A report lands in your inbox. Near the top, in bold: "Estimated reach: 4.2 million people." It's a big, satisfying number. Nobody in the meeting asks where it came from, and the report doesn't say. That gap — between a confident figure and an invisible method — is exactly what keeps social listening metrics from ever getting explained properly. Before you trust any number in any dashboard, it helps to know which of three very different buckets it belongs to.
The three kinds of numbers in every listening report
Every metric a social listening tool shows you falls into one of three categories, and they are not equally trustworthy.
- Counted. Real mentions. Each one is a post, an article, a video or a comment that actually exists, with a link you can click to go check it yourself. If a number is counted, you can recount it by hand and land on roughly the same answer. It's the only category you can audit without just trusting the vendor.
- Modeled. Estimated reach, potential audience, projected impressions. These are legitimate only when the vendor publishes the method behind them and an error margin — the same way a political poll discloses "±3 points, 95% confidence." Across the social listening category, that disclosure almost never happens. You get the number. You don't get the model.
- Branded. Dollar figures like AVE or EMV, proprietary "visibility scores," "presence indexes" — numbers built on a formula the vendor owns and doesn't show you. They look precise (two decimal places help with that), but there's no way to check them, because checking them was never the point. The point is the big number on the slide.
Counted numbers earn trust on their own. Modeled numbers can earn it too, if the method is public. Branded numbers are built to look like the first two while behaving like neither.
AVE: the number the industry keeps selling anyway
AVE — advertising value equivalency, sometimes labeled EMV, "earned media value" — takes a mention and converts it into what it would have cost as a paid ad. A story on a major outlet becomes "$40,000 in equivalent advertising." It sounds like a real number because it's expressed in a real currency.
It isn't real. AMEC, the international association for communication measurement and evaluation, has rejected AVE since it published the Barcelona Principles in 2010 — and reaffirmed that rejection in the 2.0 and 3.0 updates since. The core objection is simple: earned coverage and paid advertising are different things with different value, and multiplying a mention by an ad-rate card doesn't produce a meaningful number. It produces a number that looks meaningful.
So why does AVE still show up in listening dashboards across the industry, more than a decade after the standard-setting body said to stop? Because it works — not as measurement, as a renewal driver. A dollar figure is easy to screenshot, easy to drop into a slide for a client or a boss, and easy to watch grow month over month. Nobody renews a subscription over "we found 340 mentions with neutral sentiment." People renew over "we generated $2.1M in equivalent value." The number doesn't have to be sound. It has to be shareable.
Our own case, with the method shown
We didn't want to make this argument in the abstract, so we ran a direct test: our own product against a well-known global listening tool, same keyword, same window, same kind of count, both panels open side by side.
- Keyword: "jujuy" — a province in northern Argentina.
- Window: 14 days, July 2026.
- Method: same days, same keyword, direct count read from each panel, no adjustments.
The global tool reported 10,955 mentions. murmura360 reported 585. On volume alone, that looks like a rout — close to 19 times more mentions for the global tool.
Then we looked at where those mentions were coming from. Over a 30-day window, the global tool showed only 3 mentions from jujuy360.com — the leading news outlet in the very province the keyword names. A tool listening for "jujuy" that barely hears the outlet covering Jujuy every day isn't a rounding error, it's a blind spot. And on podcasts, the pattern flipped entirely: murmura360 found 24, the global tool found 4.
Put side by side, the story isn't "one tool beat the other." It's that raw volume and real coverage are not the same measurement:
| Metric (14-day window, keyword "jujuy") | Global tool | murmura360 |
|---|---|---|
| Total mentions | 10,955 | 585 |
| Mentions from the province's leading outlet (30 days) | 3 | — |
| Podcast mentions | 4 | 24 |
A tool can report ten times the mentions and still miss the one source your audience actually reads. Volume is not coverage.
Five questions to ask any social listening tool
You don't need a full audit to protect yourself from a hollow report. Ask these five questions — of us, or of anyone else in the category:
- Can I click through? Every counted mention should link to the original post. If it doesn't, it isn't counted — it's asserted.
- Where's the method for "estimated reach"? Ask for the published formula and an error margin, not just the number.
- Is there a dollar figure anywhere in the report? If so, ask what it's built on. If the honest answer is "an ad-rate multiplier," you now know what AVE looks like wearing a different label.
- What happens with Meta? Facebook and Instagram require page-owner permissions to monitor properly. No listening tool — including ours — can see inside Meta without them. Anyone claiming full Meta coverage without that access is describing something that doesn't exist.
- Did it check the source that matters most to you, specifically? Not the aggregate count — the one or two outlets, podcasts or accounts your actual audience follows. A tool that wins on total volume can still lose on the source you actually care about.
How murmura360 does it
We built murmura360 around the answer to those five questions, not around a bigger headline number.
Every metric in the product shows its method next to the figure — no unexplained "estimated reach" sitting alone on a card. There is no AVE anywhere in murmura360, no dollar-value conversion of a mention into hypothetical ad spend. If we can't show you how a number was built, we don't show you the number.
We also tell you what we can't see, instead of implying otherwise: Meta shows as disconnected in every listening tool, including ours, until a page owner grants access. No vendor actually gets around that, whatever a sales page suggests.
What we do count is real: spoken mentions inside YouTube videos and podcasts, timestamped to the exact minute someone says the name, with a direct link to that moment; new sources discovered automatically as the system learns which outlets actually cover a topic; sentiment and emotion read by our own in-house AI, running on our own servers; Google News links resolved back to the real publisher; alerts by email or Slack; PDF and Excel reports carrying your own logo; an API and MCP for teams that want to build on top of it; and monitoring in any language, with an interface available in both Spanish and English — which is exactly why global, English-first tools tend to miss what's said about you in Spanish.
None of that needs a dollar figure to make its case. The method is the pitch.
If a number in your report can't tell you how it was built, it isn't a metric. It's a guess wearing a metric's clothes.
If you want to see what a listening report looks like when every number shows its work, try murmura360 free for 14 days — no card required.