Observe Users, Don't Just Ask Them
Watch what users actually do, because watching your assumptions get crushed beats any interview
- Difficulty
- Easy
- Time to result
- ~days to results
- Steps
- 2
- Confidence
- 85%
Barrios argues the standard advice to 'talk to users' works only sometimes; what works best is observing them directly in research sessions and watching them use the product. Observation surfaces the false assumptions baked into your design that users will never articulate in an interview.
Origin
Sebastian Barrios contrasting Mercado Libre's practice with the ubiquitous 'talk to your users' advice.
Core principles
- 01Asking users what they want yields incomplete or misleading answers
- 02Observation reveals the assumptions your design silently makes
- 03Users adapt to broken designs rather than telling you they're broken
How to run it
- 1
Prefer observation over interviews
Run user research sessions where you watch people use the product rather than only asking them what they want or need.
Pro tip Treat observation as the primary signal and interviews as secondary.
Watch out 'Users will tell you what they want' only works sometimes; don't rely on stated preference alone.
- 2
Watch your assumptions get crushed
Look specifically for moments where a user's behavior contradicts a design assumption you didn't know you'd made.
Pro tip The cringe of seeing an assumption break in real time is the most valuable part of the session.
In the wild
Barrios describes a field labeled 'email, name' where users, unsure whether to enter their email or username, type both into the single field because that is literally what the label asked for — crushing the designer's assumption in front of their eyes.
→ Direct observation exposed a design flaw that no user interview would have surfaced, letting the team fix the ambiguity.
Common mistakes
Relying on what users say they want
Interviews capture stated preferences that are often wrong or incomplete; users can't reliably predict or report their own behavior, so asking alone misses real usability failures.
Assuming a design is clear because it's clear to you
Designers bake in assumptions (like a field being obviously for email) that users interpret differently; only watching real usage reveals the gap.
Is it for you?
Best for
Product teams and founder-engineers validating flows and interfaces before scaling them
Not ideal for
Early exploratory discovery where you need to understand motivations and context that observation alone can't reveal
From the transcript
“we like to observe users more than talk to users”
“what works best is just to observe them”
“you cringe at all the assumptions that that are being crushed before your eyes as as someone types in their email and uh their name…”
From the episode
How Mercado Libre built Latin America's most valuable company: 18k engineers, 30k deploys a day, and their own fleet of planes
Sebastian Barrios