UIUXfolio Projects

Expanding Family Discoveries through
Full-Text Search and AI

Project timeline

4 months

Year

2026

Company

Ancestry is an online genealogy and consumer genomics company that helps people discover, preserve, and share their family history.

The project itself :

Project Overview

The new Full-Text Search experience unlocks contextual narratives hidden within billions of archival images, allowing genealogists to find ancestors mentioned in secondary or supplementary roles - such as witnesses, neighbors, or bondsmen- that are invisible to traditional name-index searches.

Problem:

Millions of unindexed and under-indexed historical records were only discoverable through manual browsing, which left critical ancestral clues locked away and researchers stalled at "brick walls.

Goal:

Design and optimize a new full-text search experience along with some key enhancements:
• Launch 10 collections, totaling 51 million images
• Optimize Search UI and Guidance
• Upsell by driving subscriber upgrades using multiple entry points

My role:

UX Product designer leading the new Full-Text Search design, flow and experience

Responsibilities:

Competitive analysis

Heuristic evaluation
High-fidelity design concepts
Prototyping

Qualitative research data

Secondary customer research
Usability testing
Userflows

All about the research:

Discovery Phase

Before architecting the new experience, I structured our discovery around three fronts: understanding full-text search usability patterns, analyzing how competing discovery platforms support both casual users and power-user genealogists, and auditing baseline performance data from our search tools to benchmark success.

UX Research Plan

Mental Models:

How do users intuitively navigate full-text search, and where does it deliver the highest value in discovery flows?

Comp Analysis:

How do other platforms balance simplicity for everyday users with expert tools for geneaologists?

Baseline metrics:

What are the key metrics of our legacy search tools and what benchmarks define success for our our new search experience?

Customer Research

Synthesized secondary customer research and usability archives across our UXR libarary to identify core usage metrics and validate the underlying demand for Full-Text search.

I found that the root problem was how we read records: our search only scanned a handful of basic form fields like names and dates. This meant entire pages of handwritten wills, family histories, and story-rich books remained invisible to search queries—locking away the exact stories users came to find and driving down engagement

3.5%

Collections Under-indexed

Key records un-extracted, leaves details invisible.

15%

Low Search Initiation Rate

Only 15% of active registrants initiated a search, an unusually low engagement given our 26k+ collections.

10%

Low Result Click-Through

Only 10% of active registrants clicked a result, signaling dead-ends and low query relevance.

Competitive Analysis

Phase I - Search form features

Captured and annotated screenshots of full-text search features from 3 leading family history platforms, citing strengths and weaknesses of each.

This analysis was important to discover new opportunities, brainstorm ideas for new features and observe common layout designs.

I was able to cite some of the strengths and weaknesses of each of the search tools from our top competitors. Below are UI search landing pages from Family History and MyHeritage.

Family History had the best example of a successful, fully functional Full-Text Search tool. It provided a freeform keyword text field and search tips.

Although MyHeritage does not exclusively have a full text search tool, our competitor offered critical interaction takeaways. Their search interface prioritized a clean layout—stripping away visual noise at the primary entry point.

Phase II - AI transcription layouts

Created a competitive audit of how LLMs handle record image transcriptions across a range of formats. The top 4 AI Assistants (LLMs) reviewed for analysis were:
• Claude (Anthropic)
• Gemini (Google AI)
• Chat GPT (Open AI)
• Scribe (used in MyHeritage)

This competitive audit is to inform the redesign of our record image viewer —specifically benchmarking how leading platforms pair original document scans with AI-assisted transcriptions to create a seamless, side-by-side reading experience."

The project schematically :

Define Phase

After clear insights of legacy search pain points and competitive analysis, I partnered closely with our content team to map out end-to-end user flows, aligning cross-functional stakeholders on the new Full-Text Search journey from initial query to record discovery.

Userflows

Mapped the primary end-to-end user flow from entry point and query submission to the final record view, establishing a shared blueprint with engineering and product stakeholders around page transitions and core interactions.

Alongside the main path, I integrated the paywall and upgrade funnel for non-subscribers discovering the Full-Text Search entry point.

My goal was to map the end-to-end user flows for both subscribers and non-subscribers—identifying critical paywall interception points, evaluating drop-off friction, and ensuring an intuitive path to high-value search results.

Content Reviews

Partnered cross-functionally with Product Management and Content Strategy across 4 workshops and async Figma reviews, co-designing the content system for the new search experience—including onboarding, contextual help text, and animated guidance patterns.

The purpose of these workshops was to establish clear, user-centered copy across the onboarding modal and contextual help text into the finalized, production-ready interface designs.

