Upwork Proposal Skill
Generate proposals that beat the AI flood and actually get responses.
The Problem We're Solving
Clients receive 50+ proposals that all sound the same: robotic, vague, and stuffed with "Dear Hiring Manager." They can smell generic AI instantly. This skill creates proposals that feel human, specific, and worth responding to.
Key Stats (2025-2026)
| Metric | Value |
|---|---|
| Average proposals per job | 20-50 |
| Reply rate | 8-30% |
| Win rate | ~5% (1 in 20) |
| AI+Human hybrid win rate | 48% higher than pure AI or pure manual |
| Optimal proposal length | 150-250 words |
| First-hour applications | 48% higher response rate |
Workflow
User pastes job → Extract requirements → Match to user skills → Generate proposal + shorter versionRequired Input
| Input | Description |
|---|---|
| job_description | The Upwork job posting (user pastes it) |
Optional Inputs (Ask if relevant)
| Input | When to Ask |
|---|---|
| relevant_experience | If job matches a specific past project |
| rate/pricing | If user wants to include specific pricing |
| availability | If job mentions urgency or timeline |
| special_angle | If user has unique insight into the problem |
The 7-Part Formula (24%+ Response Rate)
See references/proposal-structure.md for details. Quick reference:
| Part | Length | Purpose |
|---|---|---|
| Hook | 25 words | Client name + their specific problem |
| Relevance | 50-75 words | Similar experience with results |
| Solution | 50-75 words | Bullet-point deliverables |
| Proof | 30-50 words | 1-2 past projects with metrics |
| Timeline | 15-25 words | Specific dates, not vague |
| Investment | 10-15 words | Clear pricing if appropriate |
| CTA | 15-20 words | One specific next step |
Total: 195-285 words (under 300 is ideal)
Anti-AI-Slop Rules
See references/standout-strategies.md for full guide. Critical rules:
| DO | DON'T |
|---|---|
| Start with "Hey there" or use client's name | "Dear Hiring Manager" or "I hope this finds you well" |
| Mention 2 specifics from their job post | Generic "I'm the perfect fit for this role" |
| Use numbers: "23 similar projects", "$12K/month results" | Vague claims: "extensive experience" |
| Natural contractions: "I'll", "you're", "can't" | Robotic: "I would be delighted to" |
| One emoji max (👋 in greeting) | Multiple emojis or none at all |
| Specific timeline: "December 15th" | Vague: "as soon as possible" |
| Lead with solution approach | Lead with credentials |
Testing Instruction Detection
Clients often add tests like "Write HOWDY at the top" to filter spam. Always scan the job for:
- Specific words to include
- Questions to answer
- Hidden instructions in the middle/end of posting
If found, follow them FIRST.
Output Format
Always provide TWO versions (user frequently asks for shorter):
## Proposal (Ready to Submit)
[Full proposal following 7-part formula]
## Shorter Version
[Condensed 100-150 word version]
## One-Liner (For quick reference)
[Single sentence pitch]
## Notes
- [Relevant past projects to mention]
- [Questions to ask client]
- [Red flags if any]User's Style Preferences
From 12 Upwork conversations analyzed:
- Greeting: "Hey there 👋" or "Hi there 👋"
- Tone: Natural, humanized, conversational
- Structure: Solution-first, then experience
- Length: Always asks for shorter - start concise
- Honesty: Never oversells, mentions learning areas
- Tech mentions: n8n, Python, AI agents, Google Workspace
User's Skill Positioning
See references/user-profile.md. Quick reference:
Highlight these:
- n8n workflow automation
- AI agents / Agentic AI (PIAIC certified)
- Python, FastAPI
- API integrations
- Google Workspace automation
- WhatsApp Business API
Be honest about:
- React, Node.js (learning)
- Make/Zapier (concepts understood, not hands-on)
Engineering Methodology (Trust Signal)
User follows Spec-Driven Development (SDD) and Test-Driven Development (TDD).
When to mention this: Only for projects where quality/reliability matters:
- Complex automation with multiple integrations
- Long-term/ongoing projects
- Clients who mention "reliable", "production-ready", "maintainable"
- Higher-budget projects ($500+)
- Clients burned by previous freelancers
DON'T mention for: Simple one-off tasks, quick fixes, low-budget gigs.
What you deliver (in repo):
specs/- Requirements & design specs (per phase)tests/- Pytest test suites (written BEFORE code)docs/ADRs/- Architecture Decision Records- Self-documenting code with clear structure
How to phrase it naturally:
I follow spec-driven development — I'll document requirements first,
write tests, then implement. You'll get working code plus specs and
tests in the repo, not just "it works on my machine."Shorter version:
I write specs and tests first, so you get reliable, documented code.One-liner (when relevant):
All deliverables include specs, tests, and documentation in the repo.See references/engineering-methodology.md for full details.
Example Transformation
Generic AI slop:
Dear Hiring Manager,
I hope this message finds you well. I am writing to express my keen interest in your project. With my extensive experience in automation and workflow development, I am confident I can deliver exceptional results. I have worked on numerous similar projects and would be delighted to discuss further.Winning proposal:
Hey there 👋
I can help you automate your Google Sheets → AI → PDF → Email workflow in n8n.
Here's the plan:
- Connect Google Sheets with scheduled triggers (twice weekly)
- Pass data to ChatGPT/DeepSeek for processing
- Generate formatted PDF (via API or Google Docs conversion)
- Auto-send via Gmail
I recently built a similar workflow connecting Sheets with ChatGPT for automated reporting — happy to share details.
When would you like to kick off?Checklist Before Output
- Read ENTIRE job post (including hidden test instructions)
- Extracted 2+ specific requirements to address
- Under 250 words
- Starts with personalized hook (not "Dear Hiring Manager")
- Includes specific approach/solution
- Has concrete proof (numbers, past projects)
- Clear CTA at the end
- Shorter version provided
- Natural tone, contractions used
- No overselling or vague claims