There's a mistaken idea that discovery is a long, expensive process, something for big companies with research teams. In practice, the discovery that pays off most is small and frequent: one focused week, before committing months of engineering.
This is the plan I use when I want to go from "I think people need this" to "I watched five people try to use this and I know what to change".
Before you start: the question of the week
Discovery without focus turns into an opinion poll. Start with one question, written as a risk:
- "Would managers of small teams swap their feedback spreadsheet for a tool?"
- "Will clinics accept patients booking on their own through WhatsApp?"
- "Does the store owner understand the report without help?"
One week, one question. If the question doesn't fit in one sentence, it's too big.
Monday: what we already know
Before talking to anyone, gather what already exists:
- Support tickets, sales messages, comments on social media;
- Usage data (if the product already exists): where do people drop off?
- Competitors and alternatives: how is the problem solved today?
At the end of the day, write down the hypotheses the week will test. A good hypothesis says who, which behavior and which signal will confirm or kill it.
Tuesday: interviews
Five 30-minute interviews with people who match the target profile. Five sounds like few, but it's enough for strong patterns to show up, and you'll have more rounds.
The golden rules of a discovery interview:
- Ask about the past, not the future. "Tell me about the last time you had to do X" is worth more than "would you use a product that does X?". Everyone says they would.
- Don't pitch your idea. Not yet. You want to understand the problem, not collect compliments.
- Follow the emotion. When the person sighs, laughs nervously or says "this is a nightmare", dig deeper.
- Look for what's already been tried. People with real pain have already tried to solve it: a spreadsheet, a workaround, an abandoned tool. No attempts is a sign of weak pain.
Wednesday: opportunity map and choice
Pull the notes together and organize them into an opportunity map: the outcome the customer wants at the top, the pains and needs below, grouped. For each opportunity, assess:
| Criterion | Question |
|---|---|
| Frequency | How often does this happen? |
| Intensity | How much does it hurt when it happens? |
| Alternative | How bad is the current solution? |
| Reach | How many people in the target audience feel this? |
Pick one opportunity and sketch two or three different solutions on paper. Having more than one solution keeps you from falling in love with the first idea.
Thursday: prototype
Build the cheapest prototype that lets you test the hypothesis. It's almost never production code:
- Clickable Figma screens with realistic content (no lorem ipsum);
- A sales page with an "I want to try it" button;
- A manual process behind a simple interface (the famous "Wizard of Oz").
The prototype has to be good enough for the person to forget it's a prototype in the first few seconds. Real content makes more difference than visual polish.
Friday: user testing
Five more sessions, now with the prototype. Give a task ("you need to give feedback to someone on your team, show me how you'd do it") and watch in silence. The signals that matter:
- Did the person complete the task without help?
- Where did they hesitate or go back?
- Did they ask "when can I use it?" or "how much does it cost?"? That's the best signal there is.
At the end of the day, the decision: continue, adjust or drop it. All three are good outcomes. Dropping it on Friday costs a week; dropping it after three months of development costs a quarter.
What comes out of the week
- The hypothesis confirmed or killed, with evidence;
- Your customers' literal phrases;
- A tested prototype, with a list of what to change;
- The scope of what's worth building first, and what stays out.
Summary
Discovery is a habit, not a phase. One week per cycle, with a clear question, five conversations and a cheap prototype, completely changes the quality of what reaches engineering. And there's a bonus: when the product launches, you already know exactly which words to use to sell it.
Murilo Anzini
Hands-on writing on engineering, product, data and marketing for people who want to understand, and build, the whole journey of a product.
Write the sentence before the website: positioning for B2B products
If you can't explain in one sentence who your product is for and why it's the best choice, no landing page will do it for you. A practical positioning guide.
Culture isn't a wall of values: it's ritual
The values on the wall say what a company would like to be. The rituals (the conversations, the reviews, what gets celebrated) show what it is. How to turn culture into a recurring practice.