From file to useful result
How to ask with this skill installed
Installing the skill gives your AI a method. Your request still has to provide the case-specific facts, constraints and expected output.
1Name the real task
Describe the decision or deliverable, not just the topic.
2Provide the evidence
Add source material, audience, limits and known facts.
3Define done
Set format, quality criteria and checks.
Source example
Replace the brackets with your case
Mistakes to avoid
- Asking hypothetical questions. "Would you use this?" is useless. People say yes to hypotheticals and no to real commitments. Instead: "Would you give me €50 to try the beta for a month?"
- Interviewing the wrong people. Talking to people who won't actually buy your product (e.g., users instead of budget holders). In B2B, the user and the buyer are often different people.
- Confirmation bias. Hearing what you want to hear. Bring a notetaker or record the conversation. Review notes with a skeptical co-founder who wasn't in the room.
- Not running concurrent validation. Waiting months for interview results before doing any quantitative testing. You should be running qualitative and quantitative experiments simultaneously.
- Small sample sizes with big conclusions. 2 interviews don't give you validation. 10 interviews give you a signal. 30+ interviews give you confidence.
- Ignoring churn and dropout. Interview your churned customers too. Understanding why people leave is as important as understanding why people join.
- Not defining success criteria before the experiment. If you don't decide in advance what result would make you proceed vs. pivot, you'll always find a way to interpret ambiguous data as "positive."
- Forgetting the Spanish payment reality. Spanish companies pay late. If you're selling B2B software to Spanish PYMEs, model cash flow accordingly. Don't assume 30-day payment terms — plan for 60-90 days.