings on complex AI-augmented work up 45% (Upwork FWI 2026). MBO Partners, whose survey counts anyone who did any independent work in the past year, put 5.6 million US independents above $100,000 in annual earnings in 2025, up 19% on 2024 (State of Independence 2025). Those two count different populations and neither is a rate table, so read them as direction rather than level. The direction is that repricing toward where demand moved is available to independents on a timescale employment does not offer.

The same market punishes standing still faster. Upwork's own data shows generative-AI creative production with contract starts up 90% year over year and earnings per contract down 13% — more work, worth less each. Repricing is a capability, not a guarantee, and it only helps if you use it.

Concentration risk becomes divisible, if you maintain the mix

An employee has one client. A redundancy is a 100% revenue event with a notice period attached. Five clients at roughly equal weight makes the loss of one a 20% event you can absorb while replacing it. That is genuinely better risk structure, and it is the part of the independence case that survives scrutiny best.

The catch is that most independents do not have five clients of equal weight. They have one client at 60% and four at 10%, which is employment with worse terms and no notice period. The diversification benefit is real but conditional, and the condition is a client mix you have to actively maintain rather than one that appears.

You get a feedback loop that employment hides

Inside a company, the connection between what you do and what it earns is mediated by several layers and usually invisible to you. Independently it is one arithmetic step: fee divided by every hour the project consumed, including scoping, revisions, invoicing and chasing. That number, the effective hourly rate, is the closest thing to a career instrument you get, because it tells you which work to stop taking. Most people who freelance for five years and price better at the end got better because they were forced to look at that number, not because they became more confident.

The employer pension contribution does not reappear in your fee

In the UK, automatic enrolment obliges an employer to put in a minimum of 3% of qualifying earnings between £6,240 and £50,270 (GOV.UK). On a £60,000 salary that is 3% of £44,030, or £1,321 a year, and many schemes pay well above the minimum. That contribution does not become part of your fee when you leave. It becomes a line you fund yourself or a hole you don't notice for a decade.

In the US the same item shows up in the aggregate: retirement and savings accounted for $1.57 of the $46.60 per hour employers spent on private-industry compensation in March 2026, or 3.4% of the total (BLS Employer Costs for Employee Compensation). Payroll tax moves too. An employee pays 7.65% in FICA and the employer pays the matching half; a self-employed person pays the whole 15.3% as self-employment tax, half of which is deductible above the line (IRS).

Group pricing was doing more work than you noticed

This is the item that got materially worse in 2026 and the one people underestimate most. KFF's 2025 employer survey, covering more than 1,800 employers with at least ten workers, put the average annual premium for family coverage at $26,993, of which workers contributed $6,850 — so the employer absorbed $20,143 (KFF, October 2025).

Leaving means buying that yourself, and the price of buying it yourself changed. The enhanced ACA premium tax credits expired after 2025. For 2026, average net premium payments for marketplace enrollees rose 58%, average net deductibles rose 37%, and effectuated enrollment fell from 22.1 million to 19.2 million by February 2026 — about three million people (KFF). Anyone modeling a US freelance budget on 2024 premium assumptions is out by a wide margin.

The UK version is smaller but not zero. There is no employer sick pay: Statutory Sick Pay of £123.25 a week is an employee entitlement (HMRC rates 2026 to 2027), and the self-employed have neither that nor an employer income-protection scheme. Statutory paid holiday of 5.6 weeks (GOV.UK) becomes 5.6 weeks you neither bill nor get paid for, which is a 10.8% reduction in billable weeks that has to be inside your rate.

One smaller change is worth naming because it is easy to miss: as an employee you filed expenses into somebody else's approval queue and the money came back. Independently every expense is your own margin, which is why the useful comparison is against tools built around your own money rather than an expense queue.

The gap on a CV, and what we honestly know about it

The claim that a long freelance stretch reads badly to employers is common and the evidence for it is thinner than the confidence with which it is stated. What we do have is a systematic review and meta-analysis of roughly 67,000 fictitious job applications, published 14 May 2026, finding that employers treat unemployed and inactive candidates less favorably than employed ones, with short spells under six months carrying no penalty and negative effects becoming noticeable after about twelve months (D'hert, Baert & Lippens, Socio-Economic Review).

That study measures unemployment and inactivity. It does not measure self-employment, and no comparable audit evidence for a freelance record exists that we could verify. So the honest position is that we do not know the size of the effect, and anyone quoting one is guessing. What the research does suggest is a mechanism: the penalty attaches to illegibility and to inferred idleness rather than to the absence of an employer's name. A record listing named clients, dated engagements and outcomes reads as continuous work. A record saying "freelance consultant, 2022–present" does not, and it is the second version that invites the inference the research is measuring.

The sector variation is real and unmeasured too. In fields that hire through structured leveling (large tech, banking, consulting partnerships) an independent stretch has no obvious slot in the ladder and re-entry is usually at a level, not a title. In agency, creative, and specialist technical markets, client work is the standard currency. Know which one you are in before you assume either way.

The ledger

Read the table as a whole and the shape is clear. Independence trades a set of predictable, employer-subsidized costs for a variable, self-funded set, and buys with it price control and divisible risk. Whether that trade is good depends entirely on whether the price control is exercised — which is why the same decision produces very different outcomes for two people with identical skills. Someone entering the market with no track record faces a steeper version of the same trade, which is a different calculation again.

The condition under which the trade pays

Here is the threshold. Independence is career-positive if, within roughly two years, your realized effective hourly rate exceeds the employed replacement cost of your old role — measured as total employer spend rather than gross salary — by enough to fund the pension contribution, the health cover, the unpaid leave and a reserve against the variance the Federal Reserve data describes.

Below that line you have taken on the debit column without collecting the credit, and you are not building a career, you are subsidizing your clients.

The way to know which side you are on is to measure the realized rate rather than the quoted one. Quote a number with the rate calculator, then check what it actually became at project close, and revise upward if the two keep diverging. If they never converge, the market is telling you something about the work you are selling and it isn't a confidence problem. The wider sequence of what has to be in place, and in what order, is set out in the milestone guide to going freelance.

Worklyn's CFO Mode runs Safe to Spend and a rolling 13-week forecast off your real numbers, which is what turns the income variance in that table from a surprise into a schedule.