ML-powered family stories built from historical records, driving a ~14% subscription lift.
Ancestry’s StoryScout
My role
Design Lead & Research
End-to-end flow, data structure & user research
Platforms
Mobile & desktop
iOS, Android, mobile web & desktop
Collaboration
Cross-functional
User Research, Product, Engineering & Data Science
Timeline
1 year
Research, design, iteration & launch
AT A GLANCE
Family history.
Without the homework.
The problem
Most new users never built a tree past their grandparents. Genealogy asked for research skills they didn’t have.
My contribution
I shaped the product into four workstreams and designed an experience that lets machine learning do the searching.
The result
A ~14% combined lift in subscriptions across verticals, beating go-to-market goals.
THE CHALLENGE
Curious about family.
Not about research.
75%
of new users never created a tree, or only added their grandparents.
“This is incredible! I have ancestors who were on the Mayflower?”
Ancestry’s core experience meant recalling details, entering them by hand, and knowing how to research. That’s where newcomers stalled.
We focused on the “Family History Curious”: people aged 25–44, mobile-first, who loved family stories but didn’t expect to do research. Potentially around 35 million new users in the US alone.
Interviews showed people expected Ancestry to already know who they were. They didn’t know how to find ancestors, struggled to interpret records, and got lost in terms like “2nd cousin twice removed.”
Four problems.
One connected journey.
01
Building a tree
Help people get started without knowing how to research.
02
Presenting lineage
Make family connections understandable without genealogy jargon.
03
Interpreting records
Turn hard-to-read historical records into something personal.
04
Organizing stories
Give discoveries a familiar home to return to.
Key decisions
Decision 01
Start with someone
you know.
Instead of asking people to search records themselves, StoryScout’s machine learning does the work. I tested several starting forms: a grandparent’s name and a place they lived drove the highest completion.
Most people know their grandparents and can verify them. I made grandparents the anchor for onboarding, lineage, and the story library.
One familiar person becomes the starting point for a much bigger story.
Decision 02
Show the connection.
Not the whole tree.
Full family trees overwhelmed novices. Looking beyond genealogy — at GitHub branching and computer folders — I designed a view showing just three key people.
Relationships use everyday language, like “Mary’s brother.” The full family tree stays a tap away for anyone who wants to go deeper.
Make the relationship clear before asking people to understand the tree.
Decision 03
Records become
stories worth reading.
Handwritten records are often difficult to read. I turned their facts into snackable narratives: 50–90 words, with one or two sentences per detail.
Maps, images, and historical context made stories feel grounded. Access to the original record built trust and engagement. But leaning too far into education caused drop-off, so we kept the experience light.
Keep the story approachable, and keep its evidence close.
Decision 04
A familiar home
for every discovery.
Discovered stories had been scattered, with no clear place to revisit them. I organized the library around grandparents so people always had an anchor for finding stories across their family.
Research showed the layout was easier to understand. The same grandparent model also helped people grasp how relatives connect.
A consistent anchor connects getting started, understanding, and coming back.
THE LEARNINGS OVER TIME
Beyond the first “wow.”
After several iterations, I proposed and ran a four-week diary study with tailored prototypes. We wanted to see how people’s feelings about StoryScout changed over time — not just at first use.
Where it worked
A powerful first impression and a real “in awe” moment.
Low effort to learn something valuable about family.
Genuine curiosity — people remembered what they learned.
A family tree that could carry into Ancestry’s core product.
Where it fell short
The awe faded after 2–3 weeks as stories grew repetitive.
Historical images sometimes looked like actual family photos.
People still needed tools to build their own family trees.
WHAT CHAGNED
Play to its strength.
StoryScout was strongest at the first impression. We brought it into marketing funnels and targeted campaigns, where an unforgettable first moment matters most.
An excellent introduction isn’t automatically a lasting habit. Research helped us find where the experience belonged.
THE IMPACT
New stories.
New connections.
14%
combined lift in subscriptions across verticals
Exceeded
go-to-market conversion goals for new users, registrants & DNA customers
Strong engagement: most people who received a story opened it. The story library and reusable patterns went on to support experiences across Ancestry.
Reused
story and design patterns scaled into DNA results and other Ancestry products