Why mention counts lie

Every listening tool opens on the same number. Mentions this week, with an arrow and a percentage next to it. It is the first thing on the dashboard because it is the easiest thing to compute, and it is the first thing a CMO stops trusting.

Here is the problem with counting mentions: the count moves for reasons that have nothing to do with whether your brand is doing well. A competitor's outage sends people to your subreddit to compare notes. A trade publication runs a listicle and forty accounts repost it within an hour. Someone with a large following makes a joke using your product name as a verb. All three spike the number. Only one of them is worth a meeting.

Volume tells you that something happened. It will not tell you what, and it will never tell you whether it was good.

The three questions a number has to survive

Before any figure goes in front of an executive, it should answer three things the raw count cannot.

Which direction is it moving in? A week with 400 mentions at 70 percent positive is a different business than a week with 400 mentions at 30 percent positive. Same number, opposite meaning. Sentiment is not a vanity layer on top of volume — it is the thing that makes volume legible at all.

Who is saying it? Ten posts from people evaluating tools in your category are worth more than a thousand from accounts that will never buy anything. Reach is not the same as relevance, and the two are frequently inversely related.

Is anyone actually shopping? A mention says your name came up. A buyer signal says somebody is trying to solve the problem you solve, right now, in public. Those are different objects and they belong in different piles.

LinkedInBuyer 84

We are reviewing our listening stack before the new budget year. Open to suggestions.

What replaces the count

The honest answer is: a paragraph. Not a number.

What a marketing lead needs on a Monday is two or three sentences saying what changed since last Monday, why it changed, and which conversations to read to check the claim. That is a summary, and until recently it was a job someone did by hand every week with forty browser tabs open.

This is the part worth automating. Not the counting — the reading. Robin writes the summary, cites the posts behind each claim, and keeps the underlying conversations one click away, so an insight is something you can open and check rather than something you have to take on trust.

The test

Try this on your current reporting. Take last month's headline number to whoever owns the budget and ask them what decision it should change.

If the honest answer is "none", you are not measuring brand health. You are measuring how much noise the internet made near your name, which is a real thing, just not a thing anybody can act on.

The useful version of that same report names the three conversations that mattered, says what each one is evidence of, and links to them. It is shorter, it is harder to produce by hand, and it is the only one that survives contact with a decision.