Should you spend the connects?

How the score is calculated

Five things you can read off the job page are each scored out of 10 against fixed bands, and the score you see is their average, clamped to the 1–10 range. Nothing is sent to a language model: the same deterministic PHP runs here and in the Chrome extension, which is why it costs nothing to use and gives the same answer twice.

The five are the client's hire rate, how many proposals the job already has, the client's history (their total spend, how many people they have hired and how long the account has existed), how many freelancers are currently being interviewed, and how many invites the client has sent.

A factor the job page does not show is left out of the average rather than counted as a zero. A brand-new client with no spend history is not the same as a client who has spent nothing across forty jobs, and scoring them identically would be the wrong answer rather than a cautious one.

The verdict is a band on the final number: 8 or above is apply, 5 to 7 is maybe, below 5 is skip.

What counts as a good hire rate

Hire rate is the single most useful number on an Upwork job page, and the one most freelancers skim past on their way to the budget.

It scores 10 at 90% and above, 8 from 80%, 6 from 60%, 4 from 40%, and 2 below that. The bands are steep on purpose. A client who hires on nine of every ten jobs they post is a client who posts when they intend to buy. One who has posted thirty times and hired four is running a market survey, and the budget on the listing tells you nothing about whether it will ever be paid to anyone.

The failure this is protecting you from is not losing to a better freelancer. It is the job being cancelled, reposted or quietly abandoned after you have spent the connects and written the proposal.

Why proposal count matters less than it feels like it does

Upwork shows competition in buckets — fewer than 5, 5 to 10, 10 to 15, 20 to 50, 50+ — and a high bucket is the thing that makes most people close the tab.

It is worth less attention than it gets. Proposal count tells you how many people clicked apply, not how many wrote something the client will read past the first line. On a job with fifty proposals and a client who hires reliably, the real competition is the handful of proposals that open with the client's actual problem.

That is why it is one factor of five here rather than the headline. A crowded job posted by a client who hires still scores better than an empty one posted by a client who does not, and that ordering is the useful part.

When a high score is still the wrong job

The score answers one question: is this client likely to hire someone. It does not answer whether they should hire you.

A job can score 9 and still be wrong for you — the budget is half your rate, the stack is one you have not touched, the timezone means standups at 5am, or the scope is written vaguely enough that it will double after you start. None of that is visible in the numbers, and none of it is something a scoring model should pretend to read.

So treat a high score as permission to spend the connects, not as a recommendation to bid. The check it replaces is the one nobody does properly: scrolling up to look at the client's history before writing four paragraphs about yourself.

What the scorer cannot see

It reads the numbers on the job page and nothing else. It does not open the client's review history, read the job description, or know anything about your profile.

That is a deliberate limit rather than a missing feature. Everything it scores is a published figure you could check yourself, which is what makes the result explainable — every factor comes back with the number it was scored on, so you can disagree with it.

The Chrome extension does more, because it can see the job description and your profile at the same time. On this page, what you get is the part that needs no account and no permissions.