A lightweight scoring method that preserves context, recurrence, behavior, urgency, and uncertainty instead of hiding judgment behind a magic number.
Start with the real decision
A single score can help a founder compare many observations, but it becomes dangerous when it hides missing context. Five vague complaints should not automatically outrank one detailed switching story.
A demand score is a review queue, not a revenue forecast.
A practical sequence
- Score specificity. Name the person, job, trigger, current workflow, and consequence.
- Check recurrence. Require independent authors, dates, and sources; copied claims are one origin.
- Look for commitment. Time, money, workarounds, and internal effort outweigh stated enthusiasm.
- Identify urgency. Deadlines, renewals, broken processes, and repeated failures make delay costly.
- Test product fit. Use only capabilities that exist today and a job the product can credibly improve.
Keep the evidence reviewable
For every retained item, record the original source and date, the customer situation, the behavior actually observed, your interpretation, and the decision it may change. Keep observation separate from inference: “the buyer described a migration” is observed; “this account has strong intent” is an inference. For consequential conclusions, add the evidence that would falsify them. This is not paperwork for its own sake. It prevents a fluent conclusion from surviving after its scope and context have been forgotten.
Recurrence also requires deduplication. Several posts may quote one announcement; one story may be reposted across platforms; multiple complaints may belong to the same event. Mark shared origins before deciding whether you see a stable pattern, a concentrated reaction, or temporary noise.
Common ways the method fails
Engagement worship. Likes may reflect reach or controversy, not demand.
Precision theatre. Extra columns do not create truth if they never change priority.
No recalibration. Compare old high scores with later behavior and revise the rubric.
Separate research, judgment, and outreach
A public discussion can reveal jobs, language, and alternatives, but it does not grant permission to build a personal dossier or send an unsolicited pitch. Finish the research judgment first, then decide separately whether participation is appropriate. If you participate, follow community rules, disclose relevant interests, and make the answer useful even without a product link.
SeeVoid’s role is to discover, group, and preserve source context so founders repeat less search work. It does not decide whether a discussion is representative, and it does not turn a high-scoring signal into automatic public posting. The next action should also consider interviews, product behavior, and the current business boundary.
End with a decision card, not another bookmark
Write the current decision, three strongest pieces of evidence, one counterexample, what remains unknown, the owner, the review date, and the trigger that would reopen the choice. Make the next step proportional to evidence: weak signals justify another interview, medium signals may justify a page or workflow test, and heavier product or market investment needs independent recurrence plus real behavior.
Stop when additional material no longer changes the decision. Continuous research should not create continuous anxiety; it should shorten the path to knowing what to do, what not to do yet, and why.
What completion looks like
Review the top five signals weekly and explain why each sits above the sixth. If the answer is only “the total is higher,” the model has become a black box.
Further reading:U.S. Small Business Administration market research guide.