UX Case Study
St. Jude
Memorial Lead
An AI-assisted workflow that finds the right person to receive a handwritten card when a St. Jude donor passes away. The AI suggests. The user decides.
Selected Household
Likely Duplicates Households
Surviving Contacts InfoAI-Extracted
Overview
Less Searching. More Connecting.
The Short Version
When a donor passed away, reaching their family depended on someone remembering to search obituaries by hand. I designed Memorial Lead around three decisions: let AI find candidates but never act on them, put the entire review on one screen, and end every flow in a handwritten card instead of a templated email. Users went from finding 25 recipients a day to 75.
The Problem
Users were manually searching obituaries, copy-pasting names, and sending generic emails, if they remembered at all. A process held together by manual effort and good intentions was failing families at exactly the wrong moment. Memorial Lead was designed to make the right thing easy.
Why It Matters
Major gift fundraising is built on relationships. Bereavement is where those relationships are tested, and the family of a donor is often the next generation of supporters.
Business Goals
User Goals
My Role
Lead designer on Memorial Lead from discovery through launch. I partnered with the product team from research through ship, and owned the AI review UI, the contact selection flow, and the handwritten card integration. Designing AI-assisted workflows means the interface has to build trust, not just display data.
Working Across the Org
Users & development associates
Interviewed the people who had been doing this manually for years, and mapped their workflow end to end before designing anything.
Product team
Partnered from research through launch, keeping every decision optimized for a quick and intuitive flow.
AI / engineering
Defined how model output surfaces in the UI, candidate names with confidence scores and source excerpts, and how the interaction behaves when the AI is uncertain.
Handwritten card vendor flow
Designed the queue handoff so a completed review ends in a physical card, not another templated email.
Research
Learning From People Who Did This by Hand
Before designing anything, I immersed myself in the user flow and talked to the users. I mapped the existing manual process end to end and spoke with the people who had been doing it for years about where it broke down. Four structural pain points surfaced consistently.
No way to know who to reach
Users had no structured process for finding next-of-kin. Outreach relied on memory, Google, and word of mouth, if it happened at all.
Manual and time-consuming
Finding a name, verifying a relationship, and drafting outreach could take over an hour per donor. Most slipped through entirely.
Wrong person contacted
Without structured data, users often reached out to the wrong family member, or the deceased themselves. It was an awkward and painful failure.
Missed relationship window
Families of major donors are often the next generation of supporters. A delayed or missed card meant a lost connection at the most important moment.
What the Workflow Itself Revealed
Reading the whole obituary was daunting
Manually scanning a full obituary for every possible name was slow and easy to get wrong. Missing a name meant a missed family.
The existing flow sent users all over the screen
Information was scattered. Users jumped between sections to complete one task, which created confusion and slowed everything down.
Scannability was the priority
With so much text on screen, users needed the relevant information at a glance, not to read everything to find what mattered.
How It Works
Six Steps From Obituary to a Card in the Mail
Obituary flagged in external app
Deceased linked, duplicates merged
AI scans obituary for recipients
User selects recipient
Added to handwritten card queue
Card mailed to recipient
Step 3 was the hard one
The AI step was the most technically complex. I designed the review UI around model output that surfaced candidate names with confidence scores and source text excerpts, so users could verify a suggestion against the obituary itself. The interaction had to work even when the AI was uncertain.
Design
The AI Suggests. The User Decides.
Prototype · One Screen. Huge Impact.
I built and iterated the prototype in Axure and tested it with users through multiple rounds. The walkthrough below shows the full review: obituary status, duplicate merge, AI-extracted family contacts, recipient selection, and the note that rides along with the card.
Outcomes
The Right Outreach, at the Right Moment
Nothing slips through
Detection stopped depending on someone remembering to check. Obituaries are flagged systematically, so high-value donors no longer disappear from the relationship unnoticed.
The right person, verified
Confidence scores paired with source text excerpts let users verify each suggestion against the obituary itself, ending outreach to the wrong relative or the deceased.
A card, not a template
Every completed review ends in a handwritten card. The outreach finally matches the moment: personal, timely, and human.
A pattern for AI features
Suggest-and-verify, confidence made visible, and a human on every consequential action became the template for how the platform approaches AI-assisted workflows.
Next Steps
Where Memorial Lead Goes From Here
Key Takeaway
"Designing AI-assisted workflows means the interface has to build trust, not just display data."
The AI suggests; the user decides. Showing confidence scores next to the source text they came from is what made users trust the suggestions enough to move fast. Since Memorial Lead, I design every AI feature around making the model's uncertainty visible instead of hiding it.
