Elisabeth SaehrProduct designer, Lausanne
Swiss humanitarian NGO · 2023–2024

A website built around what visitors came to do.

Medair’s site had to explain what the organisation does and earn enough trust for people to give. Research showed visitors got lost in the menu and in the donation process. As product designer and product owner, I led the research, the structure and the tests, and coordinated about 30 people across eight markets.

Bounce rate 59% → 27%, 2023 to 2025
Role
Product designer and product owner: research, wireframes, user testing, workshops, stakeholder coordination. Visual design and development by an external agency I worked with.
Timeline
Research Jan–Feb 2023 → launch April 2024
Platform
Responsive multi-site: 8 markets, 4 languages
Research
927 survey answers · 8 recorded user tests · donor-care interviews · menu tests with about 30 people
The launched Medair home page: the menu Our work, Get involved, About us, News & stories, a red Donate button, the headline “We save and sustain lives in times of crisis”, and a photo with a red “See our impact” card.
The launched site. Medair’s home page after the redesign, live since April 2024.

In 30 seconds

59→27%site-wide bounce rate in Google Analytics 4, 2023 against 2025, after the relaunch.
50→90%of answers “confident I completed the task”, from version 4 of the menu to version 7.
927survey answers and 8 recorded user tests before the first wireframe.
8 · 4markets and languages on one site, each with its own campaigns.
Problem

Half the test participants were confused about what Medair does. All eight found the donation process confusing. The menu followed how Medair is organised, not what visitors came for.

What I did

Ran the research with a UX/UI colleague, built personas and a new structure, wireframed seven versions of the menu and tested them, and kept 30 people across eight markets moving together.

Where it stands

Launched in April 2024 and adopted by the teams. In 2025 the bounce rate was 27%, against 59% in 2023.

The problem

People came ready to give. The site made them work for it.

The brief was simple to say: make sure visitors understand Medair’s mission, and are convinced it is a trustworthy organisation they can contribute to. In early 2023, we tested the existing site to see how far it was from that.

50%were confused about what Medair does8 recorded tests
87%had their first impression spoiled by small bugs8 recorded tests
8/8found the donation process confusing8 recorded tests
87%got lost between a story and the donation8 recorded tests
Why visitors came, survey of 927
21% came to give 13% didn’t find what they came for: 77% of them were looking for a job the rest, something else
One dot = 1% of answers
What the research found, January–February 2023. Three in four test participants also asked for easier navigation, and three in four wanted to see impact and figures.

One visitor in five arrived wanting to give. The menu asked them to know how Medair is organised first.

Research

A research sprint before any wireframe

I ran it with a UX/UI designer colleague in January and February 2023: what visitors say, what they do, and what other organisations do better.

927survey answers from site visitors: why they came, whether they found it.
8 + 1recorded user tests in German and French (ages 25–34 and 35–65), plus one moderated session.
2interviews with the donor-care teams, French and German speaking.
3competitor reviews: World Vision, Swiss Red Cross, charity: water.
3analytics sources reviewed: traffic, heatmaps, session recordings.
4key findings: first impression, donation, stories, and the need for a personal space for donors.
Persona sheets taped to an office window, each with columns of pink, blue and green sticky notes beneath it.
A persona workshop with colleagues. One sheet per persona on the window, sticky notes under each one.
From findings to structure

I turned the findings into six personas and ran workshops with colleagues to build them. They became the test for every menu label: which persona is this for, and would they use this word?

DonorJob seekerLegacy donorMediaAlumniChurch leader

Then a new sitemap, with card sorting and tree testing, and A/B tests.

Navigation

Seven versions of the menu

The navigation is a mega menu: each top-level entry opens a full-width panel with its links on one side and a featured card on the other. So every version had two layers to get right, the top-level labels and what each panel holds. Each version was checked against the personas and reviewed with colleagues; versions 4 and 7 were tested with participants. The top level got shorter and plainer as it went.

1
Our workGet involvedSupport usMedairEmergenciesStay informedDonate
5 menus, 2 buttons. One menu per partner type.
2
Our workGet involvedBring supportAbout usEmergenciesLatestDonate
Monthly giving added. Careers featured.
3
Our workGet involvedBring supportAbout usLatestDonate
Emergencies no longer a top-level entry.
4
Our workGet involvedSupport usAbout usLatestDonate
First test: 50% confident. Events, press, alumni to the footer.
5
Our workSupport us / Get involvedAbout usNewsDonate
Supporting and partnering merged.
6
Our workSupport us / Get involvedAbout usNewsDonate
Same top level, menus reorganised.
7
Our workGet involvedAbout usNews & storiesDonate
Second test: 90% confident.
The top level of each version, from the wireframes. Each entry opens its own mega-menu panel, shown in the decisions below. Search and language sit on every version and are left out here. Row 7 shows the labels as launched. The dots mark the two tested versions.
Key decisions

Three choices that shaped the site

Each one comes from a finding, and each one turned down an option that was easier to agree on inside the organisation.

