Your own channels tell you what people say when they know you are listening. That is a filtered signal, and the filter removes exactly the material you most need.
The unfiltered version exists in public, in places you do not control and cannot moderate. Learning to read it is the highest-return research skill available to a small team, because it costs nothing but attention.
Where the honest material lives
Roughly in order of signal density:
- Review site free-text. Not the star rating — the paragraphs, and especially the “what do you dislike” field, which is where the specific friction lands.
- Community threads. Reddit, Hacker News, Stack Overflow, industry Slacks and Discords, specialist forums. Someone asking “has anyone used X?” generates a thread of unprompted assessments.
- Competitors' comment sections. Underused and consistently productive. When a competitor announces a feature, their own users appear in the replies explaining what is still missing — a roadmap gap list written by the people who care most.
- Trade and procurement forums. In industrial and services categories these carry more weight than any consumer social platform.
- Support and issue trackers. Where public, a competitor's open issues are an unusually direct list of their weaknesses.
- Job postings and reviews. Employee-side sources describe internal conditions that eventually surface as product and support quality.
How to read it without fooling yourself
Public commentary is biased in known ways, and knowing them makes it usable.
Extremes are over-represented. People write reviews when delighted or furious. The middle is silent. This is why three-star reviews are the most valuable — written by people who use the product, like parts of it, and can articulate the limits without either evangelism or rage.
Recency matters more than volume. Forty complaints about an issue fixed a year ago is history, not a finding. Always weight by date, and check whether a theme is still appearing.
Count, do not impress. One vivid complaint feels like a finding and is not. Track how many independent sources raise a theme before treating it as real. Three independent mentions is a signal; one is an anecdote.
Watch for competitor astroturf, in both directions. Unusually articulate praise appearing in clusters, or a sudden burst of criticism from accounts with no history, deserves discounting.
Preserve the words
The single most common mistake is paraphrasing.
“Users report onboarding difficulties” is a summary that has thrown away the finding. “Took our ops team four days to get the first import working and we still don't know what broke” contains the severity, the affected role, the duration and the underlying problem — and it is usable directly in positioning, in objection handling, and in a product conversation.
Keep verbatim quotes, each with a link to its source. Two reasons: the phrasing is the value, and a quote without a source cannot be checked or reused with confidence.
What to do with what you find
The output should route to specific places:
- Recurring complaints about you → product backlog, with frequency attached.
- Recurring complaints about competitors → discovery questions and battlecard material.
- The vocabulary buyers use → positioning and page copy. Buyers describe their problem in words that rarely match your category label, and matching their words is most of search-intent fit.
- Objections in the wild → objection comebacks, sourced from reality rather than imagination.
- Unexpected use cases → segments you are serving without knowing it.
The limits, stated plainly
Public listening cannot tell you how common a view is among all customers — only among those who wrote something. It skews toward technical, vocal and English-speaking audiences. It underrepresents enterprise buyers, who mostly do not post. And in thin categories there may simply not be enough material to draw a conclusion from, in which case the honest output is “insufficient signal” rather than a confident theme built on four comments.
Treat it as one input with known bias, not as a survey.
Doing it repeatedly
A one-off sweep gives you a snapshot. The value compounds when you repeat it, because you start seeing themes appear, peak and disappear — and a theme that is growing is worth far more attention than one that is merely large.
Quarterly is enough for most categories. Save the searches, keep the previous quotes, and compare.
Our Social Media Audit covers this half deliberately — review write-ups, community threads, competitors' comment sections and trade forums — with every quote linked back to its source rather than summarised into a sentiment score.