Monthly Report · March 202625 min read

Where agencies wasted their connects on Upwork in March

We analyzed every job posted on Upwork in March 2026 — 173,321 of them — plus the agency proposals that chased them. One category replied 2.7× more often than the busiest one. Most agencies aren't pointed at it.

Vadym Ovcharenko
Vadym Ovcharenko
Founder, GigRadar · April 26, 2026
The headline number
2.7×
reply-rate spread

Engineering & Architecture replied at 10.9%. Web, Mobile & Software Dev (the busiest category) replied at 4.0%.

173K
jobs · +4.5% MoM
17.0%
view rate
5.3%
reply rate
Video walkthrough

The whole report in under three minutes

TL;DR: March 2026
  • 173,321 jobs were posted on Upwork in March 2026, up 4.5% from February.
  • • Market view rate 17.0%, reply rate 5.3%. Both below 6-month average.
  • Engineering replied at 10.9%. Web Dev replied at 4.0%. 2.7× spread.
  • • Web Dev took 58.5% of agency attention; Engineering took 0.6%.
  • • Fastest-growing skill: Artificial Intelligence (+18.1%).
  • • Hourly ceiling: $15–34/hr in IT & Networking; floor at $5–11.

The Upwork funnel in one picture

Figure 1

The Upwork funnel in March 2026

Every Upwork proposal lives inside a funnel. The client posts a job. Some agency submits a proposal. The client may or may not open it. If they open it, they may or may not reply.

Figure 1Aggregate Upwork proposal funnel.

17% view rate sounds low until you remember that's after the proposal made it past Upwork's relevance filter into a client's inbox. 5.3% reply rate sounds low until you remember a meaningful share of Upwork "applications" are weekend punters firing off form letters. The harder question is: which slice of that funnel were your proposals in?

Contrarian read

What other Upwork reports get wrong

Every annual freelance market report you've seen leans on the same three narratives. The data tells a different story.

✕ Myth

The freelance market is shrinking.

✓ The data

March posted 173K Upwork jobs, up 4.5% MoM and broadly recovering across 9 of 12 categories. The market isn't shrinking; agencies are crowding the wrong slice of it.

✕ Myth

If your reply rate is below 5%, your proposals are bad.

✓ The data

Web Dev averages 4.0% reply rate even for top agencies. Reply rate is mostly a function of category supply-demand, not pitch quality. Move category before you rewrite proposals.

✕ Myth

Higher hourly rate means harder to win.

✓ The data

Engineering's reply rate (10.9%) is 2.7× Web Dev's, but its hourly midpoint is half. Specialized demand wins more conversations than generalist hustle, regardless of price.

Deep dive: cover letters

What actually moves reply rate inside the cover letter

We pulled 70,260 cover letters from February 2026 (chosen over March to avoid chat-open lag), computed 19 structural features, and grouped them into reply-rate buckets. Two findings invert the conventional cold-email playbook.

Figure 2

Speed beats every other lever — bid in the first 15 minutes

Figure 2Reply rate by time-to-bid bucket. Green = above baseline. Rose = below.

Bidding within 15 minutes of a job posting nearly doubles your reply rate vs bidding the next day (6.49% vs 3.41%, 3.08pp spread, p<0.001). The effect holds in every category. In Web/Mobile/Software Dev specifically: 5.15% under 15min crashes to 2.48% after a day.

Figure 3

The U-curve: short letters win, the 150-word sweet spot is wrong

Figure 3Reply rate by cover-letter word count. The valley sits at 60 to 170 words.

Cold-email research reports a 50 to 125 word sweet spot with a bell shape. On Upwork the curve is U-shaped, not bell-shaped. Sub-40-word letters reply at 7.29%; the dead zone sits at 60 to 170 words. Either get in and out in two sentences, or write a real 220+ word proposal. The middle ground hurts.

Figure 4

Combined effect in Web/Mobile: fast + short = 3.7× the reply rate

Figure 4Time-to-bid × word count, in Web/Mobile/Software Dev (the largest segment, n=41K).

