Nguyễn Đặng Tuấn Minh
There is a simple truth that becomes clearer to us every year: the entrepreneurial path is not paved with roses; it is built on sharp questions, meaningful data, and many deliberate small stumbles (“cú vấp nhỏ”). When we first began working with Vietnamese entrepreneurs ten years ago, we, too, carried the common romanticism of innovation: that a good idea would naturally find its customers; that persistence was enough. Reality taught us the opposite: without data, there is no learning; without learning, persistence only deepens the mistake.
Looking back, what matters most over the decade is not the number of programs, workshops, or “successful” projects, but the moments when we paused at the right time, narrowed the experiment, reframed the question, and found our map again through small but meaningful pieces of data. These are what we call “lean failures” (thất bại tinh gọn): failing earlier, smaller, documented, and forward-facing.
Why does data—especially qualitative data (dữ liệu định tính)—matter so much?
In the startup world, people talk endlessly about numbers: installs, conversion rates, recurring revenue. They are essential—but they answer what, not why. When a curve doesn’t go the way teams expect, most increase budget or switch channels. Few sit down with real users, ask slowly, listen without interrupting, and rewrite assumptions using everyday language.
We learned that proper qualitative data is not a collection of impressions—it is discipline. Discipline in asking non-leading questions. Discipline in verbatim note-taking, separating “opinions” from “observed behavior.” Discipline in leaving the office often enough to hear the difference between someone who says “I like it” and someone who has actually paid.
Many major pivots came not from dashboards but from direct conversations: comforting a busy mother describing her 12-minute dinner routine; sitting with a building manager to hear invisible inconveniences; calling a churned customer to understand why they left. These fragments are rarely beautiful, but truthful—and when enough of them accumulate, they guide the numbers.
Good questions—and the power of a new one
Not every failure is worth learning from. Only failures grounded in a clear question leave traces of progress. After hundreds of interviews, we gradually abandoned “beautiful but useless” questions like “Do you like this idea?” Instead, questions anchored in past behavior always revealed the truth:
- “When was the last time you faced this problem? What happened?”
- “How did you solve it? How long did it take? What was the real cost?”
- “Why did you choose that approach? Who did you consult?”
- “What’s the best and worst part of your current solution?”
- “If there were a ‘good enough’ temporary fix tomorrow, what must it do first?”
These avoid prediction (often full of illusions) and focus on behavior already paid for. Every detail—a cost, a timestamp, an influencer—is actionable. We shift from “listening to comfort” to listening to decide.
Often, a single new question changes everything. “Who actually pays?” once moved a team from an impossible B2C dream to a viable B2B path. Another time, “If we sold only the strongest component, would customers buy?” unlocked an entirely new revenue line. A new question is frequently the true pivot.
A small framework to maintain interview discipline
We keep simple habits:
- Always go in pairs: one asks, one records. Attention is respect.
- Record verbatim: separate customer words, our interpretation, and new assumptions.
- Avoid interviewing friends—politeness corrupts data.
- Avoid closed questions or future hypotheticals unless tied to immediate commitment (deposit, sign-up, payment info).
- Prefer in-person interviews to surveys. Surveys are convenient but shallow; one hour face-to-face can save months of drifting.
These small practices give qualitative data enough reliability to guide decisions. When data is reliable, failures stay small.
Applying Steve Blank’s Four Steps with local pragmatism
We value Steve Blank’s Customer Development Model not as doctrine but as rhythm:
- Customer Discovery – interviewing for problems & existing solutions; rewriting assumptions using customer language; building “meaningful—valuable—practical” tests.
- Customer Validation – collecting behavioral evidence (deposits, paid trials, repeat use) using innovation accounting instead of vanity metrics.
- Customer Creation – growing moderately by scaling proven behaviors, not by spreading thin out of fear of missing out.
- Company Building – turning lessons into processes, data into reusable knowledge, adaptability into weekly habit.
In Vietnam, step 1 and 2 often merge: teams probe problems while rushing to sell. This is acceptable only if learning and selling remain clearly separated. When the boundary blurs, “beautiful data” drifts you away from reality.
Naming failures precisely
We learned to name failures correctly. “MVP failed” is not enough. We say:
- “wrong payer,”
- “misjudged problem priority,”
- “measured vanity metrics,”
- “messaged without immediate value,”
- “chose a conservative market for a solution requiring education,”
- “overbuilt features unrelated to target behavior.”
Once a failure is named precisely, the next experiment becomes obvious.
Some of the most valuable lessons come from closing projects using data—painful but peaceful. One urban farming team did so after confirming their competitive advantage was not defensible, the market small, and founder time non-scalable. They stopped early to move fast elsewhere. That failure saved them.
Lean + Data: the discipline of “just enough”
In the AI era, data seems cheap and abundant. The temptation is to collect everything. We choose just enough: only what will be used; only what leads to decisions. Every metric ties to an assumption and a branching threshold. Qualitative data becomes our compass: weekly “learning hours” dedicated to rereading customer voices—not to tell inspiring stories, but to change our questions. When questions change, priorities change; when priorities change, products change.
A practical Customer Discovery script (Vietnamese context)
Start with their problem and current solution—not your idea:
- “When did you last face this issue? How severe was it?”
- “How did you handle it? Total cost in time/money/emotion?”
- “Why that option? Who influenced your choice?”
- “Best and worst parts of your solution?”
- “If a ‘good-enough patch’ arrives tomorrow, what must it do first?”
End with a small commitment (deposit, payment info, agreeing to next week’s pilot). If they refuse to pay a tiny cost today, don’t lull yourself with “maybe later.”
Experiment design: failing small, learning big
Each experiment should have:
- one question,
- one signal,
- one decision.
Example: “If we insert service add-on A into one store for two weeks, will 7-day return rate increase ≥20%?” If yes, expand; if no, stop; if slight increase with same complaints, adjust one element and retry.
In multi-stakeholder markets, don’t chase the vague “first customer.” Find the first replicable cluster—a group with similar context, decision-makers, and reasons to pay. You don’t need the whole industry, just one consistent island to build a bridge.
A small promise for those exhausted
When teams are tired, they blame: “the market,” “the team,” “bad luck.” We’re allowed to be tired. But before blaming, ask whether the question was right. Many times, exhaustion comes from carrying the wrong question for too long. Change the question; energy returns.
Failure does not diminish you; silence about failure does. Tell the story with data, better questions, and humility to change beliefs when evidence speaks—that is how we continue.
After ten years, we have learned one thing: love for customers must be greater than love for the product. That love appears not in slogans but in how we listen, record, pause, pivot, and begin again with a better question.
Lean Startup does not glorify failure; it turns failure into building material for knowledge. Every piece of data—number or word—placed correctly, becomes a brick on the path. The path need not be straight; it just needs to move forward. And to move forward, we must learn.
© KisStartup. Any reproduction, citation, or reuse requires clear attribution.
