Share of Shortlist: the number that replaces ranking

Position ten links used to be the whole story

A search results page gave you a ranked list. Position one got the click; position nine barely existed. Whole disciplines grew up around moving a page from position four to position two, because that move had a predictable, measurable effect on traffic.

An assistant doesn't do that. It gives one paragraph, sometimes one name, occasionally a short shortlist of two or three. There is no position ten to fight your way up from, because there is no list of ten to begin with. The buying moment moved from a ranked list a person scans themselves to a spoken shortlist an assistant has already narrowed down. A number that describes rank no longer describes what actually happened in that moment — you need a different set of numbers, built for a different kind of answer.

One ladder, several questions

VISIBLE measures the same fixed set of Money Prompts every time, and classifies every answer on a simple ladder: Absent, Mentioned, Considered, Recommended, First Choice. That classification is what makes several different readings possible — without it, “were you mentioned” would be the only question anyone could ask, and it's the least useful one.

Whether you're included in the answer at all is the lowest bar, and the one most tools stop at — it's just the Absent, Mentioned and Considered rungs of the ladder, not a metric with its own name. Recommendation Rate measures how often AI actively recommends your brand rather than simply mentioning it, which is a meaningfully harder thing to earn. How often you're presented as the strongest option — the First Choice rung, tracked as a count of first-choice recommendations rather than a rate of its own — is the moment a buyer is most likely to act on directly. And Share of Shortlist measures your share of considered and recommended positions versus tracked competitors: the portion of the shortlist that's yours, rather than a rival's.

Why they have to be read together

A brand can show up often — named, sometimes considered — and still have a weak Recommendation Rate: visible, but rarely put forward. That's a specific, fixable pattern: the assistant knows you exist but isn't confident enough to suggest you outright, usually because the evidence backing you up is thinner than a rival's. Reading how often you're merely named, on its own, would hide that pattern completely and make the brand look healthier than it is.

The reverse also happens: a brand that rarely appears at all but earns a high Recommendation Rate when it does — rare, but it means that whenever the assistant does find enough evidence to include you, it's confident enough to put you forward. That's a discovery problem, not a trust problem, and it needs a different fix again.

Two more numbers worth knowing

Two further metrics fill in the picture rather than compete with it. Discovery Win Rate asks whether buyers can find you without already knowing your name — the same distinction that made Northwind Coffee's two scores so different in the first place. Competitive Movement asks who gained and who lost between one measurement window and the next, which is the only honest way to say a competitor is “winning” — not as a permanent verdict, but as a direction observed over time.

Why this has to be a series, not a snapshot

One run of one question tells you what an assistant said once. Assistants are probabilistic — ask the same thing twice and the wording, sometimes the shortlist itself, can shift. A single measurement can't tell you whether a low score is a real pattern or an unlucky moment, any more than one coin flip can tell you whether a coin is fair.

That's why the useful version of this repeats on a schedule: the same core Money Prompts, asked every month, across the same set of supported assistants. On Autopilot, that means 30 core Money Prompts asked across four supported AI and search surfaces, recurring every month, with a smaller set of discovery prompts free to rotate and catch new questions as the category shifts.

Money Prompts stay frozen on purpose — an instrument that keeps changing what it measures can't tell you whether anything moved.

What repetition buys you that a single pass can't

Repeat sampling on the same questions is what turns a score into a trend line, and a trend line is what lets you tell the difference between a change you made actually working and the ordinary noise of a probabilistic system answering slightly differently each time. It's also the only honest way to claim an improvement — a single before-and-after pair of answers proves nothing on its own; a pattern across several repeated readings does.

What a bad, average and good picture looks like

Numbers only mean something next to competitors. Take the same worked example VISIBLE uses across its own product screens: four coffee roasters, measured on the same commercial questions.

Read that as three readings of the same idea. Blue Bottle at 41% is a good picture — of every recommendation handed out across the four tracked brands, four in ten went to it, more than any rival. Stumptown at 24% is an average one — present, competitive, but not the default answer. Northwind at 18%, third of four and only a point ahead of Counter Culture, is a bad one: mentioned sometimes, rarely the brand an assistant leads with, and well behind the roaster that owns the category.

Traditional SEO measures rankings. VISIBLE measures whether your brand appears, whether it is recommended, and how much of the category conversation belongs to you.

VISIBLE

What's basic and what's full

A free Snapshot gives a basic preview of Share of Shortlist and Recommendation Rate — enough to see the shape of the gap. The full Intelligence Audit expands the same measurement across thirty Money Prompts and five tracked competitors, with a full classification breakdown rather than a preview. The metric doesn't change between the two; the depth and the competitor set it's measured against does.

What the number is for

Share of Shortlist on its own doesn't tell you why a competitor is ahead — that's a separate question about sources and accuracy, covered in the five numbers behind your score. What it does is turn a vague worry — “I think we're losing ground to AI search” — into a specific, trackable figure you can watch move, or fail to move, month over month, against the same competitors, on the same questions, every time.

Running that measurement properly, on your own Money Prompts rather than a generic template, starts with the free check. It won't give you the full monthly series on the first pass — that comes with repeated measurement — but it will show you, for the first time, where you currently sit.

It's also worth being honest about what none of these numbers claim to be: a guarantee. AI-generated answers are dynamic, and a brand that improves its Share of Shortlist this month isn't promised the same result next month, or on a different assistant, or against a competitor that's also changing something. What the numbers give you instead is a way to tell, reliably, whether the thing you did actually moved anything — which is a more modest and considerably more useful promise than a ranking guarantee would be.