Favente
Brand-to-creator platform · product design
I designed Favente’s creator onboarding, guided content creation, and performance reporting flows. All three went live. Guided Create brought the steps into one project, and the performance record now distinguishes missing data from a real zero.
The work, annotated
- Five measures (reach, engagement, audience, conversions, output), each read through one typed contract.
- “How this is measured”: period, scope, coverage and source sit beside every figure.
- Stale data says so: “Out of date” with the last reading's age.
- Role
- Founding Technical GTM
- Timeframe
- Jun 2026 – Present
- Team
- Founding team member
- Tools / stack
- SvelteKit, TypeScript, Figma, Claude Code, Playwright
- Status
- Live: creator onboarding, guided content creation and performance reporting (merged in PRs #15, #76)
The problem
Favente is an early-stage platform that helps brands recruit their existing users as paid creators, then guide, review and pay for their content.
Two kinds of friction came up in review. Guided Create was really five separate generators: the creator's topic reached none of them, and the script never saw the accepted format. The performance views made results look worse than they were. Accounts were summed by shared date, so the headline “fell” whenever one account had not reported, and a radar chart scored creators against invented targets.
My role
I joined Favente as Founding Technical GTM. As a founding team member I designed and shipped the creator onboarding, guided content creation and performance reporting flows, which are now live. I ran click-by-click usability audits to cut extra steps, and redesigned navigation and the mobile layouts so the product felt consistent. For each change I wrote the implementation brief and tested the build for function, layout and usability before release.
Decisions & what changed
Live
Guided Create as one connected project
I turned five separate generators into one project. Each step now carries what came before it: the brief, the concept and the accepted hook flow into the script and on to Finish. The company’s brief shows as the company’s own guidance instead of being merged into the creator’s idea. On a 390 px phone the text column went from 143 px to 303 px.
Evidence · Merged 24 Aug 2026 · live, per the résumé (“guided content creation … all now live”)
Inspect the rejected option
Five standalone generators, each choosing its own subset of context.
(open the full screen)
Close-up · Script step, dark theme: the accepted hook carried into the script (synthetic data) (open the full screen)
Close-up · the hub, dark theme: the saved project waits at its next step (synthetic data) Live
A track record that never shows missing as zero
I replaced the radar chart with one row per measure. Its axes mixed different scopes and scored creators against made-up targets, so results looked worse than they were. Now each graphic follows the arithmetic, and “not connected”, “not tracked”, “not imported” and “building history” are all kept apart from a real zero. A first import counts as a snapshot, never as growth, and comparisons need a like-for-like window.
Evidence · Track record merged 10 Oct 2026 and the Performance page 11 Oct, after an Overview aggregation fix on 24 Aug · live, per the résumé
(open the full screen)
Before · a radar scored against fixed marks (same data) (open the full screen)
After · one result and five measures in their own units (same data) Same sample record in both. Before, the headline was 1.9M “tracked views on approved claims”: the lifetime sum of creator-submitted counts (1,938,000). After, it is 1.8M “views gained since readings began”: the 1,838,120 views actually observed across three connected accounts since 6 June 2026.
(open the full screen)
Close-up · the same record in the dark theme: “Not imported” and “3 untracked, not zero” (synthetic data) Proposal
Three directions compared before one was merged
Before committing to one design I built three comparable analytics directions (A, B and C) on the same synthetic data, each handling the awkward states honestly: snapshot vs growth, not reported vs zero, a partial day, stale data. The merged design, D, combines the strongest parts. The three explorations stayed in a dev-only lab.
Evidence · An unmerged design lab, “analytics design directions A/B/C”, on synthetic data
What was built
- Guided Create session flow: brief → concept → script → finish → produce.
- Creator Performance page and the Track Record on one five-measure contract.
- Implementation briefs and visual-QA targets (populated, first import, empty, dark, 360 px).
Outcomes
- Live
- Creator onboarding, guided content creation and performance reporting flows
- Self-reported · Updated résumé
- Merged
- Guided Create overhaul, Track Record and Performance page merged to main
- Verified · source noted · PRs #15, #76