A bid placed in the first 15 minutes with a sub-60-word note converts at 7.04% against a 4.6% category baseline. The same agency, writing a 250+ word letter and waiting 6 hours, converts at 1.91%. The compound lift is the cheapest 50% reply-rate gain in the dataset.

Figure 5

Buzzwords cost replies — most painfully in Data Science

Figure 5Reply rate by buzzword count. Tracking: AI, leverage, synergy, world-class, 10x, ROI, revolutionary.

Each additional buzzword shaves measurable points off your reply rate (overall spread: 2.0pp). In Data Science it's brutal: 0 buzzwords = 7.13% reply rate, 4+ = 3.89% (-3.24pp). Sales & Marketing is the only category where buzzwords don't hurt.

Figure 6

Links by category: hurt in Design, IT, Web; help in Sales, Accounting

Figure 6Net reply-rate change when a URL is included in the cover letter.

Pasting a portfolio link is net-negative overall (-1.09pp), concentrated in tactile/visual categories where the client wants a custom answer. Links HELP in service categories where credentialing is the bottleneck (Accounting +2.45pp, Sales/Marketing +1.19pp).

What did NOT move the needle, stop optimizing
  • Greeting style. "Hi [Name]" did not measurably beat "Hello there". 0.67pp spread.
  • Question count. Asking 1 question = 6.11%, asking 3+ = 5.36%. Under 1pp.
  • Emoji. With emoji 5.59%, without 5.91%.
  • "You-focused" pronoun ratio. Heavy-you letters underperform balanced.
  • Surface keyword overlap. Stuffing job-description keywords. Flat across deciles.
  • First-line length. All hook lengths within 0.6pp.

Sample: 70,260 GigRadar agency cover letters submitted in February 2026. Buckets with n<500 dropped. No verbatim cover-letter text used; structural features only.

Where the eyeballs go

Figure 7

View rate by category: Engineering opens 2.9× more often than Web Dev

If reply rate is whether the client wanted to talk, view rate is whether the client even noticed. The category divide is the most-cited finding in this report:

Figure 7Share of submitted proposals that were opened by the client. March 2026, by category.
"In Engineering, four in ten proposals get opened. In Web Dev, one in seven."

This isn't a quality argument: across these 12 categories the proposal pool is pulled from the same agencies, often the same writers. It's a scarcity argument. Engineering clients are hiring something specialized; the proposal stack on their job is shorter; their attention budget per proposal is bigger. Web Dev clients are drowning in applicants and skim accordingly.

Figure 8

View rate by category over six months

Figure 8Hover any sparkline to see exact monthly values. Yellow dot = March 2026.

View rate hasn't reset, it's hardened. Engineering & Architecture held above 35% for the entire six-month window, peaking at 39.1% in March. Writing stayed above 28%. Web, Mobile & Software Dev compressed from 14.7% in October to 13.6% in March, the lowest line in the grid. Customer Service, Admin Support, and Sales & Marketing all sit in a tight 19 to 22% band: clients still notice the proposal, but the contract is commodity-priced underneath.

From the GigRadar Academy

Diagnosing a low view or reply rate inside GigRadar

Where the conversations start

Figure 9

Reply rate by category: same divide, sharper edge

Figure 9Reply rate (chat opened) by category, March 2026.

The shape rhymes with view rate but the spread is sharper. Engineering replies 2.7× more often than Web Dev. Writing has the second-highest reply rate at 9.3%. The remaining clients are quality-sensitive, not price-sensitive, because the bottom of the market has already left.

Figure 10

Reply rate by category over six months

Figure 10Reply rate is compressing across nearly every category. Yellow dot = March 2026.

Reply rate has compressed almost everywhere, but the order hasn't moved. Engineering & Architecture printed 10.9% in March, Writing 9.3%, Accounting & Consulting 8.8%, IT & Networking 8.0%, Sales & Marketing 7.7%. The market average sits at 5.3%, but only three categories underperform it: Web, Mobile & Software Dev at 4.0%, Data Science at 4.9%, and Legal at 2.6%. If your team is anchored to the platform average and bidding into Web Dev, you're below the floor of every category that isn't Web Dev.

