Product Design · Systems Thinking · Cross-Functional Leadership

Candid Search — one platform for the sector's most comprehensive nonprofit and funder data.

Candid holds the most complete, most trusted data about the social sector. The challenge: design a single search experience that makes that data feel accessible — serving four distinct user types, reconciling two merged product legacies, and keeping the design consistent across multiple feature teams.

My Role
Design Director & IC
Approach
Systems thinking · Cross-functional leadership
Collaborators
Product, Engineering, UX Research
Method
Scaled Agile (SAFe)
Launched
app.candid.org
Candid Search
1.9M
Org profiles unified
30M+
Grant records indexed
$180B
Annual grants tracked
4
Distinct user personas served
14K
Beta users before launch
#1
Most comprehensive nonprofit dataset
Context

The most comprehensive data about the social sector. One platform to access all of it.

01
Two separate platforms, two separate data models
GuideStar and Foundation Directory Online had different information architectures, taxonomies, and UX patterns. Merging them wasn't additive — it required fundamental rethinking of how users navigate between nonprofit profiles and grant data.
02
Radically different user goals within the same session
A fundraiser researching a funder needs to switch instantly between grant history, mission alignment, and contact information. A compliance officer needs quick verification. A donor needs trust signals. Search had to work for precision lookup and open-ended discovery simultaneously.
03
Data at a scale that resists simplicity
2,500+ data points per organization. 6 million public records ingested annually. A proprietary five-facet taxonomy that needed to feel learnable. Making this feel like a search product — not a database query tool — required every design decision to earn its place.
Role & Approach

Design direction and IC work — across the whole product, not a slice of it.

Systems thinking across feature teams
Feature teams in a SAFe environment go deep by design — which means they can drift into incompatible assumptions. I held the whole product in view: spotting when two teams were solving the same problem differently, when a local decision would create a global inconsistency, and when a new feature needed to be designed as part of a system rather than in isolation.
Cross-functional leadership
I was a consistent presence in conversations with PMs, engineers, and UX researchers — not just in design reviews. I translated between disciplines, made sure design intent survived into implementation, and represented the user perspective in product and engineering decisions before designs were ever finalized.
IC work embedded with UX research
I worked closely with our UX researcher throughout — not just receiving synthesis reports, but shaping research questions and translating findings into design decisions with each team. I also contributed directly to the design work: brainstorming and ideating alongside designers on every major feature area.
The systems challenge
User Research

Four personas. Four fundamentally different relationships with the same data.

The Fundraiser
Development Associate · Grants Manager · Prospect Researcher
Primary goal
Find and cultivate institutional grantmakers whose priorities align with their organization's mission — fast. Every hour in a search tool is an hour not spent writing grants or building funder relationships.
Grant history by funder Subject area filtering Geographic focus Board connections LOI requirements Contact info
Key friction
"Have to look in many different places to find information on one foundation. Can't sort by common grant amount. No CRM integration."
The Funder
Program Officer · Grants Manager · VP of Program Administration
Primary goal
Vet grant applicants and understand the nonprofit landscape in a given subject area or geography. Conduct due diligence before disbursing grants. Build lists aligned with funding priorities.
IRS compliance status 990 financials Staff size trends Past funders Demographic data Landscape analysis
Key friction
"No single source of data for due diligence. Relying on 990s is a habit — profile data is often outdated. Candid not integrated with our grants management system."
The Individual Donor
Cause-driven · Research-before-donating · Often arrives via Google
Primary goal
Quickly verify that a nonprofit they're considering is legitimate, effective, and transparent. Most arrive via search engines — unaware of Candid as a brand.
Mission & impact info Financial ratios Seal of Transparency Program descriptions IRS ruling year
Research insight
76% conduct research 10 or fewer times per year. Most turn to Google first — none mentioned GuideStar or Candid unprompted.
The Commercial User
Marketing · Sales · Wealth Management · Legal Services
Primary goal
Find and contact nonprofits and foundations that could benefit from their products or services. Identify prospects by size and subject area, then locate relevant decision-maker contacts.
Email addresses Asset filters Revenue filters Contact names & titles List export CRM integration
Key friction
"Need better quality email data. Export limits mean duplicates can't be identified until after export. No person-level search."