Decision 1

Name the menus after what visitors want to do.

Chose
Four plain entries: Our work, Get involved, About us, News & stories, each opening a mega-menu panel. Every way to help sits in the Get involved panel, and the partner types are grouped behind its featured card, “Partner with us”.
Rejected
The first structure: one entry per type of partner (foundations, cantons, corporates, churches, schools) and Medair’s own split between sectors, countries and emergencies.
Because
Three in four test participants asked for easier navigation. In the menu test, people said the split between countries, emergencies and what Medair does felt arbitrary, and a participant asked to donate as a foundation wrote that they would look for Donate, not Philanthropy.
Version 1 · “Support us”
Support us
Philanthropy
Foundations
Cantons, municipalities
Corporates
Churches
Schools
Version 7 · “Get involved” mega menu
Get involved
All the ways to get involved
DonateTask 1
Volunteer
Fundraise
Sponsor a project
Pray with us
Partner with usYour organisation shares our commitment to humanitarian aid? Let’s explore how we can work together.

On the left, a visitor has to know which kind of partner they are. On the right, they pick what they want to do. Same content, organised by the visitor’s question.

Decision 2

Put jobs where job seekers look.

Chose
Careers as the featured card of the About us menu, “Work with us”, with its own image and call to action.
Rejected
Careers as one link among six under “Medair”, next to events, ethics and the press centre.
Because
In the survey, 13% of visitors didn’t find what they came for, and 77% of them were looking for a job. Job seekers are one of the six personas.
Visitors who didn’t find what they came for
77% looking for a job23%
13% of all survey answersSurvey, 927 answers
Version 1 · “Medair”
Medair
About
Events
Careers1 of 6
Press centre
Ethics
Contact us
Version 7 · “About us”
About us
About Medair
Leadership
Finance & report
Press centre
Contact us
Work with usHave a significant impact by starting a career with Medair
Decision 3

Keep the donation flow that worked. Adapt it to each market.

Chose
Keep the multi-step flow from earlier work on donations, with a page that tells donors what their gift does in concrete terms. Then adapt it to each market: country, currency and local ways to pay.
Rejected
Redesigning the donation flow along with everything else.
Because
The flow had already been improved before the redesign, and those practices held. The research said donors get overwhelmed easily, a donation can’t be interrupted, and three in four want impact and figures. With eight markets, the risk was in the local variants, not in the steps.
1Amount
2Details
3Address
4Payment
5Confirmation
1. Choose amount
Switzerland ▾CHF · Swiss francs ▾
OnceMonthly
CHF 30CHF 60CHF 120Other
You’ll be offering around 20 meals to people in need.
Next step →
Currency
CHFEURGBPUSD
Ways to pay
CardPayPalTWINTPostFinanceQR-billiDEALBancontactSEPA Direct DebitBank transfer

Same five steps everywhere. What changes per market is the country, the currency and how people pay, plus a tax note where it applies.

Testing and what it changed

From 50% to 90% confident, in two rounds

About 30 people, colleagues among them, worked through the menu prototype online: three tasks, then open questions on what confused them. After each task they said how confident they were that they had completed it.

“How confident are you that you completed this task?”

Version 4
50%
Version 7
90%
  1. Task 1Can you find the “Monthly Giving” page?
  2. Task 2How would you find information on Medair’s emergency response?
  3. Task 3Please navigate to the “Finance & Report” section.

Share of positive answers. Self-reported confidence, not measured task success.

“There is some overlap between the categories, in particular the emergencies/where we work/what we do bits.”

Participant, open answer

“Donate as a foundation – I should look for donate and not for philanthropy or partnership.”

Participant, open answer

“‘News’ as a term for the Latest stories is not very clear, we can find better!”

Participant, open answer

“Perhaps too focused on the internal perspective, not enough on the client, user or donor.”

