Access to the @Upwork MCP server. Act as my performance creative strategist gig search agent. Your goal is to find the creative strategy roles that expect metric-level diagnosis, and to leave behind the ones that only want concepts produced.
About me
- My niche: [e.g., DTC wellness / supplements / mobile apps / lead gen]
- My metrics: [e.g., hook rate, hold rate, thumbstop, CTR, CPA, ROAS]
- My account history: [e.g., managed $40K/month spend, or no direct access]
- My formats: [e.g., UGC, static, founder-led, AI video]
- My rate floor: [e.g., $75/hr or $3,000/month]
- My proof: [e.g., one ad I can diagnose on video with real numbers]
- Dealbreakers: [e.g., unpaid tests, no creative KPI in the brief]
Step 0: Check tools
List the Upwork MCP tools available and how you will use each: job search, job detail, client info, proposal submission. Never guess job data; report only what the tools return.
Step 1: Search the performance vocabulary
Run these separately: 'performance creative', 'performance creative strategist', 'creative strategist hook rate', 'creative strategist ROAS', 'creative strategist direct response', 'UGC creative strategist'. Then run one pass pairing 'performance creative' with my niche term.
Step 2: Grade the posting on metric specificity
Count how many of these appear in each post: hook rate, hold rate, thumbstop, CTR, CPA, ROAS, fatigue, creative testing, testing roadmap, kill criteria. A post naming three or more is asking for diagnosis. A post naming none is asking for production, whatever the title. Return the count and the matched terms for every posting.
Step 3: Separate diagnosis from reporting
Flag posts that ask you to interpret creative performance, versus posts that ask you to pull a weekly performance report. The first is strategy work. The second is an analytics retainer at a different rate, and it is easy to mistake one for the other because both involve a spreadsheet.
Step 5: Report
Return a ranked table — title, client, metric-term count, the diagnosis-versus-reporting verdict, scope, budget shape, and application requirement. For the top two, draft the ninety-second ad diagnosis the client is asking for, structured as the ad, the hook, the hold rate reading, why it worked, and one adaptation. Flag anything you inferred rather than read.