Do Upwork Jobs Actually Hire? An Outcome Analysis of 617,336 Resolved Postings
Published July 7, 2026
Published July 2026. Based on continuous monitoring of publicly visible Upwork job postings since July 2025: 1,506,251 postings from 588,474 client accounts as of July 3, 2026. Outcome numbers use the stable cohort: postings published October 2025 through February 2026 whose outcome has resolved, n = 617,336. Full method: methodology.
Every proposal a freelancer sends rides on two bets. The visible one: will this client pick me? The invisible one, the one almost nobody prices in: will this client pick anyone? Freelancer forums have argued that second question for years, armed with anecdotes (“half these jobs are fake”) and with published statistics that disagree so badly (22.8%? 38%?) they might as well describe different marketplaces.
We follow postings until they reach a final state, which means we can answer the question directly instead of guessing from snapshots.
The short answer: 43.9% of resolved Upwork job postings end in a hire. The other 56.1% never hire anyone: 41.3% close unfilled, 8.7% are cancelled by the client, and 6.1% expire. For a freelancer the practical reading is blunt: slightly more than half of resolved postings burned every connect spent on them.
What actually happens to an Upwork job posting
Most published Upwork statistics count postings when they appear. Ours follow each posting through its life until it lands in one of four final states:
| Outcome | Share of resolved postings | What it means |
|---|---|---|
| Filled | 43.9% | At least one freelancer was hired |
| Unfilled | 41.3% | The posting closed without any hire |
| Cancelled | 8.7% | The client withdrew the posting |
| Expired | 6.1% | The posting aged out without resolution |
Stable cohort: 617,336 resolved postings published October 2025 through February 2026.
The most useful mental model this table supports: a posted job is not a job. It is a coin flip weighted slightly against anyone being hired at all, before your proposal is ever read.
Why you have read 22.8%. And 38%. And now 43.9%.
Three hire rates for the same marketplace circulate publicly, and they disagree by more than 20 points:
| Source | Published number | What it measures |
|---|---|---|
| GigUp free dashboard | 22.8% | Roughly 69,000 postings from the last month, measured immediately |
| Vollna 2024 annual report | 38% | More than 5 million projects over a two-year window, settled and unsettled mixed |
| This analysis | 43.9% | Resolved postings only, from cohorts old enough to have settled |
All three are computed from real data. The disagreement is about when you look.
Hiring outcomes take time to settle. Postings that hire tend to close within days (next section), while postings that never hire linger open for weeks before closing, being cancelled, or expiring. That lag creates two opposite distortions:
- Measure a young pool and count everything, unsettled postings included, and the rate reads far too low, because most outcomes have not settled at observation time. That is the 22.8%.
- Measure a young pool and count only what has settled, and the rate reads far too high, because fast, successful closures dominate the settled subset. Our own April 2026 postings, measured in early July, show an artificial 55.8% fill rate for exactly this reason. We publish that number as a warning label, not a market fact.
- Average across a long mixed window and you land somewhere in between. That is the 38% (36% in their 2023 edition).
The stable answer requires patience: cohorts old enough, roughly four months, that outcomes have fully resolved. That is the 43.9%.
Worth saying plainly: the three sources agree on the shape of the market even while disagreeing on the level. GigUp’s category ranking largely matches ours, with translation at the top and sales, legal and accounting in the bottom cluster. Vollna also finds translation the top category (54% against our 51.3%) and entry-level postings filling most often (42% against our 47.5%). The shape is real. The level depends on when you measure, and most published levels are measured too early.
If a job is going to hire, it hires fast
Time from posting to hire, for postings published in April 2026, our first cohort with reliable closure timing (n = 15,970 fills):
| Time to fill | Share of fills |
|---|---|
| Within 24 hours | 14% |
| 1 to 3 days | 14% |
| 3 to 7 days | 47% |
| 7 to 14 days | 14% |
| 14 to 30 days | 10% |
Roughly 75% of all fills happen within 7 days of posting, and about 90% within 14 days. (Closure detection runs on a roughly two-hour cycle, so figures are accurate at day granularity.)
Two consequences follow. For freelancers: hiring happens at the front of a posting’s life. Nine out of ten postings that will ever hire have already done so two weeks in, so fresh postings are where the action is. For anyone reading market statistics: this asymmetry, fast fills and slow failures, is exactly the mechanism that makes early-measured hire rates unreliable in both directions.
What moves the odds
The 43.9% baseline hides wide, consistent spreads. Every cut below uses the same stable cohort.