The saturation paradox

Figure 11

The saturation paradox: where the agencies aren't, the eyeballs are

Figure 11View rate vs reply rate vs share of agency attention. Categories sorted by view rate.

Web Dev pulls 58.5% of all agency attention in March and converts it to a 13.6% view rate. Engineering pulls 0.6% and converts at 39.1%.

Put a number on it: at Upwork's typical $0.30/connect and a $200/month connect budget, an agency that points 60% of its bids at a 4% reply-rate lane is buying ~$120/month of impressions where 96 in 100 land unread. Pointed at a 10%+ lane, the same budget converts ~2.5× more first replies. That's the cost of fighting in the wrong arena.

Reply rate heatmap by skill and budget

Figure 12

Reply rate heatmap: top skills by budget tier

A reply rate is meaningful only against its budget tier. Here are the 15 highest-volume skills on Upwork in March, color-coded by reply rate and tagged with the budget bucket they typically sit in.

Figure 12The 15 highest-volume skills, color-coded by reply rate intensity. Darker = better.

Three patterns repeat across the heatmap. Quick-fix briefs under $80 convert at 7 to 10% reply rate when scope is tight (Video Editing leads at 10.4%). Small-project budgets between $80 and $119 are the sweet spot for retainers — Social Media Marketing replies at 9.5%, Graphic Design at 4.1%. Standard projects between $120 and $199 are where Adobe Premiere Pro and Explainer Video clear 10%. Cells above $200 with reply rates under 5% are the trap quadrant: high effort, low conversion, mostly Web Dev sub-skills like Full Stack Development (3.6%) and JavaScript (3.7%).

Budget ranges by category

Figure 13

Fixed-price budget ranges by category

Figure 13P25–P75 interquartile range. Yellow tick = median. The spread tells you where deal size is concentrated.

Categories with tight IQR bands (Translation, Customer Service) are commodity-priced. Wide bands (IT & Networking, Web Dev) mean the same category contains both $200 fix-its and $5K+ projects. Position toward the high end if you can.

Patterns in client stats

Figure 14

Client quality patterns by category

Reply rate is half the picture. The client on the other end determines whether the conversation is worth your time. Here's what March's clients looked like by category: bankroll, verification, hiring history.

Figure 14Payment-verified rate, average lifetime spend, average client feedback, and average client hire rate. By category, March 2026.

Engineering & Architecture and Accounting & Consulting clients pay-verify above 80% and average lifetime spend over $30K — high-bankroll, high-trust accounts. Customer Service clients verify at one of the highest rates in the dataset, but average lifetime spend stays under $5K: reliable, just not big-ticket. Design, Sales, and Web Dev clients sit in the middle on every metric. Legal is the outlier at 2.6% reply rate with otherwise solid client stats — meaning your proposals reach qualified buyers who simply don't write back.

Hourly vs fixed price mix

Figure 15

Hourly vs fixed-price mix by category

Figure 15The structure of how each category is contracted. Hourly = retainer-friendly; fixed = scope-defined.

Customer Service (74% hourly) is structured for retainers. Design and Writing (~50% fixed) reward productized packages. The contract type isn't a preference: it's encoded in the category.

Volume sweep

Figure 16

Job posting volume by category over six months

Figure 16Where the market is actually growing. 9 of 12 categories posted positive MoM in March.

Nine of twelve categories posted positive month over month in March. Design & Creative recovered to 46.9K jobs (+7.6% MoM). Web, Mobile & Software Dev printed 37.7K, still the largest single category but down from October. Engineering & Architecture is the standout — 8.2K jobs in March, up roughly 12% for the second consecutive month. Writing, IT & Networking, and Admin Support each grew quietly. Translation and Legal are the only categories that printed negative MoM in this window.

Skills on the move

Figure 17

Fastest-growing skills, MoM

Figure 17Top 10 skills by job-volume change, March vs February 2026.

