The polite mask
People put on their nicest face for the brand and say what sounds good, not what they really think.
How we avoided it
- Present ourselves as independent researchers
- “We test the website, not you”
- Trust behaviour over opinions
Problem
A new homepage and design system looked right internally, but had never been tested with customers before rollout.
Approach
6 moderated interviews on a live, non-public build, over 3 months, with customers from every segment.
Role
I helped shape the protocol and the panel, moderated some sessions and scribed others, then reviewed every recording, built the analysis and the client presentation.
Outcome
The new design direction was adopted based on positive user feedback, highlighting its potential to go beyond the shopping experience by encouraging exploration, discovery, and engagement with new types of content.
Chapter 01
Nespresso's homepage is where its premium promise first meets customers online.
Nespresso France was preparing a new one, with an enhanced design system, to elevate the digital experience and open the site to content beyond e-commerce: curated coffees, capsule recycling, masterclasses.
The team believed in it. But to get the green light from the decision makers, the project's internal sponsors needed hard evidence from real customers.
The real challenge was not the design, but the decision: leadership would only back it if customers did. So we tested it with them, on a non-public version of the site.
What leadership needed to see
Would customers understand the key messages and benefits the page put forward?
Would the design still feel premium, unmistakably Nespresso?
Would users notice the new content, and if they did, would they care?
Before
After


Homepage: before vs. after
Chapter 02
We needed to see what customers did on the new homepage, and to understand why. Moderated interviews gave us both: the moderator could follow up the moment someone hesitated, and people speak more openly in a real conversation than in an unmoderated test.
The website is where almost every Nespresso customer meets the brand, so we wanted every segment in the room. Using the client’s CRM, we recruited across four criteria:
Panel
6 participants
Recruited from the client CRM, with at least one customer per segment.
Geography
Paris and beyond
Customers from Paris and from other regions, to avoid a Paris-only view.
Segments
How they shop
Ambassadors and Connoisseurs, buying online only or both online and in boutique.
Devices
Mobile and desktop
Participants joined on their own phone or computer, so we saw both experiences.
As consultants, part of our job is to tell clients how far they can trust the results. User interviews come with two well-known biases, so we designed the sessions to keep them in check:
People put on their nicest face for the brand and say what sounds good, not what they really think.
How we avoided it
People nod along with whatever the interviewer suggests, even when they don’t really agree.
How we avoided it
A good interview starts before the call. Each participant went through the same preparation, so nobody lost time on logistics on the day.
A pilot session first
Before the first interview, we ran a test session with colleagues and the client team to rehearse the guide and the tasks, and fix anything unclear.
The session, set up the same way every time
Every Teams meeting was configured identically: recording on so the session could be analysed, transcript on so quotes were easy to find and tag, and the lobby enabled so participants were admitted one by one.
Consent and data, handled from the start
Participants were told the session would be recorded three times: in the invitation, in the confirmation email and at the start of the call. Recordings stayed on the company’s Google Drive during the analysis. Transcripts were anonymised: no names or personal details, only age, gender and city.
Chapter 03
I switched between two roles: I moderated some sessions and was the scribe in others.
Moderating meant keeping the conversation natural while sticking to the guide and the timing. Scribing meant watching what participants did, said and showed on their faces, and writing it all down.
We welcomed each participant, explained what the session was for, and reminded them we were testing the website, not them. We also told them how to claim their reward afterwards.
Participants saw the homepage for five seconds, then told us what they remembered, how it looked to them and what they thought the page was for. A quick way to check whether the key messages landed at first glance.
We built five use cases and gave each participant two that matched their segment and how they used the site. They talked us through every step while we watched for hesitations, wrong turns, moments of surprise and what caught their eye along the way.
Example tasks
To close, participants rated the whole experience on pairs of opposite words: simple or complicated, beautiful or ugly, clear or confusing, premium or low-end, captivating or boring.
Chapter 04
Then came the analysis, and turning it into a story the client could act on.

I reviewed the session recordings to catch the details that notes miss: hesitations, exact wording, the moment someone gave up on a path.

All observations went into a structured spreadsheet, one row per insight, tagged by participant and segment, so nothing lived only in someone's memory.

With the analysis already shaped by goal, building the client presentation was a matter of telling the story: each slide had its findings and its quotes waiting.
Chapter 05
Clarity
4/6
participants
Customers understood the page right away and read it as an invitation to discover new coffees and buy.
The page is intuitive at first glance. It makes me want to buy.
Brand
4/6
participants
The new look felt elegant and premium: simple, beautiful, clear, tasteful. One participant found the pages too alike.
This new homepage is much nicer to look at.
Engagement
2/2
missed the offers
Participants wanted to explore, but the new content was hard to find from the homepage: offers, the subscription and masterclasses.
On the homepage, you can’t tell there are any offers at first glance.
Offers weren’t visible on the homepage.
Add a dedicated offers block high on the homepage.
Implemented
The subscription felt hidden, and some didn’t know it existed.
Give the subscription its own homepage push that explains the benefit.
Adapted
Buying compostable coffee felt long, with too much text before ordering.
Put a direct order button at the top of the Compostables page.
Implemented
Masterclass information was hard to find.
Link the masterclass details and booking from the chef and new-range content.
Adapted
Six interviews is a small sample, so we read the results with a few caveats in mind.
Everyone came from Nespresso’s customer base, and most had been buying for years. People who already like a brand tend to go easy on it.
Mitigation
We recruited across segments and paid more attention to what people did than to what they said.
The new homepage later went live in other countries, where shopping habits may be different.
Mitigation
We mixed participants from Paris and the regions, on both mobile and desktop.
We tested on a staging version. It was close to the real thing, but a generic error stopped two participants partway through a task.
Mitigation
Every session was built around real tasks, like ordering coffee or finding an offer, to keep it as close to everyday use as we could.
Chapter 06
Customers confirmed what the team believed: the new homepage felt more elegant and more Nespresso. That gave the sponsors the evidence leadership needed to approve the rollout.
“It’s inspiring, it makes you want to go and see what’s going on.”
Chapter 07
Three things I’m taking from this project into the next one.
UX research is a hard sell
Clients usually expect something they can see from a consultant: screens, prototypes, a new interface. Research delivers insights instead, and insights are easy to overlook or cut from the budget. This project was the exception: the research was the deliverable, and it gave the project's sponsors the proof they needed to convince leadership.
I should have used an analysis tool
Every transcript was reviewed, tagged and sorted by hand. The client's data policy kept recordings on the company's Google Drive, which ruled out uploading them to a third-party tool. It worked, but it took hours a dedicated tool would have saved. Next time, I'll get that question cleared before the project starts.
My NN/g training paid off
I'm a self-taught designer, but I trained in qualitative research with Nielsen Norman Group. It made this project far easier to plan and run: I knew how to write a neutral interview guide, moderate without leading, and turn hours of recordings into findings, instead of going in blind.