Expand obituary detection beyond the current source into regional and paywalled outlets, closing the remaining coverage gaps.
Feed user corrections back into the model, so every rejected suggestion improves the next one.
Add gentle timing intelligence: queue the card to land inside the outreach window rather than the moment the review completes.
Extend the suggest-and-verify pattern to other relationship moments, like major life events surfaced from public records.
Interactive Prototype
Try Memorial Lead Yourself
This is a working model of the review screen. Mark the donor as deceased, merge the duplicate accounts, check the AI-extracted family contacts against the obituary, pick the recipient, and leave a note for the card. The whole review, on one screen.
* Selected Margaret Moulton DMS Account Info
| Selected DMS Account Name | Likelyhood | Lifetime Giving | Lifetime Donations | Street Address | Auto Generated Date of Death | |
|---|---|---|---|---|---|---|
Margaret m. Moulton | High | Yes | $968,150 | 109 Elmwood Dr, Toledo, OH 12345 | View Profile |
Likely Duplicate DMS Accounts for Margaret Moulton Accounts
| DMS Account Name | Likelyhood | Lifetime Giving | Lifetime Donations | Street Address | DMS Date of Death | ||
|---|---|---|---|---|---|---|---|
| Margaret m. Moulton | Moderate | Yes | $600,150 | 109 Elmwood Dr, Alden, OH 12345 | None | View Profile | |
| Margaret m. Moulton | Low | Yes | $781,453 | 3455 Sparkel Drive, Spring Feild, OH 12666 | 09/23/2000 | View Profile |
Obituary Auto Generated Info
Nickname
Catherine Mary Ullrich
Last Place Lived
Alden, State Unknown
Date of Death
August 14, 2021
Spouse
John Ullrich
Possible Siblings
Unknown
Possible Children
Unknown
Deceased Obituary from Legacy.com
First Name
Margaret
Last Name
Moulton
Funeral Location
Starburg, Ohio
Published Obituary Date
8/18/2021
Margaret M. Moulton Born: August 25, 1955 Died: August 14, 2021 Catherine Mary Ullrich, 65, of Alden died Saturday, August 14 in the arms of her beloved husband. She came into this world August 25, 1955, to William E. Grunewald and Mary Ellen Campbell in Las Cruces, NM. She was the second of nine children. Her creative spirit and talents were obvious to those around her from a young age. On May 24, 1987, she married her soul mate John Ullrich in Wauconda, Illinois. Through this marriage, Catherine also became stepmother to four children. Catherine was an accomplished American Contemporary Pen & Ink Folk Artist and author. She depicted her rural life of growing up on a farm in her many lithographs and children's book. Her artwork has been exhibited in the Museum of American Folk Art, Smithsonian Institute, and galleries nationwide. She was commissioned by the President and invited to the White House as a featured artist on two occasions. Catherine was a collector of antiques. Her and Johnnie, as she liked to call him, visited antique shops, flea markets, and estate sales as they traveled; always finding a new prize for the collection. These collections fueled her second wave of art being mixed medium sculptures. She crafted fanciful creations out of her personal antiques. Catherine and John often traveled to valley exposures meeting box collectors, but her favorite place was home. Her gardens were legendary in the neighborhood.
* Family Contact Info From Obituary
| Name | Relationship | Donation | Date of Creation | Lifetime Giving | Lifetime Donations | Memorial Giving | Memorial Gifts | ||
|---|---|---|---|---|---|---|---|---|---|
| John Ullrich | Spouse | $1,250 | 4/24/2021- Auto | $968,150 | $64,500 | $98,761 | 12 | View Profile | |
| Kenndey Family | ? | $3,780 | 4/26/2021- Manual | $30,145 | $45,670 | $40,761 | 163 | View Profile | |
| Mathew Williamsmith | Son | $2,250 | 4/24/2021- Auto | $765,132 | $66,500 | $30,423 | 34 | View Profile |
Donations Made In Selected Marget Moulton Name
| Total Memorial Giving | Total # Donations | Last Donation Date | Memorial Gift Fund Page |
|---|---|---|---|
| $66,500 | 163 | 01/23/2022 | Yes |
| Recipient (Letters Mailed To) | Memorial Giving | # of Donations | Donation Date |
|---|---|---|---|
| Jackson Hampher and Mary Hampher | $1,250 | 1 | 12/22/2021 |
| Jamie Williamson Family | $4,250 | 12 | 9/24/2021 |
| Jackson Reed | $9,250 | 34 | 8/12/2021 |