peaksly explains · 08

Zero-click

How do you assess AI visibility without clicks?

An AI answer names a brand, answers the question, and nobody clicks through. This visibility leaves no trace in web analytics. It can still be assessed: in the answer itself, with a denominator, across the same search prompts and several points in time.

6-minute readUpdated 5 Sep 2026with a fictional example

Answer to prompt 9 of 16 · Auwald Dämmstoffe · fictional and shortenedFictional teaching example, not a measurement

How thick does wood-fibre insulation in a pitched roof need to be?

For a pitched roof, 160 to 240 millimetres of wood-fibre insulation are common depending on the target value; what matters is the planned insulation value of the whole build-up.¹ Boards in these thicknesses are offered by Tannhof, Auwald² and Bergfaser, among others.

1daemmen-magazin.example/holzfaser-steildachtrade medium
2auwald.example/steildachown page

1Answer read
2Question answered
3No click, no visit

In the example the path ends here: the own page is linked and is not visited. The mention remains visible all the same.

Observation

Mention: Yes
Position: 2 of 3 named suppliers
Source of the mention: own pageauwald.example/steildach
Visit: not part of the observation

What is counted is what the answer says, not what happens afterwards.

All names, addresses, answers and figures are fictional; domains end in .example. The observation shows a way of reading, not a measurement and not a promise of results.

In short

What is counted is the answer, not the click.

AI visibility without clicks is assessed in the answer itself: whether the brand is named, at which position, with which source and in which context, and how often across the same prompts and several points in time. In the example, 20 of 48 observations name Auwald Dämmstoffe; web analytics shows 4 visits from AI answer surfaces in the same period. Both figures are correct and count different things.

Step 1 · Two countersFictional example: Auwald Dämmstoffe

Web analytics counts visits, not mentions.

A click from an AI answer shows up in web analytics, if at all, as a referral from the respective platform. For overlays inside classic search it cannot be separated from a normal search click. A mention without a click leaves no trace at all.

In the example, 4 visits from AI answer surfaces in nine weeks stand next to 20 mentions in 48 observations. Neither figure follows from the other: four visits say nothing about how often the brand was named, and twenty mentions say nothing about visits.

Both figures are correct. They simply count different things: one counter visits to the page, the other answers on an answer surface.

Web analytics · auwald.example · 9 weeks

4 visits with a referral from AI answer surfaces

counts only clicked referrals
overlays inside search cannot be separated
mention without a click: no trace

Observation · 16 prompts × 3 points in time

20 of 48 observations name Auwald Dämmstoffe

read from the answer, not from the visit
per point in time 6, 7 and 7 of 16
holds whether anyone clicks or not

Step 2 · Four valuesFictional example: Auwald Dämmstoffe

What is assessed is what the answer says.

Without a click, the answer itself remains the evidence. Four values can be read from it: whether the brand is named, at which position among the named suppliers, which source is attached to the mention and which statement the brand is linked to.

The fourth value is the context of the mention. In the example the answer links Auwald to boards up to 240 millimetres; the range has extended to 280 since spring. The mention is correct, the statement is outdated. That is exactly what no click count could reveal.

Every value carries its denominator. Position 2 of 3 means: among three named suppliers, in this one answer. How the first three values come about is explained in the fundamentals article.

Answer read · prompt 9 of 16 · one point in timefictional

Boards in these thicknesses are offered by Tannhof, AuwaldMention and Bergfaser, among others.Position 2 of 3

Footnote 2 at the mention: auwald.example/steildachSource: own page

“Boards in these thicknesses”, that is up to 240 millimetresContext: outdated, page states 280

Four values · one point in time

Mention: Yes
Position: 2 of 3 named suppliers
Source: own page, auwald.example/steildach
Context: “up to 240 mm”, page states 280: outdated

All four values come from the answer. No value needs a click.

Step 3 · Denominator and points in timeFictional example: Auwald Dämmstoffe

One mention is a find. 20 of 48 are a finding.

Without click counts, the denominator decides how much a figure says: the same 16 prompts, the same answer surface, the same measurement path, three points in time four weeks apart. Every cell is an observation, every observation its own row.

In the example, 6, 7 and 7 of 16 answers name the brand. Four prompts name it at all three points in time, six never, six sometimes. That is neither a curve nor a trend, but three counts with a denominator that can be compared.

Three points are not a line. Only repeated measurement over a longer period shows whether a pattern holds; how that happens continuously is described by the monitoring.

16 prompts × 3 points in time · one answer surface, direct API · denominator per row 16fictional

12345678910111213141516

Week 16 of 16

Week 57 of 16

Week 97 of 16

no mentionmentionat all three points in timecolumns = prompts 1 to 16, hover or tap shows the prompt

Count · 16 prompts × 3 points in time

Mention: 20 of 48 observations
Per point in time: 6, 7 and 7 of 16
Stable: 4 of 16 prompts at all three points in time
Varying: 6 of 16 at one or two points in time
Never: 6 of 16

Three counts with a denominator, no curve and no trend.

Step 4 · AssessmentFictional example: Auwald Dämmstoffe

Four values, three points in time, no overall score.

The assessment sets the values side by side instead of blending them into one number. At the last point in time: mention 7 of 16, of which 4 stable across all three points in time; position 1 to 4 among two to four named suppliers; source of the mention 3 of 7 own page; context outdated in 2 of 7 answers.

From this follows a next step that needs no click count: bring the page auwald.example/steildach up to date and observe the same 16 prompts again in week 13. Whether the answer then says 280 instead of 240 is shown only by the measurement.

No score. One number made from four values would hide exactly the differences that justify an action: four stable prompts say something different from two outdated statements.

Assessment · week 9 · 16 prompts, one answer surfacefictional

Mention: 7 of 16 promptsof which 4 at all three points in time, 1 at two, 2 only in week 9denominator: 16 prompts, week 9

Position: 1 to 4among two to four named suppliers; position 1 in 2 of 7 answersown denominator per answer

Source of the mention: 3 of 7 own page4 third-party: 3 trade media, 1 directoryone source per mention, not per answer

Context: 2 of 7 outdated“up to 240 millimetres” instead of 280; 5 of 7 currentread by hand, with a date

Next step · without a click count

1Update the page: 280 instead of 240 millimetres
2Week 13: observe the same 16 prompts again
3Read the context again, compare the values

Whether the answer then says 280 is shown only by the measurement.

Honest limitFictional example: Auwald Dämmstoffe

What a mention without a click is not.

A mention without a click is visibility, but not a visit, not an enquiry and not revenue. The article does not convert mentions into reach and does not estimate clicks that nobody measured.

What stays comparable is what was counted: the same prompts, visible denominators, four separate values, documented points in time. Which prompts belong in a set is shown in the article on choosing prompts; how six answer surfaces answer the same question differently, in the article on answer surfaces.

Not claimed

visits or enquiries from a mention
a reach extrapolated from mentions
that the click has become unimportant
a trend from three points in time

What is counted

mention with a denominator
position among named suppliers
source of the mention
context of the mention

Count first, then interpret.

The fundamentals article explains mention, position and source in detail. Managed monitoring observes the same prompts continuously.

AI visibility explained View the monitoring →Look up the terms →