📖 Book
✓ Link checked
Paid
Beginner
Why we picked it
The single best thing ever written on customer conversations. It teaches you to ask about the customer's life and past behaviour, not your idea, so you can't be lied to. If a founder reads one thing before talking to a single customer, it's this.
From
momtestbook.com
by Rob Fitzpatrick
~130 pages
- Talk about their life, not your idea.
- Ask about specifics in the past, not opinions about the future.
- 'That's so cool, I'd totally buy it' is a compliment, not data, dig for commitment and evidence.
Open
momtestbook.com →
📖 Book
✓ Link checked
Paid
Beginner
Why we picked it
This is the short, practical manual for actually doing a test yourself, no training required. Krug walks you through running a simple session in a morning a month, and hands you the scripts, checklists, and a recording setup so you are not inventing the process from scratch. It pairs perfectly with the Nielsen article: one tells you why five users is enough, this one shows you exactly how to sit down and run it.
From
Steve Krug (New Riders)
by Steve Krug
About 168 pages
- You can run a useful test in roughly a morning a month, no lab and no research team needed
- Comes with ready-to-use scripts and checklists so you copy a working process instead of guessing
- Focus on finding the few most important problems, then fix them with the least you can do
Open
sensible.com →
📖 Book
✓ Link checked
Paid
Beginner
Why we picked it
The most accessible, practical intro to usability ever written, you can read it in a weekend and immediately fix your product. The definition of 'make it obvious, not clever.'
From
sensible.com
by Steve Krug
~200 pages
- Self-evident design is the goal, kill anything that adds thinking.
- Users satisfice: they scan and click the first reasonable option.
- Cheap, frequent usability testing beats large formal studies.
Open
sensible.com →
📖 Book
✓ Link checked
Free
Intermediate
Why we picked it
Savoia's pretotyping approach forces you to define, before you show anyone anything, exactly what a real reaction would look like versus a polite one. His XYZ hypothesis format ("at least X% of Y will Z") gives you a measurable line in the sand so you're not relying on your own read of a fuzzy conversation. It's the clearest antidote to a founder convincing themselves a shrug was enthusiasm.
From
pretotyping.org
by Alberto Savoia
~80 pages
- Decide what counts as a real signal before you run the test, not after.
- Watch for what people do (the Z), not what they say they'd do.
- A cheap pretotype that answers the real question beats a polished one that doesn't.
Open
pretotyping.org →
📄 Article
✓ Link checked
Free
Intermediate
Why we picked it
GV's method for building a realistic facade and testing it with five target customers in a single week, which is the interview flavoured cousin of the fake door. Their fake it philosophy, a convincing front is all you need to learn, is the same principle as a landing page test applied face to face. Use it when you want qualitative reactions to a prototype alongside the quantitative click data.
From
GV (Google Ventures)
by Jake Knapp
~15 min read plus resources
- A convincing facade is enough to test; you do not need the real thing
- Five target customers surface most of the problems worth knowing
- Compress research into a week so you decide fast, then move
Open
gv.com →
📄 Article
✓ Link checked
Free
Beginner
Why we picked it
A clean primer if you have never run a test. It defines the three pieces that matter (a facilitator who does not lead, realistic tasks, and participants who look like your real users) and explains moderated versus unmoderated and qualitative versus quantitative in plain terms. Good to read before your first session so you set it up right.
From
Nielsen Norman Group
by Kate Moran
~12 min read
- Give realistic tasks, not questions about opinions
- The facilitator observes and never rescues the user
- Recruit people who resemble your actual target users
Open
nngroup.com →
📄 Article
✓ Link checked
Free
Beginner
Why we picked it
This is the cleanest, most concrete catalogue of the exact question shapes that quietly steer a person to the answer you already want. It names five specific traps (rephrasing what you think you saw, assuming a problem and blaming the user, suggesting the answer, naming interface elements the user never used, and assuming their emotions) and gives a neutral rewrite for each, so you can pattern-match your own script before an interview. It stays practical rather than academic, which is the right altitude when you are about to sit across from a real customer.
From
Nielsen Norman Group
by Amy Schade
about 7 min read
- Leading questions bury the answer you want inside the question, so people mirror your words instead of telling you what actually happened.
- The fixes are specific: swap suggested outcomes ("how well did this save you time?") for open ones ("what was easy or hard about this?"), and never name a feature the user has not named first.
- Watch for questions that assume an emotion or a problem ("when you were struggling..."), because they quietly plant a reality that may not be true for that person.
Open
nngroup.com →
📄 Article
✓ Link checked
Free
Intermediate
Why we picked it
This is a catalogue of the specific ways founders sabotage their own sessions, including talking too much, asking hypothetical questions, and reacting visibly to what the user does. Skim it after your first test and you'll probably recognize at least one mistake you just made.
From
Nielsen Norman Group
- Poor or leading questions produce data you can't actually act on.
- Asking someone to predict their future behavior is unreliable.
- Recruiting the wrong participant undoes everything else you do right in the room.