Personas

Design tension

"When users are looking for a specific nonprofit, they want name matches at the top of results. When they want to discover new organizations, they want to match other profile data and easily filter results."

This research insight shaped the entire information architecture of search results. We couldn't optimize for one intent without degrading the other — so we built dual-mode search logic, surfaced through the UI.

User Flows

Mapping the paths through complex data.

Search query "homeless shelter" 1 Results page Precision + discovery 2 Intent detected by search engine Exact match Name matches first 3a — precise lookup Discovery mode Filter-first, broad set 3b — open exploration Apply filters Subject, geo, amount 4 Funder profile Grants, mission, LOI 5 Save to Project 6 Iterate / refine search based on profile findings
Receive grant application via GMS or email 1 Search org by name / EIN Candid Search 2 Review profile + 990 data Mission, financials 3 Charity Check compliance IRS status + alerts 4 Flags found? 5 No Proceed to funding Export report 6a Yes Escalate for review 6b
Wireframes

From structure to surface.



Wireframes
Design Decisions

The decisions that shaped the experience.

Dual-mode search: precision vs. discovery
Research surfaced a fundamental tension: some users know exactly what they're looking for (an org by name), others are exploring broadly. We designed the search engine to detect intent from query patterns — exact matches surface name results first; descriptive queries prioritize faceted filtering and semantic matching. One bar, two behaviors, invisible to the user.
Progressive filter disclosure (15+ dimensions)
The platform supports 15+ filter categories: subject area, population served, geographic area, org demographics, revenue, assets, IRS status, and more. Surfacing all filters simultaneously would overwhelm new users. We introduced a progressive disclosure pattern — most common filters always visible, advanced filters accessible via expand. Power users get everything; beginners see what they need.
Context-aware result cards per persona
A nonprofit search result card that's useful for an individual donor (mission, Seal badge, financial health) differs from one useful for a commercial user (assets, revenue, email contact). We designed a modular card architecture with context-sensitive information hierarchy — same record, different emphasis based on user type and subscription tier.
PCS taxonomy as a learnable UX layer
Candid's Philanthropy Classification System uses five facets (subject, population, org type, transaction type, support strategy) — far more nuanced than the old NTEE single-code system. We designed search-as-you-type autocomplete that teaches the taxonomy through use, maps natural-language queries to PCS codes, and exposes the underlying structure without requiring users to learn it upfront.
AI features as workflow acceleration, not replacement
The AI-powered funder recommendation engine and LOI writer were designed to reduce time spent on known-tedious tasks — not to replace human judgment. We positioned them as a first draft, not a final answer. The LOI writer explicitly draws on the user's own nonprofit profile data and the funder's public priorities, making the output grounded and editable rather than generic.
Compliance data as a first-class surface
Funder users needed quick, trusted verification of IRS compliance status — but were bypassing profile features entirely, going straight to 990s from habit. We elevated compliance indicators (501c3 status, alert flags, Pub78 verification, OFAC status) to the top of the profile card, reducing the cognitive overhead of the compliance check workflow and making the profile the reliable single source of truth.


Candid Search

New Capabilities

Designing the AI features — acceleration without abdication.

AI-Powered
Funder Recommendations
Analyzes a nonprofit's mission, subject areas, geographic focus, and organizational profile against historical grantmaking patterns across 3M+ transactions to surface funders that have actually funded similar organizations — not just ones with matching keywords.
AI-Powered
Letter of Intent Writer
Generates a first-draft LOI by pulling directly from the user's Candid nonprofit profile (mission, programs, demographics) and the target funder's public priorities and grant history. Designed specifically to benefit non-native English speakers and under-resourced fundraisers.
New
Projects Workspace
A digital workspace for tracking funding prospects across their grant-seeking lifecycle — with saved searches, funder prospect lists, task management, and deadlines. Replaces the spreadsheet workflow that most fundraisers had been using.
Outcomes

What shipped — and what it means.

14K
Beta users before launch
230+
Platforms powered by Candid API
62%
More donations with Seal
2K+
Community partner libraries
86K+
Orgs sharing demographic data
← Previous
Next →