For years, Upwork worked in a frustrating but understandable way. You applied to dozens of jobs, most went nowhere, but a few clients opened your proposal, some replied, and a handful became calls. One or two turned into paid work.
It was never easy, but it made sense.
But in the last 12-18 months something changed.
No one even reads or opens your proposals anymore. It feels like shouting into a void.
Upwork has changed.
Upwork used to be a human inbox
In the old version of Upwork, a client posted a job. Freelancers applied. The client skimmed proposals, compared candidates, and replied to promising ones.
Competition was already fierce. Good jobs often had 50+ proposals. There were always cheap freelancers, agencies, spammy applications, and generic templates.
A freelancer with a strong profile, solid reviews, relevant work history, and a personalized proposal could still win jobs.
The proposal itself mattered because there was a reasonable chance a human would read it.
That assumption is now weaker.
The new Upwork model is closer to algorithmic shortlisting
Upwork does not simply show proposals in the order they arrive. Upwork's own explanation of why proposals are not viewed says that, when a post receives many proposals, Upwork shows the closest matches first. It points freelancers toward profile fit, skills, work history, Job Success Score, profile title, overview, and other match signals.
Clients also see preview information before they open the full proposal: your name, title, rate, badges or JSS where shown, and the beginning of your message.
That means "not viewed" does not necessarily mean your proposal was never surfaced anywhere.
It more likely means the client never opened the full proposal.
And that distinction matters.
Your proposal may have appeared somewhere in a client-side list, but the client may have looked only at the top few matches, a shortlist, or the summary view. If you do not survive that first ranking and preview stage, the quality of your full proposal is almost irrelevant.
In other words, Upwork is no longer just a writing game.
It is a ranking game, a preview game, and increasingly an AI-shortlisting game.
AI has made the proposal flood worse
Freelancers are not imagining the increase in polished-but-generic proposals.
Upwork has been investing heavily in Uma, its AI system. Upwork's Uma page explicitly pitches custom proposal drafts for freelancers, and Upwork reported in its Q1 2025 financial results that 52% more users engaged with Uma in Q1 2025 than in Q4 2024.
That has a predictable side effect.
It is now easier for a mediocre freelancer to send a proposal that looks decent at first glance. More people can apply faster. More people can personalize just enough to seem relevant. More people can submit polished applications without spending much time thinking.
So clients are not just seeing more proposals. They are seeing more proposals that look superficially competent.
That makes the client's job harder.
And when the client's job gets harder, they rely more on filters, summaries, ranking systems, AI tools, invites, and shortcuts.
Which means fewer full proposals get opened.
Clients increasingly do not need to read every proposal
Upwork has also been adding AI tools for clients.
Uma can help compare proposals, evaluate talent, recommend freelancers, create shortlists, and support the hiring process. Upwork's Uma page says clients can use Uma for side-by-side proposal comparison, and its Spring 2026 update says Uma Recruiter shortlisting is now available on the Basic plan, where it automatically identifies and surfaces relevant professionals for a project.
This is probably one of the biggest reasons freelancers feel invisible.
A client with 80 proposals no longer has to manually read 80 proposals. The platform is moving toward a model where the client sees recommendations, rankings, summaries, and shortlists.
That may be useful for clients.
But for freelancers, it changes the game completely.
A carefully written proposal is only useful if it gets past the first layer. If the system or the client's preview screen does not put you into the small set worth opening, your actual message may never be read.
There may also be fewer serious clients to compete for
Another thing has changed: the client side of the marketplace has been under pressure.
Upwork reported over 872,000 active clients in Q1 2024, 812,000 in Q1 2025, 796,000 in Q2 2025, and 785,000 at the end of 2025. Revenue did not collapse, but client count declined while spend per active client increased.
That suggests a marketplace with fewer active clients, but larger or higher-spending clients.
For freelancers, this can feel brutal. There may still be money on the platform, but not necessarily more accessible jobs. Good jobs attract huge competition. Small clients may be less active. Serious clients may rely more on invites, Talent Scouts, shortlists, or AI recommendations.
More freelancers are applying with AI-assisted proposals while fewer clients are active and more filtering happens automatically.
That combination explains why many good freelancers are seeing fewer proposal views.
Boosting does not solve the real problem
Many freelancers respond by spending more Connects and boosting proposals.
That is understandable. If no one reads your proposals, paying for visibility seems logical.
But boosting is not the same as being chosen.
Upwork says boosted proposals can appear in prominent slots, but they can lose that placement if you are outbid or after certain client interactions. The boost can get you into a better slot, but the client still decides whether to read your proposal.
Boosting won't guarantee a full read, fix a weak match, overcome a vague profile, or necessarily beat an AI shortlist.
Boosting is probably most useful when you are already an unusually strong fit for a job.
It is much less useful when the job is broad, generic, and flooded with applicants.
Lowering your rate can backfire
Another common reaction is to lower your hourly rate.
Sometimes that helps, especially if you were clearly outside the client's budget. But for experienced freelancers, lowering your rate can create a different problem.
