Redesigned ALS-LM website showing the homepage with hero image of three smiling adults, impact statistics, and the Support Us page with donation options.

Rescuing a Nonprofit's Donation Flow

5/5 A/B testing participants preferred the redesigned donation flow for confidence. NPS improved from 1.25 → 7.0 average.

Visit Original Site

Visit Figma Prototype

Technology

Figma

Scope

UX Research, Interaction Design, Usability Testing, A/B Testing

A semester-long UX redesign of Adult Learning Systems — Lower Michigan's donation flow, using the UX Research method to form, test, and revise design hypotheses.

The Problem

ALS-LM is a 501(c)(3) serving adults with disabilities in South East Michigan. Their donation page requires mailing physical cash, and no digital payment exists. The site's overall design lacks professional visual design, having long and information dense paragraphs, and may lack trust signals. These issue creates a trust and conversion barrier.

Diagnosis

Persona: John, a middle-aged donor with disposable income, altruistic, interested in the cause
Peak-End Rule applied: the ending (users seeing mail-only) is the worst moment, and per Peak-End, users disproportionately remember the ending
Psych scores table (compact): Steps 1–5 with scores (+10, +5, -15, -10, -20)

Line graph of user emotional response across the donation flow: scores rise to +15 at step 3, then drop sharply to −30 at the final step when users discover mail-only donations.

Hypothesis

I made six falsifiable hypothesis and predictions, each grounded in a named design principle.

H1: A less dense paragraph with icons makes people more willing to read (Reduces Cognitive load)
H2: Displaying impact metrics on the homepage increases persuasion (social proof from Cialdini)
H3: Adding client testimony increases persuasion.
H4: Providing a digital donation flow increases task completion likelihood
H5: Adding a staff group photo on the last page increases personal connection and increases return intent (Peak-End Rule)
H6: Social media CTA post-donation increases follow behavior

The Redesign

Redesigned homepage section displaying impact statistics: 300+ disabled individuals achieved independent living, 25+ years of service, and 600+ free equipment distributed.

Added Impact statistics (300+, 25+, 600+) on homepage → tests H2

Client testimony section showing a smiling man in a wheelchair outdoors, with the quote: 'ALS-LM helped me to achieve independent living within 6 months.' — Tony Smith.

Added client testimony with photo → tests H3

Digital donation form with frequency toggle for one-time or monthly giving, preset amount buttons from $10 to $200, a custom amount field, and a Donate button.

Added digital donation form with preset amounts → tests H4

Redesigned Support Us page with three illustrated cards for Financial Support, Donate Equipment, and Volunteering, each with a brief description and call-to-action button.

Re-arrange support Us layout with 3 illustrations → tests H1

Thank You page displayed after donation, featuring a group photo of staff members outside a lodge and social media links to Instagram and Facebook.

Add a thank You page post donation with group photo and social media links → tests H5, H6"

Before

After

Reduced CARF paragraph text → tests H1

A/B Testing

Due to Time constraints, I could only recruit users from or close to academia. But I tried to list the criteria that are potentially most likely to donate.

I recruited and tested 5 users, 3 of whom are younger adults in their early career phase, and 2 of them are mid to late adults. 2 of them are currently students who have had full-time job experience, 1 is currently full-time at a religious/non-profit industry, 1 researcher/professor at UM, and 1 software developer.

Test protocol summary

Think-aloud, 3 tasks per version, post-test survey with NPS/WoMI/trust/confidence/engagement/conversion/retention

Key Findings

H1: ⚠️ Falsified - 3/5 expressed opposite: they wanted MORE detail (contradicts assumption)
H2: ✅ Verified (3/5), with no opposites
H3: ⚠️ Weak (1/5), with no opposites
H4: ✅ Strongest (4/5), with no opposites
H5: ❌ Unfalsifiable from data
H6: ⚠️ Weak (1/5), the participant expressed that he cares about where his money goes after donation, and he needs to follow their social media to find that out.

Key Discovery: The Information Density Paradox

The initial hypothesis assumed less text = more engagement. Testing revealed a tension: reduced text improved scannability but decreased trust. I originally assumed that with less text, people are more willing to read paragraphs, but with reduced text, people are less likely to trust the organization. 3 participants expressed that they want to see the detailed information because they want to see if the organization has real impacts or not.

Additional Findings

Findings on what increases trustworthiness

4 of 5 participants expressed that trustworthiness comes from knowledge and familiarity with the organization, so that they are more willing to donate
1 participant expressed that adding a text that "we won't sell your information" or "we won't spam your email" would increase trustworthiness.
1 participant (SW Engineer) expressed that adding the payment platform name, such as who backs up the payment process, increases trust and credibility.
1 participant (a master's student) expressed that they like the tax-deductible label on the old website, suggesting this may increase persuasion to donate.

Findings on the unexpected constraints of this study

The About page is not available when testing, because one participant expressed that they want to know more about the organization, and wants to click on the About page to know more about them, so they feel safer donating.
Some images are generic images from Pexels/Unsplash; one participant commented that these images look fake.
One A/B testing was conducted after a class; one participant may be tired and give more generic thoughts.
Figma does not allow input; therefore, it's hard to tell if the checkout form would decrease the motivation of the users when they are filling out these forms.