Direct answer
Four: cold-query presence (do the engines name you when a stranger asks for your category?), description accuracy (when they do, do they describe you correctly?), corroboration count (how many sources you don’t control say the same thing?), and share of the answer (of the firms named, how often are you one?). Traffic, conversion and pipeline can’t detect a discovery problem — they only count buyers who already arrived.
Boards think a discovery problem would show up in the reporting. It won’t. Every metric a marketing team presents is built from people who already made it to the business. The buyer who typed the category into an answer engine, read a paragraph naming three firms that weren’t you, and closed the tab left no session, no bounce, no lost-deal record. He is not in the deck because he was never in the data.
That is why declining discoverability is usually noticed late, and noticed as something else — a soft quarter, a sales team that says the market has gone quiet. The dashboard is not lying. It is answering a question nobody asked it.
What the four metrics measure that the dashboard can’t
Each of the four is taken from outside the business, by asking the same systems a buyer asks and reading what comes back. None requires access to your analytics. That is the point: a measurement of whether strangers can find you cannot be sourced from people who already did.
| Metric | The question it answers | How you read it |
|---|---|---|
| Cold-query presence | Are we named at all when the category is asked about? | Run the unbranded buyer query across the major engines; record named or not named |
| Description accuracy | When we are named, is what’s said about us correct? | Compare the sentence the engine returns against what you actually sell |
| Corroboration count | How many sources we don’t control say the same thing? | Count independent mentions that describe you in your category |
| Share of the answer | Of the firms named, how often are we one of them? | Across a fixed set of queries, your appearances divided by the total |
Cold-query presence is the only binary on the list
You are either in the answer or you are not, and the second state has no gradations. This is the metric to put in front of a board first, because it is the one senior people grasp instantly and cannot argue with. Ten queries, four engines, one column of yes and no. The number that matters is how many cells say no.
A number built from the buyers who arrived cannot tell you anything about the ones who didn’t.
Description accuracy is where most firms are surprised
Being named is not the same as being represented. A machine that describes you with a service you dropped two years ago, or a market you left, is putting you on shortlists you can’t win and off ones you should. This metric costs nothing to read and is often the fastest thing to fix, because it is usually caused by stale material you still control.
Corroboration count is the leading indicator
The first three months of a discoverability decline show up here before they show up anywhere else. Sources outside your control — independent mentions, listings, references that describe you in your category — are what an engine leans on when it decides whether to repeat a claim. When that count stops growing, presence usually follows it down a quarter or two later. It is the same mechanism behind why qualified buyers never discover you while the product keeps improving.
Share of the answer is the competitive one
Presence tells you whether you exist. Share tells you whether you are winning. Fix a set of twenty queries, run them quarterly, and count how often you are among the firms named versus how often a specific competitor is. It converts a vague worry into a trend line, and it is the number that makes a board treat this as a commercial exposure rather than a marketing complaint — which is precisely what declining discoverability costs before revenue drops.
Field note: in the diagnostics I run, the most common pattern is not absence but misdescription — the firm is named, and the sentence attached to it is wrong or three years out of date. Leadership teams almost always find this more alarming than being missing, because it is easier to see the deals it costs. An observation from client work, not a published statistic.
Put them on the same page as the revenue numbers
These four belong beside pipeline, not in a separate marketing appendix, because they move first. Arrival metrics report the consequence; these report the cause, and they do it early enough to be worth acting on. If you want the scope of a formal read rather than a self-run one, the range is set out in what drives the price of a discoverability audit.
- Run ten unbranded buyer queries across the major engines and record, per query, whether you are named.
- Read the sentence each engine attaches to you — is it what you actually sell today?
- Count the sources outside your control that describe you in your category. Track the number quarterly.
- Fix twenty queries and measure how often you appear versus one named competitor.
- Put the four numbers on the same page as pipeline, not in a marketing appendix.
If you want to see the four numbers for your own category before you decide whether they belong on the board pack:
Keep reporting on the buyers who arrived, and the first hard evidence that the others stopped coming will be a revenue number you can no longer do anything about.
Book a free 15-minute session See the proof