> ## Documentation Index
> Fetch the complete documentation index at: https://docs.trybluemoon.com/llms.txt
> Use this file to discover all available pages before exploring further.
# Measure Commercial Reach — Visibility Weighted by Buyer Intent
> Commercial reach re-weights the same AI mentions by how close each tracked question sits to a buying decision, so deliberately unwinnable questions stop dragging your headline number down.
Commercial reach is a second visibility number on your Overview. It counts exactly the same mentions as your headline visibility rate — no extra data, no modelling, no adjustment for anything the AI did not actually say — but it weights each tracked question by how close that question sits to a buying decision.
It exists because a plain mention rate treats every question in your library as equally winnable, and your library is deliberately built so that some of them are not.
The two numbers sit side by side on purpose. Commercial reach does not replace visibility, and neither one is the "real" number. **The gap between them is the insight.**
## How to find it
Navigate to [app.trybluemoon.com](https://app.trybluemoon.com) and open **Overview**. **Commercial reach** is the tile immediately to the right of **Visibility** in the KPI strip at the top of the page.
It respects the Overview filter bar exactly as the visibility tile does — time range (30 days by default), platforms, topics, and personas. Change a filter and both numbers recompute over the same slice of polls.
The tile only appears when the current slice carries commercial signal. If every question in your filtered slice is bottom-tier calibration, the tile is not rendered at all rather than printing `0%`. See [When it shows nothing](#when-it-shows-nothing).
## Why a plain mention rate under-reports
When Bluemoon sets up your account, the topic generator does not produce a flat list of questions about your product. It builds a **four-tier library ordered by buyer proximity**, and each tier has a different job:
Someone is choosing a vendor right now. "best invoicing tool for a 10-person agency". You should appear here. If you do not, that is a commercial problem.
Someone has the problem your product solves but is not yet shopping for the product. You may or may not surface.
Wide category questions where any single vendor appears among dozens, or not at all.
Bare conceptual questions. The generator's own instruction to the model is that a brand should **never** appear here. These questions exist to prove the scoring is not rigged.
The peripheral tier is not a mistake in the library — it is what keeps the whole measurement honest. Without questions you cannot win, a visibility score drifts toward 100% and stops meaning anything. Every account gets them.
But they still land in the denominator of the plain mention rate. Counting questions that were designed to be unwinnable structurally depresses every customer's headline number, and it depresses it by an amount that depends on how much ballast their library happens to carry — which makes the raw number harder to compare across accounts than it looks.
Commercial reach removes that distortion without touching the raw number.
## The four tiers and their weights
Each topic in your account carries a tier, set during onboarding and editable afterwards on the **Topics** page (the **Category** field). Every poll inherits the weight of its topic's tier.
| Tier | Weight | What it represents |
| ------------------------- | ------- | --------------------------------------------------------------------------------------------------------------------- |
| `core` | **1.0** | Buyer-intent questions. Counted at full value. |
| `adjacent` | **0.6** | Problem-led questions adjacent to the purchase. |
| `broad` | **0.3** | Wide industry and trend questions. |
| `peripheral` | **0.0** | Calibration ballast. Contributes nothing — not to the numerator, not to the denominator. |
| *(blank or unrecognised)* | **0.5** | A deliberately neutral fallback, so a missing category can neither silently erase a question nor silently promote it. |
The weight is a **stationary property of the question**. It is assigned from the topic tier and is never derived from what an AI answered. A weight that moved from sweep to sweep would move the denominator of the rate, which would make every week-over-week comparison meaningless.
This is the **tier weighting**, and it is what Commercial reach is computed on today. The per-question intent score described in [Question intent](/guides/question-intent) does not enter this number unless a per-question weighting has been enabled for your account; that variant exists behind a feature flag and is off by default. See [What it does not tell you](#what-it-does-not-tell-you).
## How it is calculated
Both numbers are produced by the same pass over the same set of poll results, so they can never disagree about which polls they saw.
```
visibility = polls where you were mentioned ÷ all polls
commercial reach = Σ weight of polls where you were mentioned
─────────────────────────────────────────
Σ weight of all polls
```
Two things hold for both numbers:
* **Branded questions are excluded from both.** A tracked question that already contains your brand name trivially guarantees a mention, so those polls are dropped before either rate is computed.
* **A mention is a recorded mention.** Both count a poll as a hit when the scoring pipeline recorded any mention of your brand in that answer. Neither number is re-derived at read time.
