Individual comments are customer service. The aggregate is research. It's the cheapest research you will ever have access to, because your audience does the fieldwork and files it in public.
Almost nobody harvests it. Replies get handled one at a time and then forgotten, so the same question gets answered forty times and never gets made into anything.
Here's how to turn the pile into a shortlist.
The four kinds worth logging
Not every comment carries information. Four do.
The repeated question. The single most valuable signal available. If ten people ask the same thing, your content has a hole in it and they've pointed at the exact spot. This is a content brief someone else did the work of discovering.
The correction. Someone who knows more than you did about one detail. Painful, and usually the start of a better piece than the one they corrected.
The unexpected use. "I actually use this for X." You've just learned about a segment you weren't addressing, in their own words.
The objection. "Looks good but I'd never pay that." A real buying objection, stated publicly, unprompted. Sales teams pay for this.
Everything else (praise, emoji, tangents, arguments between strangers) is noise for this purpose. Handle it per comment triage and move on.
Log it in one line
The discipline is not "read comments carefully". It's write it down at the moment you see it, because you will not remember it a week later and there is no searching for a feeling.
One line per item, four fields:
| Field | Example |
|---|---|
| What they said | "how do you stop it rusting in coastal air" |
| Which kind | repeated question |
| Where | YouTube, the workshop tour video |
| Date | so you can spot when something starts |
That's it. A note file, a spreadsheet, whatever you'll actually open. The format matters far less than the habit of capturing at the moment of noticing.
Counting is the whole method
Now the part that separates this from confirmation bias.
One comment is an anecdote. A single person asking something is a person asking something. It might be a great idea; it is not evidence.
Three or more of the same thing, from different people, on different posts, is a pattern. Same-post repetition is weaker: people read each other's comments and echo them.
The reason to be strict here is that comments are a self-selected sample. Commenters are the most engaged fraction of your audience, and the loudest fraction of that fraction. What they want is not automatically what most viewers want. That's fine as long as you're honest about which population you're generalising from. It's tier 1 evidence with an extra selection bias stacked on top.
Turning a pattern into a piece
Once something clears the bar, the conversion is mechanical.
A repeated question becomes a piece that answers it. Directly, in the title, using their words. If eleven people asked "how do you stop it rusting in coastal air", the title is close to that sentence, because that's what the next hundred people will search for.
An objection becomes a piece that addresses it head-on. Not defensively. The bravest content most brands could make is a straight answer to their most common objection, and almost nobody makes it.
An unexpected use becomes a piece for that segment. You've found an audience you weren't serving. Sometimes it's bigger than the one you were.
A correction becomes a follow-up. Publicly getting something right, having got it wrong, is unusually good content: it's proof you're paying attention, which is otherwise hard to demonstrate.
Then it enters the plan like anything else, with the comment log as its why now (see from a pile of ideas to a sequenced plan). "Eleven people asked this across four videos" is one of the strongest why-nows you can write.
Two traps
The loudest commenter. One person who comments on everything, at length, with strong opinions, is not a segment. Their volume is a personality trait, not a signal. Count people, not comments.
Answering in the comments and stopping there. You reply, they're satisfied, and the knowledge dies in a thread nobody will read. The reply is the courtesy. The piece is the point.
Where Acumin fits
The Engagement inbox surfaces only comments with strong signal (strong criticism and substantive praise), which is the right filter for replying and a deliberately narrow one for harvesting. Mild-but-repeated questions are exactly the kind of thing it drops, because they don't need a human urgently.
So the log stays a human habit. What the product does help with is the next step: a pattern you've spotted goes into a Concept Read to get a confidence band before you commit, or straight into a Creative Brief if you're already sure.
Being straight about it: nothing here reads your comments and hands you a content calendar. The counting is judgement, and the selection bias is real enough that it should stay judgement.
How to use this tomorrow
Open your three most-commented posts. Read only the questions. Skip everything else.
Write down each one, one line. You'll almost certainly find the same question twice before you're through the second post, and that's your next piece.
Related: Comment triage covers which comments need a human at all. From a pile of ideas to a sequenced plan is where a confirmed pattern goes next.