Review of the results, translated from French
What participants saidWhat changed by version 7
Countries, emergencies and “what we do” feel arbitrarily separated.Sectors left the menu for a “Discover our work” page. Countries stayed, and Emergencies became “Emergency response approach”.Shipped
A foundation would look for Donate, not Philanthropy or Partnership.“All the ways to get involved” leads the menu. Partnerships became one featured card instead of a list of partner types.Shipped
Prayer next to philanthropy “makes it really Christian and most of the philanthropy audience is secular.”“Pray with us” stayed, at the end of Get involved, apart from partnerships.Shipped
“News” reads as a newsletter or press, not as Medair’s stories.Renamed “News & stories”.Shipped
“The donate button feels quite small.”A filled, coloured Donate button on the launched site.Shipped
What didn’t work

Thirty people, eight markets, and every decision took time

The website belonged to everyone: finance for donations, HR for recruitment, CRM, fundraising and marketing, management, and the affiliate offices. Each market ran its own campaigns; the UK’s are not France’s. Add a content writer, translators and proofreaders, and the external agency that designed and built the site.

Adoption went very well, and the launch went without many hitches. But with that many people, everything took time, and it took a year. Roles and responsibilities were never set out clearly enough. And an earlier version of the menu still read, in the review of the test results, as too internal.

Fundraising & marketingCampaigns, stories, donor journeys
ManagementSign-off, priorities
FinanceDonations
HRRecruitment, careers pages
The website8 markets · 4 languagesMe · product designer & owner
CRMDonor data
Affiliate officesEach market, its own campaigns
Content, translation, proofreadingEvery page in four languages
Design & development agencyVisual design and build, external
About 30 people around one website. I was the link between them and the agency.
What I’d do differently

Clarify everyone’s roles and responsibilities at the start: who proposes, who is consulted, who decides, market by market. With that many people involved, it is the fastest way to save months.

My part, and everyone else’s

I owned the product. An agency designed and built it with me.

I worked with an external agency for the visual design and the development. Everything before that was mine: the research, the structure, the wireframes and the tests. So was the coordination around it: working with the agency, and getting each step agreed by the teams and the markets.

  1. Research sprintSurvey, recorded tests, interviews, competitors
  2. Personas and sitemapWorkshops, card sorting, tree testing
  3. Menu reviewed, persona by personaWhere would each persona click?
  4. Two menu tests50% on version 4, then 90% on version 7
  5. LaunchEight markets, four languages
  6. 27% bounce rateGoogle Analytics 4

What I did

  • Research, with a UX/UI designer colleague: survey, recorded tests, interviews, analytics.
  • Structure: personas, sitemap, card sorting, tree testing.
  • Wireframes for the menus, the pages and the donation flow in each market.
  • Testing: the menu tests and A/B tests.
  • Workshops with colleagues, and the coordination of about 30 people.

What the agency and others did

  • Visual design and design system, by the agency, from my wireframes: new typeface, fewer colours, components.
  • The build, by the same agency.
  • Content: a content writer, translators and proofreaders.
  • Market specifics: each affiliate office’s campaigns.
Results

The bounce rate fell by more than half

Both figures come from Google Analytics 4, for the whole site: 59% in 2023, before the relaunch, and 27% in 2025. Other things change in two years too, so I read it as a strong signal for the new site, not as its effect alone.

Bounce rate · Google Analytics 4
59%
27%
2023
2025
The launched Medair home page: the headline “We save and sustain lives in times of crisis”, the menu Our work, Get involved, About us, News & stories, a red Donate button, and a photo with a red “See our impact” card.
59% → 27%bounce rate, 2023 against 2025Google Analytics 4
50% → 90%confident they completed the menu tasksSelf-reported, about 30 people
Adoptedlaunched without many hitches, accepted across the teamsProject review
What the research left open

Where I would go next

I left Medair in 2024. These come straight from the findings the redesign didn’t fully cover.

  1. A personal space for donorsFinding 4: reporting and a personalised space
  2. The path from a story to a donation87% got lost on it in 2023
  3. Measure task success, not only confidenceA tree test on the launched menu
  4. Follow the bounce rate per marketEight markets, eight sets of campaigns
What I learned

Three lessons

An organisation chart makes a poor menu.

Visitors don’t know how Medair is organised; they know what they came to do. It took several versions and outside eyes to see how much internal language the menu had kept.

Confidence is a signal, not proof.

Nine in ten felt sure they had finished the task. Alongside tree testing and analytics, that is convincing. On its own, it isn’t.

On a multi-market site, governance is part of the design.

Every page had several owners. Deciding who decides is as much a deliverable as the wireframes.

Medair · 2023–2024 · Product visuals redrawn from my wireframes. Quotes are participants’ written answers; one is translated from French. People are not named. Next: Pro Juventute →