What I got wrong about onboarding
PRACTISE 2
A stakeholder, let's call her Mrs. Omáčková, walked in one day and said the e-commerce onboarding wasn't working. Users didn't get it, the numbers were down, and could I please fix it. Deadline? "Yesterday was already late." This is the article where I admit how many things I got wrong before I got anything right.
Speculative
Futurism
Design challenge
IKEA
The brief
Post-rebrand, the onboarding had been repainted into the new visual identity but nobody was sure it actually worked. Analytics suggested it didn't. The design challenge framed itself: how do we improve the onboarding for new users on our e-commerce platform?
I had four methods in mind. Stakeholder Interview, Service Safari, Prototyping, Moderated User Testing. Clean, familiar, nothing fancy.
Service Safari, or: it's worse than I thought
I went through the onboarding myself, as a regular user, writing down everything that confused me. From those notes I built a list of assumptions about where users would probably get stuck.
The list was long. The polite version of my conclusion was that we were doing "really well." The honest version was that the rebrand had given a broken onboarding a fresh coat of paint, and nobody had checked whether the underlying flow still made sense.
Prototyping, or: I thought the prototype was bulletproof
I mapped the entire flow in Figma from start to finish, attached helper questions to each section, annotated the assumed pain points. I felt good about it. I had a coherent story to tell users, clear hypotheses to test, and a clean prototype.
It wasn't bulletproof. It had typos. It had a couple of small interaction bugs I hadn't caught because I was so close to it. I only found these when participants started pointing them out during the actual test, which is the worst possible time to find them.
Recruitment, or: I thought we had contacts
The plan was to recruit four participants from our target audience. I assumed we had a contact database. We didn't, or at least not one I could use cleanly. I ended up scraping together ten participants, partly to compensate for the chaos, and some of them weren't strictly target audience.
I treated the non-target participants as additional assumption-testers rather than primary data, so their input wouldn't distort the qualitative findings. It worked, but only because I caught the recruiting problem early enough to adjust. If I hadn't, the whole research round would have produced confident-looking but misleading results.
The presentation, or: I thought my deck was bulletproof too
I won't dwell on this one. I walked into the stakeholder presentation thinking the findings were clear and the deck was tight. Stakeholders found things to push back on that I hadn't anticipated. Some pushback was fair, some was politics, and I didn't always know which was which in the room.
The specific mistake I made was presenting "X out of Y" percentages from a small qualitative sample. With ten participants, percentages create an illusion of statistical weight that the methodology can't support. I should have framed findings as patterns and quotes, not numbers.
What I'm taking with me
Four things, written down so I actually do them next time.
Start research earlier. "Yesterday was already late" is never going to stop being the answer, so the only defence is starting before someone asks.
Maintain a recruitment database. Scrambling for participants the week of testing is how you end up compromising on who you talk to.
Test the prototype before the test. Run it past a colleague for typos and broken interactions before it ever goes in front of a real participant.
Don't quote percentages from qualitative research. With small samples, share patterns, share quotes, share what you saw. Numbers belong in quantitative work.
The onboarding got better. I got better faster.
© MUNI
(WDX® — 02)
Design
I help companies shape their digital strategy from the ground up. Through audits, service design, and workshop facilitation, I turn complex requirements into clear, functional solutions.
2025 - 2026 ©
Powered by Framer
Alterna