If you have been on a freelance marketplace for a few years, you have probably had the same experience: the same profile, the same work, the same proposals, and steadily worse results. It is easy to read that as a personal failing. It is not. Three specific things changed, and none of them are about you.
What actually changed
1. Supply grew much faster than demand
The number of people offering services online rose sharply, driven by remote work, currency arbitrage and a decade of "start freelancing" content. Demand grew too, but nothing like as fast.
In an auction, price is set at the margin: by whoever is cheapest and available, not by the average. So even if most freelancers hold their rates, a small number of new entrants pricing at the bottom is enough to reset what a buyer expects to pay for the whole category.
This is the part people describe as "race to the bottom". It is more precise to say that in a market with low switching costs and near-infinite supply, the buyer's reference price is set by the cheapest credible option they can see.
2. AI removed the last barrier to entry on proposals
Writing a decent proposal used to take fifteen minutes and cost something. That cost was doing real work: it separated people who cared from people spraying applications.
That signal is gone. A competent, personalised-sounding proposal now takes seconds, so a job that once drew twelve applications draws sixty, and the median quality of those sixty is higher than the median of the twelve.
The consequence is counter-intuitive and it matters: effort no longer signals seriousness, because effort is no longer expensive. Anything a buyer can no longer use to tell candidates apart stops being a differentiator, and the buyer falls back on the things they can still read quickly. Price, reviews, and whether you look like a specialist in exactly their problem.
3. Platforms started selling visibility
Every marketplace eventually monetises attention, because attention is the scarce asset it owns. Connects, boosted proposals, promoted gigs, paid membership tiers: the mechanisms differ and the effect is the same. Ranking stops being purely a function of quality and becomes partly a function of spend.
That is not a conspiracy, it is the business model working as intended. But it changes what you are competing on, and it means "get better at the work" is no longer sufficient to get seen.
Why the usual advice does not fix it
You will be told to improve your profile, write better proposals, apply faster, lower your rate to build reviews, and niche down.
Four of those five are optimisations inside an auction where the number of bidders keeps rising. They can move you a few places up a list of sixty. They cannot change the structure.
The fifth one, niching down, is genuinely correct, and it is usually explained badly. On a marketplace a niche helps because it is the one thing a buyer can still read at a glance when every proposal is well written. Off a marketplace, a niche is not a marketing preference. It is the thing that makes the alternative possible at all, and here is why.
The alternative, and why niching is the hinge
The alternative to applying for posted jobs is approaching businesses directly. That works only if you can answer one question precisely:
Which specific kind of business do you serve?
Not "small businesses". Not "startups". Something you could use to build a list: independent dental practices with two or more locations. Roofing contractors in the Southeast. Shopify stores selling physical goods in the outdoor category. Boutique hotels in coastal towns.
The reason this is the hinge is mechanical. If your answer is "anyone who needs a website", there is no list to build, no shared context to write about, and nothing to say in a first email that is not generic. If your answer is specific, then a list exists, the businesses on it share a problem, and you can write one genuinely relevant email that works for all of them with small variations.
Narrow does not mean small. There are tens of thousands of dental practices in most large countries. You need a handful of clients.
The method, in five steps
This is the entire thing. It is not complicated and it is not fast to set up.
Step 1: Define the target precisely
Write one sentence: business type, size signal, geography. If you cannot imagine searching for it, it is not specific enough yet.
The best input is your own history. Look at the clients you enjoyed and made money from, and find what they had in common that was not personality. That commonality is your niche, and it is more reliable than any market research you could do.
Step 2: Build a live list
Sources that are current, in rough order of usefulness for local and service businesses:
- Live map and directory data. Businesses maintain these themselves, because customers cannot find them otherwise, which is why they stay accurate.
- Trade association and chamber member lists. Pre-qualified by construction, since membership costs money.
- Review platforms. Current, and they tell you what customers complain about, which is an opening line.
- Official company registers, where the country has a good one.
What each of those looks like depends on where you sell, and it varies more than people expect. Our country-by-country directory guide covers the USA, the UK, Australia, Canada and the UAE.
What you should not do is buy a list. The reasons in full here, but the short version is that a purchased list is old, resold, and full of addresses that will damage your ability to send email to anyone at all.
Step 3: Get a contact, and verify it
A business name and a website are not a lead. You need a reachable person, and you need to know the address works before you send to it.
This matters more than it sounds. Sending to dead addresses raises your bounce rate, and a high bounce rate is one of the fastest ways to have your mail filtered for every recipient, including the ones who would have replied. Bounce rates and what is normal.
Step 4: Write something only you could have written
The first line is the whole email. If it could be pasted into a message to any other business, it will be read as a template, because it is one.
You do not need deep research. You need one true, specific observation: they have three locations but one booking page, their reviews mention the same complaint twice, they are hiring for a role that implies the problem you solve. What personalisation at scale actually looks like.
Step 5: Follow up, five times, politely
Most replies do not come from the first email. They come from the third or the fourth, because the first arrived on a bad day.
This is the step people skip, and skipping it is why they conclude cold outreach does not work. A follow-up sequence that does not annoy people.
What the numbers look like
Being straight about this, because the internet is full of implausible figures.
For a well-targeted list, verified addresses and a genuinely specific message, a reasonable expectation is a reply rate in the mid single digits to low teens, and a positive-reply rate meaningfully lower than that. On 300 businesses that is a handful of real conversations, and for most independent sellers a handful of real conversations is a good month.
Compare that with a marketplace: sixty applicants per job, one winner, and the client is not yours afterwards. Our benchmarks post has the fuller picture.
Should you leave the platforms entirely?
No, and this is where a lot of "quit Upwork" advice goes wrong.
Marketplaces are good at one thing that direct outreach is bad at: producing demand that already exists, today, with the purchase decision already made. That is genuinely valuable when you need cashflow this month or you are testing whether an offer sells at all.
The sensible position is a portfolio. Keep one marketplace for cashflow and proof. Build direct outreach in the background so that in six months your pipeline does not depend on a ranking algorithm you cannot see. The full comparison of every channel is here.
What you should stop doing is treating a marketplace as a business. It is a lead source with a commission, and the commission is not the expensive part. The expensive part is that the client relationship, the thing that pays you for the next five years, belongs to somebody else.
Where Leads Ranger fits
Steps 2 to 5 are exactly the part where people quit, because doing them by hand is four hours of copying, pasting and forgetting.
Leads Ranger does them in one place: describe the kind of business you want in plain English and it harvests real companies from live map and directory data, finds and verifies contact details, scores which are worth your next hour, and runs the follow-up sequence across email, WhatsApp and LinkedIn from your own accounts. Replies land in one inbox on a built-in CRM, so nothing dies in a tab.
It will not make a vague offer sell, and it will not tell you who your niche is. Those are still yours. It removes the repetition, which is the actual reason most people never get past step one.