← Back

Fave Work

A portfolio platform where one body of work composes many audience-specific portfolios.

Project overview

Type: Product · 8-week 0→1
Project type: 0→1 · Creator Platform · Multi-portfolio Composition · Conversion
Role: Lead Product Designer · Systems Design · Interaction Design · UX Research
Methods: JTBD interviews · audience-panel testing · usability testing · A/B
Tools: Otter.ai · Framer · Figma + Dev Mode · Amplitude · Statsig
Case thesis: Designing a portfolio platform where a multi-skilled professional builds many audience-specific portfolios from one library of work, each framed for its viewer, so reviewers see only the work relevant to them.

The context

Portfolio platforms like Behance and Dribbble assume one profile per person, where all work accumulates in a single stream. More professionals now hold more than one marketable skill, and the spread of AI is widening that further. A designer who also writes, builds sites, or makes content has to pour everything into one profile, where the mix reads as unfocused to an employer or client deciding whether this person does the specific thing they need.

The problem

A single mixed portfolio loses opportunities. In a reviewer test, a portfolio that mixed disciplines was rated a fit for a specific role only 34% of the time, against 72% for a portfolio focused on that role's work (behavioral, reviewer panel). Meanwhile 64% of surveyed professionals held more than one marketable skill but maintained a single portfolio (attitudinal), because tailoring meant building and maintaining separate sites, so they sent one diluted link and let the viewer dig for relevance.

The goal

Let a professional present only the work relevant to a given audience, framed for that audience, without maintaining duplicates, measured by tailored portfolios created and sent, reviewer fit and likelihood-to-contact, and retention, rather than by total work uploaded.

Empathize — Reviewers judged fit in seconds, and a mixed portfolio rated a fit for a role only a third as often as a focused one

In this section: Research foundation · Key insights

Research foundation (method)

Phase 1 — Creator interviews (n=20, ~40 min, multi-skilled designers, writers, and marketers, recruited via dscout, transcribed in Otter.ai): how they present work to different audiences.
Phase 2 — Reviewer panel (n=22 hiring managers and clients): rated focused versus mixed portfolios for a specific brief on fit and likelihood-to-contact.
Phase 3 — Survey (n=180; 17.5% response, 12.4% completion; select-all and single-select labeled per question): on skills held and portfolios maintained.
Phase 4 — Prototype pilot (Amplitude-instrumented, 60 creators): behavior composing and sending portfolios.

Key insights

1. Reviewers decide fit fast, and a mixed portfolio reads as unfocused. Hiring managers and clients scanned for evidence the person does the specific thing they need, and a portfolio spanning disciplines diluted that signal. Triangulation:

Behavioral: a role-focused portfolio rated a fit 72% of the time; the same person's mixed portfolio, 34%.
Verbatim — coded: Dilution: "If I'm hiring a product designer and half the work is illustration, I start wondering which one they actually are."

2. Professionals avoid tailoring because the cost is maintaining duplicates. Creators knew a focused portfolio would land better, but building a separate site per audience, and keeping each current, was more than they would do, so they sent one link for everything.

Attitudinal: 64% held more than one marketable skill while maintaining a single portfolio; the top reason was the upkeep of multiple.

3. A tailored portfolio needs its own framing, or it reads as a generic gallery. A relevant subset of work still needed a headline and a call to action aimed at that audience to feel purpose-built and prompt a response.

Verbatim — coded: Framing: "Even the right work, with no headline saying what it's for, just looks like a gallery and doesn't tell them what to do next."

Dashboard — Focus changes whether a reviewer responds

Focus changes whether a reviewer responds
Scope: reviewer panel (n=22) + survey (n=180)
Guiding question: Does a focused portfolio change a reviewer's response?

  Rated a fit for the role
    Mixed-discipline portfolio ......... 34%
    Role-focused portfolio ............. 72%

  Hold >1 skill but maintain 1 portfolio . 64%

Key Insight: Relevance to the viewer is the conversion lever, and the cost
of maintaining duplicates is what keeps professionals from using it.

Define — A professional had to present only the relevant work, framed for the viewer, without maintaining duplicate portfolios

In this section: POV · How Might We · Principles · Insight→decision map

POV statement. A multi-skilled professional needs to present only the work relevant to a given audience, framed for that audience, without maintaining duplicate portfolios, because reviewers judge fit in seconds and a diluted portfolio reads as unfocused.

How Might We

How might we let someone assemble an audience-specific portfolio in minutes from work they have already uploaded?
How might we keep one item live across many portfolios so nothing is duplicated or goes stale?
How might we make each portfolio feel purpose-built, with its own headline, call to action, and look?