The fastest-growing skills cluster around video and AI work. Artificial Intelligence leads at +18.1% MoM (2,654 jobs), but reply rate sits at only 3.8% — supply has flooded ahead of demand. Explainer Video (+17.2%, 11.2% reply rate) and Telemarketing (+13.1%, 10.4% reply rate) are the rare cells where supply is rising AND clients are still talking back. Video Post-Editing (+16.8%, 10.2% reply), Adobe Premiere Pro (+12.4%, 9.9% reply), Motion Graphics (+11.3%, 8.3% reply), and Illustration (+11.1%, 7.5% reply) round out the growing-and-converting quadrant.

Figure 18

Fastest-declining skills, MoM

Figure 18Top 10 skills by job-volume drop, March vs February 2026. Note: Pay Per Click Advertising shows -100% but this reflects an Upwork skill-taxonomy change (the active cohort moved to "PPC Campaign Setup & Management," which is up 3% MoM). Real PPC demand is stable.

The Pay Per Click Advertising row showing -100% is a taxonomy artifact: the active jobs in this niche moved to a renamed skill ("PPC Campaign Setup & Management") that is up 3% MoM. Real demand for paid acquisition is stable. The genuine declines cluster in the Web Dev sub-tree — Full Stack Development -9.7%, CSS -5.0%, PHP -4.5%, Front-End Development -2.9%, JavaScript -1.7% — all carrying reply rates under 4%. The shrinkage is a category mix-shift away from generalist web work toward narrower specializations.

Hourly rates

Figure 19

Hourly rate ranges by category

Figure 19Median min–max hourly rate from posted jobs. Yellow dot = midpoint estimate.

The hourly ceiling correlates with category specialization, not with headline reply rate. IT & Networking and Web, Mobile & Software Dev both range up to roughly $30/hr at the high end, but their floors sit between $5 and $11. Engineering & Architecture and Legal price tighter, reflecting commodity pressure on simpler tasks within otherwise specialized lanes. Customer Service, Admin Support, and Translation cluster in the $5 to $15 band — the commodity floor of the platform. If your average billable rate sits inside that floor, you're competing on price by default.

Read next

Go deeper on the categories you choose to fight in

Synthesis

Three plays for your team this week

01

Run your March numbers against this report

Pull every proposal you submitted in March, group by category, compute view rate and reply rate. Compare against figures 2 and 3. The categories where your team beats the chart are structural advantages: pour bids in. Where you underperform, the problem is positioning, not pitch.

02

Add one Engineering scanner this week

Even with zero engineering specialists in-house, set up a narrow Engineering & Architecture scanner with a $20/hr+ floor. Route matches to your most senior bidder. Reply rate will be 2.5–3× your Web Dev average. Cost: ~30 connects/week. Upside: your first close pays for the year.

03

Cut one declining-skill lane

Pick the worst-performing skill in your March mix: lowest reply rate AND steepest MoM decline. Stop bidding it for 4 weeks. Route the recovered connects to one of the growing-skill lanes from figure 8. Explainer Video and Telemarketing are the highest-yield candidates by reply rate.

Methodology

Job-volume data. 13M+ Upwork job postings collected via public crawl since 2023, deduplicated and filtered to the calendar month.

View rate and reply rate. Computed from the GigRadar customer-agency proposal pool using the same definitions Upwork uses internally for "viewed" (proposal opened) and "replied" (chat thread opened). Outbound proposals only.

Recent-month bias. Reply rate for the most recent month is typically biased ~1pp low and continues to climb for 30 to 60 days as late chats arrive. View rate is more stable.

Vadym Ovcharenko
About the author

Vadym Ovcharenko

Founder of GigRadar. Spent 8 years building Upwork agencies before turning the bidder pipeline into a product. Writes monthly market reports from the same data 2,000+ agencies use to scan Upwork.

Stop bidding into the wrong category.

GigRadar's AI bidder watches 173K+ Upwork jobs per month and scores every match by reply rate, not just keyword fit. Book a demo to see your team's category mix audited against this report's data.