Why "Just Read the Reviews" Doesn't Scale
Most brand sentiment checks start informally: someone skims recent reviews, scrolls a few Reddit threads, maybe checks the replies on a recent post. That works for a rough gut check, but it doesn't hold up as evidence: it's a handful of anecdotes, not a sample, and it's easy to unconsciously cherry-pick the comments that confirm what you already believed.
A proper audit takes the same instinct (go read what people are actually saying) and makes it systematic enough to trust and share. Here is a version that takes under an hour, start to finish.
Step 1: Define the Question (5 minutes)
Before pulling any data, write down the specific question you are answering. "How do people feel about us?" is too broad to action. Better versions:
- "How did the community react to [specific launch/announcement]?"
- "What do people say about us compared to [named competitor]?"
- "Is there a specific complaint pattern behind our recent support ticket increase?"
Step 2: Pick Your Sources (5 minutes)
Match the platform to where the conversation actually happens, not where it is easiest to search:
- Reddit: best for candid, long-form opinion, especially in niche subreddits related to your category.
- YouTube: best for reactions to a specific video (review, launch trailer, tutorial).
- Instagram / TikTok: best for reactions to visual/lifestyle content and influencer posts.
- X (Twitter): best for real-time reaction to news or announcements.
For most brand audits, two sources are enough: one for depth (Reddit or YouTube comments on a relevant video) and one for breadth (X or Instagram around a specific hashtag or mention).
Step 3: Pull a Real Sample, Not a Convenient One (10 minutes)
Pull enough volume that a handful of loud outliers cannot swing the read. A few hundred comments per source is a reasonable floor for a directional read; more for anything you plan to present externally. Avoid only sampling from a single post or thread if the topic spans several; that's the same cherry-picking problem as reading reviews manually, just at higher volume.
Step 4: Classify, Don't Just Skim (15 minutes)
This is the step manual review usually skips because it is tedious at volume: actually scoring each comment as positive, negative, or neutral, and ideally tagging the underlying emotion and topic, not just the polarity. A tool that runs this automatically (Sentyment's Market Intelligence module does this across Reddit, YouTube, Instagram, TikTok, X, and Facebook from a single URL) turns what used to be a day of manual reading into minutes, at 1 credit per Reddit/YouTube URL on the free plan.
Step 5: Look for the Pattern, Not Just the Split (10 minutes)
A 65/20/15 positive/neutral/negative split is a headline number, not an insight. The useful part is what is driving each bucket:
- Which specific aspect keeps showing up in negative comments: price, a specific feature, customer service, shipping?
- Is negative sentiment concentrated in one time window (tied to a specific event) or spread evenly (a chronic issue)?
- Are the most-upvoted or most-liked comments representative of the overall split, or is a single loud comment distorting the visible conversation?
Step 6: Package It With Evidence (10 minutes)
A sentiment percentage without supporting quotes is not persuasive to a stakeholder who was not in the room. Pull three to five representative comments (genuinely representative, not just the most dramatic ones) for each major finding. This is the difference between "sentiment is 68% positive" and a slide someone actually acts on.
What You End Up With
In under an hour: a specific answer to a specific question, backed by a real sample, with the actual language customers used to back it up, reusable in a stakeholder deck, a product review, or a crisis-response decision, and repeatable next quarter to track whether anything changed.
Conclusion
The gap between "eyeballing a few reviews" and "commissioning a research study" used to be wide. It doesn't have to be. A structured, hour-long process gets most of the rigor of the latter without the timeline or budget of either extreme.