You may move yourself into the same comparison bucket as cheaper, less experienced freelancers. Instead of looking like a specialist, you look like one more affordable option in a crowded list.
Upwork's own client-side advice warns that underpriced freelance work can signal inexperience or lower quality.
For many freelancers, especially skilled developers, designers, consultants, writers, marketers, and technical specialists, the answer is not to become cheaper.
The answer is to become easier to categorize as the right person for a specific problem.
The real problem: you are not being opened because you are not being shortlisted
This is the key point.
If your proposals are not being viewed, the problem may not be the body of the proposal.
The problem may be that you are not making it into the client's first serious consideration set.
That first set may be shaped by:
- Upwork's ranking algorithm
- profile keywords and category fit
- your title and first two proposal lines
- your rate compared to the client's expectations
- your work history in that specific niche
- badges, JSS, and reviews
- whether you were invited or applied cold
- whether the client uses AI shortlisting
- whether the job is already flooded with boosted proposals
By the time the client would read your carefully written explanation, the decision to ignore you may have already happened.
That is why writing "better proposals" is no longer enough.
What freelancers should do instead
Start with positioning.
A generic profile like "Full-stack developer," "Graphic designer," "Copywriter," "Marketing expert," or "Virtual assistant" is too broad. Even if you are excellent, broad categories are where competition is worst.
You need to be legible to both the algorithm and the client.
Examples:
Instead of "Full-stack developer," try "SaaS MVP rescue and rebuilds for React/Node apps."
Instead of "Copywriter," try "Landing page copy for B2B SaaS demos."
Instead of "Designer," try "Conversion-focused Webflow design for funded startups."
Instead of "Marketing consultant," try "Google Ads cleanup for ecommerce stores wasting budget."
Your goal is to make the client instantly understand when you are the obvious choice.
Optimize the first two lines, not just the full proposal
Clients often see the beginning of your proposal before they open the full thing. So the first two lines matter enormously.
Do not start with:
"Hi, my name is Alex and I am an experienced freelancer with 8 years of experience."
That wastes the most valuable space.
Start with a diagnosis, a relevant result, or proof that you understand the job.
For example:
"Your main risk is getting the permissions and data model right before the UI work starts."
Or:
"I have fixed this exact kind of slow Shopify store before. The issue is usually a mix of app bloat, theme code, and unoptimized product media."
Or:
"This looks more like a migration project. The dangerous part is preserving SEO, redirects, and analytics continuity."
Those lines give the client a reason to open the proposal.
Apply to fewer jobs, but with much stronger fit
The old strategy was often: apply to many relevant jobs and hope a few convert.
That strategy is getting worse.
Now, the better strategy is to be much more selective.
Avoid jobs where:
- the description is vague
- the client has no hiring history
- the budget is unrealistic
- the job already has 50+ proposals
- the client invited many freelancers
- the post looks AI-generated and generic
- the scope is so broad anyone could apply
- your fit is "pretty good" but not obvious
Prioritize jobs where:
- the problem is specific
- the client has a payment history
- you can diagnose the issue quickly
- your profile directly matches the job
- you have proof or past work in that exact area
- the client needs expertise, not just labor
- fewer freelancers can credibly claim the same fit
On modern Upwork, "I can do this" is weak.
"This is exactly the problem I solve" is much stronger.
Do not rely only on Upwork
This may be the most important practical takeaway.
Upwork can still work. But for many freelancers, it should no longer be the whole business.
The platform is becoming more filtered, more competitive, more AI-mediated, and more expensive to use. That makes it risky to depend entirely on cold proposals.
Freelancers with real skill should build channels where trust matters more than ranking:
- referrals from past clients
- LinkedIn posts and direct outreach
- niche landing pages
- partnerships with agencies
- local business networks
- industry communities
- following up with past leads
- content that proves expertise
- direct relationships with founders, teams, and businesses
Upwork is a marketplace. Marketplaces are useful, but they are also crowded and controlled by someone else's rules.
Your own network and reputation are harder to algorithmically bury.
The bottom line
If no one reads your Upwork proposals anymore, it does not automatically mean you are a bad freelancer, your work is weak, or your proposal writing is terrible.
It may mean the marketplace has changed.
Clients are overwhelmed. AI has increased proposal volume. Upwork is using more ranking and shortlisting systems. Boosting has made visibility more expensive. The number of active clients has declined. And many clients now make decisions before ever opening most full proposals.
So the question is no longer:
"How do I write a better proposal?"
The better question is:
"How do I become one of the few freelancers the client or the system decides is worth opening?"
That requires sharper positioning, stronger preview lines, better job selection, more specific expertise, and less dependence on Upwork as your only source of clients.
The painful truth is that many proposals are not being ignored after careful consideration.
They are never reaching careful consideration at all.
What I'm building
Delegate tasks. Get software.
Give Vroni a GitHub issue, bug report, spec, or rough idea. It reads the repo, plans the change, writes code, runs checks, and works toward a review-ready pull request.
Take a look at vroni.comI respect your privacy. Unsubscribe at any time.