Design principles (each traceable to an insight)

Work is one shared library. A piece is uploaded once and lives in a single library. (Insight 2)
Portfolios are compositions, linked from the library. An item is linked into a portfolio, so editing it once updates it everywhere and nothing is duplicated. (Insight 2)
Each portfolio is framed for its audience. A portfolio carries its own headline, description, call to action, profile, and look. (Insight 3)
Relevance is the unit of conversion. Composing a portfolio is choosing exactly what a given audience sees. (Insight 1)

Insight → decision map

Insight (from Empathize)	Concrete design decision
Mixed portfolios rate a fit 34% vs 72% focused	A creator composes a portfolio by selecting the relevant items, so a given viewer sees only work that matches their need
Tailoring is avoided because of duplicate upkeep	Items are linked into a portfolio: uploaded once, edited once, present in any number of portfolios without going stale
A relevant subset still needs framing	Each portfolio has its own h1, description, CTA, profile, social links, grid style, and aspect ratios, and its own link

Ideate & Craft — Work lived in one library; portfolios were compositions of it, each with its own framing

In this section: Design execution · Before → after · Other deliverables

Design execution

One work library — every piece is uploaded once as an image (lightbox), a video (lightbox), a written case study (typed or loaded as .md), or a display-only item that shows in the grid without opening. The library is searchable by category and tag.
Portfolio composer — a creator selects items from the library into a portfolio, where each item is linked rather than copied, and sets the grid (1, 2×2, 3×4, or 4×4), the item aspect ratios (4:5, 5:4, 3:2, 1:1), titles shown beneath items, and tag filters for the viewer.
Per-portfolio framing — each portfolio carries its own h1, description, CTA (text and destination), profile photo, and social links, and lives on its own link, with one set as the base at fave.work/username. A product-design portfolio can open with "Software people choose, use, and pay for" and a "Start your product" CTA, while a content portfolio opens with "Turn a following into a customer base" and a "Let's plan your content" CTA, from the same library.
Coexisting portfolios — area-based portfolios (illustration, product design, websites) and one-off portfolios assembled for a specific client or role all coexist, each on its own link.

Before → after

	Before (single-profile platforms)	After (Fave Work)
Portfolios per person	One stream of everything	Many, one per audience or purpose
Tailoring to an audience	Build and maintain a separate site	Compose from the library in minutes
Duplication	Re-upload per site	Items linked, uploaded and edited once
Framing	One headline for all work	Each portfolio's own h1, CTA, and look

Other deliverables

Built in Figma with Dev Mode handoff: the work-library and item-type model, the portfolio composer with per-portfolio grid and aspect-ratio settings, the per-portfolio framing controls, and the linked-item architecture that keeps one piece live across portfolios.

Dashboard — Composing from a library multiplies tailored portfolios

Composing from a library multiplies tailored portfolios
Scope: Last 30 days · pilot (60 creators)
Guiding question: Did creators build and send audience-specific portfolios?

  Tailored portfolios per active user .... 1.0 → 3.4
  Median time to assemble a new portfolio  ~9 min
  Sends that were a tailored portfolio ... 22% → 78%

Key Insight: With items linked from one library, building a portfolio for a
specific audience took minutes, so creators sent the relevant one.

Prototype / Test — Saved filters were fast and read as a generic gallery; composed, framed portfolios gave creators control over what each audience saw

In this section: The experiment · What it taught

The simplest model was to make a portfolio a saved tag-filter over the one library: a portfolio would be a dynamic view that updated as the library grew. It was A/B tested against the explicitly composed, separately framed portfolio in Statsig across the pilot.

The failed variant. Saved filters set up fastest, but every portfolio looked like the same gallery with different items and carried no audience framing, and adding new work to the library leaked into every filtered portfolio, so creators rated their control over "what this audience sees" at 48%. A dynamic filter could not be aimed at a viewer, and creators would not send something they could not control.

A saved filter is fast and can't be aimed at a viewer
Scope: Statsig A/B · pilot · n=2 variants
Guiding question: Which model lets a creator control what an audience sees?

  Variant A — Portfolio as a saved filter
    Setup speed ..................... fastest
    Perceived control of the view ... 48%
    Rated "purpose-built" by reviewers low

  Variant B — Composed, separately framed portfolio
    Setup speed ..................... +a few minutes
    Perceived control of the view ... 91%
    Rated "purpose-built" by reviewers high

Key Insight: A composed portfolio with its own framing let creators aim a
portfolio at a specific audience, which a dynamic filter could not do.

What it taught. When the value is controlling exactly what a specific audience sees, a portfolio has to be a deliberate composition with its own framing; a dynamic filter is lower-effort and reads as a generic gallery the creator cannot aim. The composed model shipped, with items still linked from the shared library.

Outcomes & reflections

In this section: Causal chain · Reflections

Causal chain (pilot, 60 creators)

The composer let creators assemble audience-specific portfolios from work uploaded once, each carrying its own headline, CTA, and grid, so tailored portfolios per user rose from 1.0 → 3.4 and the share of sends that were a tailored portfolio went from 22% → 78%, putting relevant work in front of each viewer; in the reviewer panel a role-focused portfolio raised the rated fit from 34% → 72%, the response a creator could expect, and day-30 retention rose from 43% → 61%.

Metric	Before	After	Δ
Tailored portfolios per active user	1.0	3.4	~3.4×
Sends that were a tailored portfolio	22%	78%	+56 pts
Reviewer rated-fit (focused vs mixed)	34%	72%	~2.1×
Day-30 retention	43%	61%	+18 pts

Scale note: a portfolio is judged one high-stakes view at a time, so putting only the relevant work in front of a specific employer or client is what earns a response, and the payoff compounds across every targeted send.

Reflections (transferable principles)

For a portfolio, relevance to the viewer is the conversion lever: a reviewer judges fit in seconds, so showing only the work that matches their need outperforms showing everything.
Separating the work from its presentation lets one library serve many audiences: a shared body of work and many composed, framed views means tailoring no longer costs duplication or goes stale.
A targeted view needs its own framing to land, with a headline and call to action aimed at the audience, so a portfolio is a deliberate composition and a dynamic filter alone reads as a gallery.