Open
nngroup.com →
📄 Article
✓ Link checked
Free
Intermediate
Why we picked it
This explains why people behave differently the instant they know you're watching, which is the deeper reason your own presence in the room is already a source of bias before you say a single word. The mitigation tactics here (building rapport, giving people a real task instead of a demo, stepping back physically) are direct, practical fixes.
From
Nielsen Norman Group
- Being watched changes behavior even if the observer never speaks.
- A real goal and enough time in the task reduce self-consciousness.
- Some observer effect is unavoidable; the goal is to shrink it, not eliminate it.
Open
nngroup.com →
📄 Article
✓ Link checked
Free
Beginner
Why we picked it
If a cofounder or teammate is going to sit in on the session, this is the two-minute brief you send them beforehand. It spells out exactly what an observer should never do (comment, correct, answer questions, react visibly), which is often where a supposedly neutral test quietly gets ruined by the person who isn't even running it.
From
Nielsen Norman Group
- Observers should not talk to or correct the participant, ever.
- Unplanned comments from the room can leak clues just as easily as the moderator's.
- Brief every observer on the rules before the session, not during it.
Open
nngroup.com →
📄 Article
✓ Link checked
Free
Beginner
Why we picked it
This is the piece that gives you permission to stop waiting for a big study. Nielsen shows that just five people uncover roughly 85 percent of the usability problems in your app, and that running several small tests beats one large one because you can fix things between rounds. For a founder with no budget, it reframes testing from a scary research project into something you can do this week.
From
Nielsen Norman Group
by Jakob Nielsen
About a 10 minute read
- Five users find about 85 percent of usability problems, so you do not need a large sample to learn what is broken
- Running three small tests and fixing between them beats a single big study
- The rule holds for qualitative testing of a fairly similar user group, so split into small groups if your users are very different
Open
nngroup.com →
📄 Article
✓ Link checked
Free
Advanced
Why we picked it
This names biases beyond the obvious one: task-selection bias (users assume whatever you ask them to find must exist), social desirability, and the effect of paying someone to participate. It's a good gut check after a few sessions, when you start wondering why the data still feels off even though you kept quiet.
From
MeasuringU
by Jeff Sauro
- Users assume a task is possible just because you asked them to attempt it.
- People tend to under-report criticism, especially face to face.
- Paying or thanking participants can subtly push them toward pleasing you.
Open
measuringu.com →
📄 Article
✓ Link checked
Free
Beginner
Why we picked it
A founder-friendly crash course on planning and running discovery interviews, with tactics from researchers at Zoom, Zapier, and Dropbox. It is especially good on picking the one or two things you must learn and writing questions that do not lead the answer. Concrete and skimmable.
From
First Round Review
by Jane Davis
20 min read
- Decide the two things you must learn before the call.
- Reframe important questions from several angles.
- Check your own bias before interpreting answers.
Open
review.firstround.com →
📄 Article
✓ Link checked
Free
Beginner
Why we picked it
This lays out the practical mechanics of a feedback session, including being neutral when you present the prototype and adapting your questions as you go without leading the witness. It's a good primer before your first few tests, especially if you're coming from a design-thinking background rather than a UX research one.
From
Interaction Design Foundation
- Present your prototype the same neutral way to every participant.
- Let participants add their own ideas instead of only reacting to yours.
- Actually change your next prototype based on what you saw, don't just collect notes.
Open
ixdf.org →
📄 Article
✓ Link checked
Free
Beginner
Why we picked it
Useful once you've run a few in-person sessions and want to test with more people than you can personally sit with. It's honest about the tradeoff: unmoderated tests remove you (and your bias) from the room entirely, but you lose the ability to probe follow-up questions, so you need a tighter, clearer script going in.
From
Maze
- Unmoderated testing removes the moderator's bias entirely, at the cost of follow-up questions.
- Write your task instructions extra carefully since no one is there to clarify them.
- Pilot an unmoderated test on one or two people before sending it to everyone.
Open
maze.co →
📄 Article
✓ Link checked
Freemium
Intermediate
Why we picked it
Testing with the wrong five people teaches you nothing, so this piece helps you decide who your five should actually be. Margolis lays out a one-day approach to find the specific customer most likely to adopt now. It sharpens the recruiting side of your usability test so the feedback is worth acting on.
From
Lenny's Newsletter
by Michael Margolis
- Narrow to the subset of users most likely to adopt today.
- A tight bullseye makes five sessions far more decisive.
- You can find and interview them in a single focused day.
Open
lennysnewsletter.com →
Why we picked it
A short, practical field guide to customer discovery: who to talk to, how to find them, and how many. Pairs perfectly with The Mom Test, Fitzpatrick teaches the questions, Constable teaches the logistics.
From
talkingtohumans.com
by Giff Constable
~80 pages (free PDF available)
- Decide who you need to learn from before you start.
- Recruiting interviewees is a skill, scripts and channels included.
- Get out of the building; patterns emerge faster than you expect.
Open
talkingtohumans.com →
