Interviews excel at depth, buying process, and politically sensitive truths. Public threads excel at language, alternatives, and recurring pains across strangers.
If they disagree, assume both are partially wrong until usage or revenue breaks the tie.
Strengths and blind spots
| Source | Strengths | Blind spots |
|---|---|---|
| Interviews | Procurement, security review, why-chain | Small-n, courtesy bias, sponsor effects |
| Community | Volume, organic phrasing, competitor set | Selection bias, missing silent users |
| Revenue/usage | Ground truth | Lagging; needs enough volume |
Qualitative research methods literature on interview bias and sampling still applies - threads are a non-random sample, not a census.
When interviews mislead
- Polite participants agree your idea is "interesting"
- Individual contributors without budget authority
- Hypothetical futures ("we would use AI for…") with no current spend
- Sponsor present in enterprise calls
Fix: Ask "what did you buy last quarter for this job?" and "who signs?"
When community evidence misleads
- Vocal minority dominates subreddit
- Astroturfing or affiliate spam
- Category rarely discussed publicly (stealth enterprise)
- Meme WTP mistaken for budget (false positives)
Fix: Triangulate with support tickets, churn reasons, and sales loss notes.
Triangulation rules
- Threads say X, interviews say not-X → segment by persona or company size.
- Both agree → higher confidence for messaging; still validate with usage.
- Neither mentions hypothesis → spike test, not full roadmap quarter.
- Revenue contradicts both → trust revenue; revisit segmentation.
Decision matrix
| Question | Start with |
|---|---|
| Why did we lose deal Y? | Interviews + CRM |
| What words do buyers use? | Threads |
| Is problem Z frequent? | Threads + support |
| Will they pay this quarter? | Interviews + pipeline |
| Should we build feature A? | Usage + paying customer calls |
Weekly rhythm
| Day | Activity |
|---|---|
| Mon | Search threads for new phrases |
| Wed | 2 customer or prospect interviews |
| Fri | Reconcile doc: messaging vs roadmap |
Use Search and Trending Problems for public evidence.