The content workshops focused around first educating users around the new tool, what it entails, and what value it delivers. The second step was to provide contextual help text and sample prompts that listed legal and official terminology that were commonly used in historical records.

The goal was to design an accessible onboarding modal for both novice users and expert genealogists, replacing text-heavy walkthroughs with animated images.

Starting with an empty keyword box created friction because historical archives rely on naming conventions most researchers never encounter. To bridge this gap, I teamed up with our content strategists to craft proactive helper patterns and suggestion prompts

The clear version :

Design phase

Under tight release timelines, rapidly iterated from mid-to-high-fidelity mockups using our design system, translating PRD requirements into an adaptable search interface.

Mockups

Phase I - Search landing page

Designed 4 high-fidelity mockups for the Full-Text Search interface—featuring a refreshed layout, an engaging custom illustrated hero banner, contextual help text, and guidance animations—to align Product, Engineering, and Design early, de-risking technical constraints and visual hierarchy before final development handoff.

These desktop variations explore distinct hero banner treatments, typographic hierarchies, and visual pairings designed to set clear user context at the entry point. Alongside visual explorations, I architected advanced search controls—including dynamic date-range selectors and Boolean keyword operators (exact match, term exclusion, and required qualifiers)—strategically categorizing power-user capabilities to stage as fast-follow enhancements for Phase 2.

Phase II - AI transcription layout explorations

Designed 10+ high-fidelity layout variations for the record detail transcription page, synthesizing competitive patterns into 4 standardized, content-responsive templates that streamlined data entry and established clear UI specifications for engineering and data partners.

Our legacy transcription tool was constrained to a narrow side panel inside the record viewer, forcing dense, unformatted copy into a cramped reading space. The redesign reimagines this panel into a responsive, highly legible reading experience—replacing text blocks with structured, scannable layouts tailored to diverse historical record types.

High-fidelity prototype

It's the detailed, interactive version of designs that closely match the look and feel of the final product.

This prototype is a phase I of the FTS flow. It includes text transcription with a bounding box highlighting all the keywords from the initial search query.

Selects Full-Text Search at main navigation
Views the keyword help text
Enters a keyword following a first and last name
Selects Search
Results view page
User selects a result and lands in the record image view page
Image view page highlights all the keyword queries in the image
Image view page auto opens sidebar with AI transcription (AI text format TBD)

The project schematically :

Outcome

Full Text Search went live in Feb 2026 marking the definitive transition of the genealogy industry from a digital archiving phase to an AI-driven discovery phase. The feature also debuted at RootsTech (largest family history conference in Salt Lake City, Utah) as part of Ancestry's "What's New" series, presented by Corporate Genealogist, Crista Cowan.

all -access user adoption

users clicked on search results

7-day return rate after usage

Takeways

This was a huge lift for any one team, it took 7 teams to ship this experience in such a short time and get it ready to demo for our annual conference.

Impact:

• 4.9% of all-access users adopted the new feature

82% of users continue searching after landing on the results page
85% of users clicked on search results after searching
83% 7-day return rate
Number of images released: 50 million
Showcased the feature through live, interactive demos for hundreds of users at RootsTech, driving feature awareness and hands-on engagement

What I learned:

If we had more time, I would've loved to include an advanced keyword search parameter feature including a typeahead and smart AI prefilled form to showcase at the family history convention.

Next Steps

Although our Full Text Search feature has been a success amongst our users, this MVP was meant to be a first step to get the ball rolling as we continuously add collections and make improvements.

Implement proposed AI layouts for record view transcription as a fast follow

Upcoming plans to launch an additional 300 million image transcriptions for more discoveries. We will alert our subscribers as new information is available

Continue to monitor usage of Full-Text Search and its quantitative data to find ways to improve any friction

Expanding Family Discoveries through Full-Text Search and AI

Expanding Family Discoveries through Full-Text Search and AI

Project Overview

Problem:

Goal:

My role:

Responsibilities:

Discovery Phase

UX Research Plan

Mental Models:

Comp Analysis:

Baseline metrics:

Customer Research

3.5%

Collections Under-indexed

15%

Low Search Initiation Rate

10%

Low Result Click-Through

Competitive Analysis

Phase I - Search form features

Phase II - AI transcription layouts

Define Phase

Userflows

Content Reviews

Design phase

Mockups

Phase I - Search landing page

Phase II - AI transcription layout explorations

High-fidelity prototype

Outcome

all -access user adoption

all -access user adoption

users clicked on search results

users clicked on search results

7-day return rate after usage

7-day return rate after usage

Takeways

Impact:

What I learned:

Next Steps

Expanding Family Discoveries through
Full-Text Search and AI

Expanding Family Discoveries through
Full-Text Search and AI