Starting as a freelance writer in a market that just repriced

Most advice on starting as a freelance writer tells you to take cheap work first. Write for a content mill, bid low on a marketplace, build a portfolio, then raise your rate.

That ladder had its bottom three rungs removed. This post is about what to do instead, and it assumes you have no clients and no published bylines.

The entry route closed, and there is data on it

Demirci, Hannane and Zhu measured job posts on a large freelancing platform and found that writing and coding posts fell 21% in the eight months after ChatGPT launched, relative to manual-intensive work. The finding that matters more for a beginner is the second one: the posts that survived were more complex, better paid, and had more competition. The simple ones went away. The hard ones got busier.

Upwork's own 2026 data shows the price side. In generative AI and creative production, contract starts rose 90% year over year while earnings per contract fell 13%. There is more of that work than ever and each piece of it is worth less.

So the traditional first year (write a lot of cheap things, get good, charge more) now runs through the one part of the market where volume is up and price is down. You would be competing on the axis that automated, against everyone else doing the same, to build a portfolio of the work nobody is paying a premium for.

The route that's left is narrower and, oddly, shorter.

Your subject knowledge is the only scarce thing you have

You are not going to out-write anyone on fluency. Fluency is free now. What is not free is knowing how a claims adjuster actually decides, or what breaks in a warehouse management system rollout, or which questions a hospital procurement committee asks in month four.

That knowledge is the moat, and this post keeps returning to it. Every decision below is downstream of picking one subject area you already know from the inside: what to write about, what samples to make, what to charge, who to email.

Three questions to pick it:

1. What have you been paid to understand? Previous jobs, a degree you actually used, a family business, six years of a serious hobby with money in it. Not "I'm interested in fintech." What have you done.
2. Does someone in that area sell something with a marketing budget? Software, professional services, medical devices, industrial equipment, financial products. Charities and hobbyists have content needs and no budget.
3. Would a smart generalist writer get it wrong in a way an insider would notice? If the answer is no, the area is too general and the moat isn't there.

If nothing clears all three, pick the closest and accept a slower start. A weak niche beats no niche, and no niche beats a niche you invented from a trends article.

Samples that demonstrate judgement, not fluency

You do not need a portfolio. You need three pieces that prove you'd make good calls on someone's behalf. Write them yourself and publish them anywhere you control.

One teardown. Take a real published piece in your area, say a vendor's blog post or a whitepaper, and rewrite the parts that are wrong or vague, with your reasoning shown. This is the single most convincing artefact a new writer can produce, because it proves you can tell good from bad, which is the thing a buyer is actually purchasing.

One explainer written for a specific job title. Not "a guide to X" but the thing a procurement lead needs to know before a renewal meeting. Name the reader at the top. Judgement is visible in what you leave out.

One short piece with a primary source in it. An interview with a former colleague, a public dataset you actually opened, a regulation you read rather than summarised. This one demonstrates that you'll do the work a generation tool can't do, and it's the sample that gets quoted back to you in calls.

Skip the personal essay, the "why I went freelance" post, and the general writing-craft content. None of them price your specialism.

Set the floor before the first conversation, not after

The most expensive mistake available to you in month one is quoting low to get started. A rate isn't an opening bid, it's an anchor, and the client who hired you at $0.10 a word will not later agree to $0.40 — they'll find another beginner.

Work out your floor from your costs rather than from what you hope to charge. Annual income you need, billable hours per week you can realistically sell as a beginner (twelve to fifteen, not forty), weeks off, monthly expenses, and the percentage you set aside for tax. The rate calculator takes exactly those inputs and returns an hourly and a day rate. Whatever it gives you is the number below which the work costs you money to do.

Then convert to whatever unit your buyers use. If the market you're entering quotes per word or per piece, price the piece by estimating hours and multiplying, and add the calls and revisions you already know are coming. Beginners undercount revisions by roughly one round, every time.

There is a second reason not to start low. Upwork's 2026 index found freelancers doing AI-related work earn 34% more per hour, and its skills report put AI Data Annotation and Labeling up 154% and AI Integration up 178% year over year on marketplace earnings. Those are growth rates from an interested party, not salaries, and the report publishes no declines — treat it as a direction, not a promise. But the direction is consistent with everything else: the premium is attached to work that requires you to know something, and a low anchor makes it very hard to move into that tier later.

