Pay each card down before its statement-close date — that's the balance the bureaus see. Target under 30%, then under 10%.

Most software/AI tools are charged to your personal cards — they're deductible business expenses and a big reason the cards stay maxed. Reimburse them from the business or move them to a business card.

Supplemental, lumpy income — not a salary. The big hourly rates are real but infrequent for a design profile, so results come in spikes. Realistic combined: a few hundred dollars most months, with occasional bigger spikes. Most reliable: Respondent + Mercor. Don't pour energy into the low-probability ones — keep the prime focus on client work + the job search.

Update "Earned to date" whenever money lands — that's how we see what actually works and cut what doesn't.

What we learned (Jun 23): H1 (ultra-low proposals = best) was mostly wrong. The "under 10 proposals" cap traps you in low-budget junk, because good-paying jobs attract 10+ proposals fast. Also Upwork's "Product Design" category is physical/industrial design, not digital UX.

Revised recipe (H2): Category = Web & Mobile Design. Sort = Best Matches. Payment verified ON. Keyword = product designer / UI UX / Figma / SaaS / landing page. Do NOT cap proposals at 5; allow up to ~15. Then screen at the gate.

10-second gate (apply only if): real digital design work, payment verified, clear scope, healthy budget (hourly $40+ or fixed $500+), client has spend or a good rating, and it names your skills. Bonus: ongoing or retainer potential.

After ~20 applications, lean into the buckets with the best interview rate and drop the rest.

This is the running list so nothing gets lost between sessions. Tags: now = today/this week, soon = next few weeks, ongoing = recurring habit.