Category: a 16-point spread
| Category | Fill % | Resolved postings |
|---|---|---|
| Translation | 51.3 | 11,275 |
| Design & Creative | 48.8 | 159,114 |
| Engineering & Architecture | 48.3 | 24,414 |
| Writing | 46.2 | 21,272 |
| Data Science & Analytics | 45.1 | 19,569 |
| Admin Support | 45.0 | 54,757 |
| Web, Mobile & Software Dev | 44.5 | 146,707 |
| IT & Networking | 40.6 | 13,941 |
| Accounting & Consulting | 40.4 | 25,905 |
| Legal | 38.1 | 7,440 |
| Sales & Marketing | 36.4 | 121,040 |
| Customer Service | 35.2 | 11,902 |
Translation postings hire at 51.3%, customer service at 35.2%. The biggest categories split: design and creative (159,114 resolved postings) fills at 48.8%, while sales and marketing (121,040) fills at 36.4%, near the bottom. Web, mobile and software development sits almost exactly on the baseline at 44.5%.
Experience level: expert-tier postings hire least
Entry level 47.5% (48,085 resolved postings), intermediate 45.4% (392,922), expert 39.7% (176,329). Postings that ask for expert talent are the least likely to hire anyone, almost 8 points below entry level. Pickier clients, sticker shock at expert rates, and exploratory postings aimed at expensive skills are all plausible mechanisms; we publish the number, not a cause.
Fixed price beats hourly by 10 points
Fixed-price postings fill at 50.5% (291,487 cohort postings) against 40.6% for hourly (253,888). A further 123,159 cohort postings (18.4%) carry no recorded budget type in our data; those fill at 35.1%. The gap fits a simple story: a scoped deliverable with a price on it usually means the client knows what they want.
The strongest green flag: the client has spent money before
| Client’s prior spend | Cohort postings | Fill % |
|---|---|---|
| Nothing yet | 152,032 | 36.0 |
| Under $1K | 140,618 | 44.0 |
| $1K to $10K | 178,985 | 47.3 |
| $10K+ | 196,899 | 46.8 |
The jump happens at the first dollar. Clients who have never spent on the platform fill at 36.0%; clients with any spending history fill at 44 to 47%. Roughly 11 points, about a quarter of the baseline rate, separates never-spent clients from everyone else. This is the posting-level echo of our client-quality analysis: spending history, not wording or account polish, is the signal that predicts outcomes.
Geography: a 20-point spread among the top client countries
| Client country | Cohort postings | Fill % |
|---|---|---|
| United States | 295,583 | 47.5 |
| United Kingdom | 55,069 | 44.4 |
| Australia | 38,173 | 46.6 |
| Canada | 34,369 | 43.4 |
| India | 25,764 | 26.7 |
| Netherlands | 14,334 | 37.0 |
| Germany | 14,271 | 42.2 |
| United Arab Emirates | 13,601 | 39.7 |
| Pakistan | 12,750 | 41.4 |
| France | 9,957 | 38.3 |
Top 10 client countries by posting volume.
The standout is India: postings from India-based clients fill at 26.7%, barely more than half the United States rate of 47.5%. The English-speaking markets cluster in the mid-40s; continental Europe sits in the high 30s to low 40s.
The tactical read
Every signal above is visible on a posting before you spend a connect:
- Check the client’s spend history first. Any prior spend moves a posting from the 36% pool into the 44-47% pool.
- Prefer fresh postings. Three quarters of all hires happen in the first week.
- Fixed price with a stated budget outperforms hourly by 10 points, and postings carrying no budget information at all fill worst of the three groups.
- Budget for the base rate. Roughly 56% of resolved postings hire nobody. A connects budget that assumes every posting is real is planning around the best case, not the base rate.
One warning against over-multiplying: these are one-dimensional cuts of the same cohort, and the dimensions overlap (client geography skews category mix, expert postings skew hourly, and so on). Read them as observed associations, not a scoring formula.
Caveats
- Fill rates are shares of resolved postings. Outcome labels are only as current as our last observation of a posting, and a posting can hire after unusually long delays.
- Time-to-fill is published only for April 2026 onward. Earlier closure timestamps in our data are retroactive and unusable for duration analysis.
- Each table is a one-dimensional cut; the dimensions overlap, and none of the gaps should be read as causal.
- Budget type is unrecorded on 18.4% of cohort postings, disclosed above rather than silently dropped.
- These figures describe the postings we monitor and may differ from Upwork’s internal accounting. We publish absolute counts and shares, never coverage claims.
Bottom line
A bit more than two in five resolved Upwork postings hire someone; the rest go unfilled, get cancelled, or expire. The lower numbers you have read elsewhere (22.8%, 38%) measure the same marketplace earlier in the settling process, not a different reality. When a posting does hire, it hires fast: three quarters of fills land inside a week. And the odds shift visibly with signals you can read before spending a connect: the client’s spending history above all, then pricing model, category, experience tier, and client geography. The 56% of postings that never hire anyone are not spread evenly; they cluster along exactly these lines.
Data: continuous monitoring of publicly visible Upwork job postings, July 2025 to present. See the methodology page for outcome definitions, cohort rules, and the list of statistics we deliberately do not publish.
Niche reports are coming
Per-niche deep dives: demand volume, real rates, fill rate, time-to-fill, and client quality for the niche you work in. Join the waitlist and tell us your niche; the most-requested niches get built first.