The first month, week by week

Week one: pick and build. Choose the subject area. Write the teardown. Set up somewhere to publish that you own, which can be a plain single-page site.

Week two: finish the samples and build the list. Write the explainer and the primary-source piece. Then list fifty organisations in your area that publish content — vendors, consultancies, trade publications, industry bodies. For each, find the person whose name is on the content or who runs marketing. Fifty is a morning's work and it is the asset that carries the next three months.

Week three: write to twenty of them. Not a blast. Twenty individually adjusted notes, five a day, each one referencing something specific that organisation published.

Week four: follow up, and write to twenty more. Roughly half your replies will come from the follow-up rather than the first message. Send it seven days later, in the same thread, two sentences long.

Expect the first month to produce conversations rather than contracts. If it produces one paid piece, that is a good month. If it produces none but four replies, the message is working and the list needs to be bigger.

The first-contact note, annotated

Short enough to read on a phone. This is the version for an in-house marketing lead; for a trade publication editor, swap the offer for a specific article angle, which is a different sale with its own terms and is covered in pitching editors: what actually gets commissioned.

Subject: your [specific piece] — one thing it left out >Hi [name], >I read your piece on [topic] last week. It's the clearest thing I've found on [narrow aspect], and it stops right before the part practitioners actually argue about: [the thing]. >I spent [N] years in [role] at [type of organisation], so I've been on the receiving end of that decision. I wrote up how it actually goes here: [link to your primary-source sample]. >If a piece on that would be useful to you, I'd be glad to outline one. If not, no reply needed. >[name]

What each line is doing:

• Subject line names their thing, not yours. It survives a phone preview and it isn't a pitch until it's opened.
• "It's the clearest thing I've found on…" — one sentence of specific praise. Two sentences reads as flattery. Zero reads as cold.
• "stops right before the part practitioners actually argue about" — this is the whole message. You've shown you know the territory past the point their content reaches. Cut this line and the email is worthless.
• "I spent N years in [role]" — credentials in nine words, positioned as why you know the thing rather than as a CV.
• A link, not an attachment. One link. Not a portfolio, not three samples, not a rate card.
• "If not, no reply needed." — removes the social cost of ignoring you, which measurably increases the number of people who don't ignore you. Keep it.
• No rate, no availability, no "I'd love to work with you." The first message sells a conversation.

Money hygiene from client one

Two things to set up before the first invoice, because both are painful to retrofit.

Take a deposit on every project until a client has paid you twice. A 50% deposit is normal for a new relationship, and the client who refuses one has told you something useful for free.

Record hours against every project from the first one, even on fixed-price work. You are not doing this to bill hourly. You are doing it so that in three months you can divide fee by hours and find out which kinds of work actually pay, which is the only way a beginner learns pricing quickly.

You don't need to buy software for this in month one. Bonsai's cheapest tier is $15 per user per month, or $9 on annual billing, and against your first three invoices that is real money. A free plan that covers contracts, invoices and time tracking is the honest starting point, which is the argument set out in the Bonsai comparison. Buy the paid tier when bank feeds and reporting start saving you an hour a month, not before.

The wider sequence of financial milestones runs past this: first repeat client, the month your income covers fixed costs, the point you need a tax reserve system. The guide to going freelance organised by money milestones lays them out in order. And when you're ready to widen beyond your own list, the checked list of where writing work is actually posted has what's live and what it costs.

Your first hour and your first month

In the next 30 minutes. Answer the three niche questions in writing. Not in your head — in a document, with the actual answers. Most people who stall at the start stall here, and the stall is invisible until you try to write it down.

This month. Publish the teardown, build the list of fifty, and send twenty notes. If you do only one of those, do the list. Samples without a list is a portfolio nobody sees; a list without samples still produces conversations.

Worklyn's free plan covers unlimited clients, proposals, contracts, invoices and time tracking at €0 a month, which is enough to run your first year properly and find out what your work actually earns per hour. The relevant starting point is the freelancer setup.