### A worked example
An invented B2B invoicing tool, 800 polls over 30 days, evenly spread across the four tiers:
| Tier | Weight | Polls | Answers naming the brand |
| ---------- | ------ | ------- | ------------------------ |
| Core | 1.0 | 200 | 60 |
| Adjacent | 0.6 | 200 | 30 |
| Broad | 0.3 | 200 | 10 |
| Peripheral | 0.0 | 200 | 0 |
| **Total** | | **800** | **100** |
**Visibility** = 100 ÷ 800 = **12.5%**
**Commercial reach** = (60 × 1.0 + 30 × 0.6 + 10 × 0.3) ÷ (200 × 1.0 + 200 × 0.6 + 200 × 0.3 + 200 × 0.0)
\= 81 ÷ 380 = **21.3%**
Same 100 mentions. Same 800 polls. The tool is mentioned in roughly one in eight AI answers overall, and in roughly one in five answers weighted toward the moment a buyer is choosing.
## Why it is usually higher than visibility
Commercial reach is normally the larger of the two numbers, and that is arithmetic rather than optimism. Nothing is added to the numerator — the 100 mentions in the example are the same 100 mentions.
What changes is the denominator. The 200 peripheral polls leave it entirely, and the broad and adjacent polls count for a fraction of a poll each. Meanwhile the tier where a brand typically performs best — core — is the only tier that keeps its full value. Removing questions you were never going to win, from a set you are being scored against, raises the ratio.
Read it as *"the calibration ballast has been taken out"*, not as *"the number went up"*.
**It can also be lower.** If your mentions concentrate in broad and peripheral questions and you are largely absent from core, the weighting works against you: the tier that carries the most weight is the tier where you are missing. In the example above, if all 100 mentions had landed in broad questions instead, commercial reach would be 30 ÷ 380 = 7.9% against a plain visibility of 12.5%. That inversion is a finding, not a glitch.
## Reading the gap
It answers "across everything we track, how often does an AI name us?" It is the number that is comparable over time and the one that stays honest because it includes the questions you cannot win.
It answers "and how much of that sits near a buying decision?"
Neither number alone tells you where you stand. Their relationship does.
| What you see | What it means |
| ------------------------------ | ------------------------------------------------------------------------------------------------------------------------------------------------------------------------ |
| Reach clearly above visibility | Your mentions are concentrated in the buyer-intent tiers. The headline number is under-reporting your commercial position, mostly because of ballast in the denominator. |
| Reach close to visibility | Mentions are spread fairly evenly across tiers, or your library carries little peripheral weight. |
| Reach below visibility | You surface in general and educational questions but not where someone is choosing a vendor. This is the case worth investigating first. |
## When it shows nothing
If the total weight of the filtered slice is zero — every poll in it belongs to a peripheral topic — there is no commercial signal to report, and the tile is omitted rather than shown as `0%`.
This is the same discipline used across the product: **"not measured" and "measured as zero" are different claims, and rendering them identically would be a false statement.** Elsewhere in the app an unmeasured rate renders as a dash for the same reason, and a genuinely measured rate that rounds down to zero is shown as `<0.1%` rather than `0%`, because a handful of mentions is not the same finding as none at all.
The most common way to hit an empty tile deliberately is to filter the Overview down to a single peripheral topic. The absence of a number there is the correct answer.
## What it does not tell you
Commercial reach is a weighting and prioritisation signal. It is not a forecast.
* **It does not predict visibility.** The tier weight is assigned from buyer proximity before any polling happens, and it is not a probability of being mentioned. Across the whole library the tiers do separate on measured mention rate — that is why the weighting is worth applying at all — but the tier weight is constant inside a tier by construction, so it says nothing about which questions within one tier you will be named in. Held against the finer per-question intent score, that within-tier relationship measures out as indistinguishable from nothing. Use it to decide what to work on. Do not use it to project what a number will become.
* **It does not explain why you are absent.** Bluemoon measures whether an AI mentioned your brand in an answer. It does not observe whether you publish anything addressing that question shape, and a low commercial reach is not evidence that you do not. Treat a gap as a place to look, never as a diagnosed cause.
* **The weight is per topic, not per question — today.** Every question filed under a topic inherits that topic's tier. A sharply buyer-intent question sitting inside a topic you categorised as `broad` is weighted 0.3 like everything else in that topic. This is the known granularity limit of the tier weighting, and it is what Commercial reach is computed on today. The per-question [intent score](/guides/question-intent) is a separate layer: a per-question weighting built on it exists behind a feature flag that is **off by default**, and it does not drive the Commercial reach tile you see unless it has been switched on for your account. Until then the practical lever is the **Category** field on the Topics page: if a topic's tier is wrong, the weight is wrong for every question inside it.
* **The tile carries no trend.** Unlike visibility, commercial reach currently shows no week-over-week delta and no sparkline. Compare it across time ranges using the Overview filter bar.
* **The weights are fixed.** 1.0 / 0.6 / 0.3 / 0.0 are not configurable per account, so the same scale applies to every domain and the number means the same thing everywhere.
Both numbers are computed from recorded polls only. When a platform fails to answer, the poll is recorded as a failure and simply is not counted — Bluemoon never fills a gap with an assumed result.