What Is Search Visibility? The Definition, the Formula, and a Real CTR Curve

The Short Version

Search visibility describes how often and how prominently a site appears in search. A ranking-based visibility score estimates a share of potential clicks across a defined keyword set. It is a weighted score, not a ranking position. You take the queries you care about, weight each one by how many clicks that position historically earns, and express the result as a share of what you would get if you owned position one for all of them.

The formula below is a transparent relative visibility model. Vendor scores can use different keyword baskets, weights, and index scales, so their numbers are not automatically interchangeable.

Here is the calculation, followed by a dated click-through snapshot from our own properties. The model estimates potential clicks; observed traffic remains a separate measurement.

The Formula

Search visibility for a single query is straightforward:

Visibility = expected CTR at your position, divided by expected CTR at position one.

Rank first and you score 100 percent. Rank somewhere that historically earns a tenth of the clicks position one earns, and you score 10 percent. Rank nowhere and you score zero.

For a set of queries, you weight each one by its search volume and average the results:

Visibility = SUM(volume x CTR_at_your_position) / SUM(volume x CTR_at_position_1)

A worked example. Three queries, all with 1,000 monthly searches. You rank 1st, 6th, and 40th. Using the curve further down this page, that is 21.90 percent, 0.80 percent, and 0.19 percent expected CTR.

Numerator: (1000 x 0.2190) + (1000 x 0.0080) + (1000 x 0.0019) = 228.9.
Denominator: 3 x (1000 x 0.2190) = 657.
Visibility: 228.9 / 657 = 34.8 percent.

Notice what that number is doing. Two of your three rankings contribute almost nothing. The single first-place ranking is carrying 96 percent of the score by itself. That is not a quirk of the example. That is the shape of the curve, and it is the single most useful thing search visibility tells you.

The Curve Everything Depends On

This model depends on a CTR-by-position table. Different populations can produce different weights, so we measured and published a curve from our own properties.

The September 7 export below contains 368,474 disclosed-query impressions and 1,284 clicks across 120 properties, pulled from Google Search Console between 2025-09-05 and 2026-09-04. Position is Search Console's fractional average, rounded before bucketing.

The dated dataset, including methodology and raw buckets, is published at CTR by Search Position. Google explains how impressions and average position are counted; an average-position bucket is not a controlled sample of individual rankings.

Read the Sample Sizes Before You Use This

Look at the impressions column, not just the CTR column. That is where the honest reading of this table lives.

Position one rests on 1,671 impressions. Position two rests on 1,079. Those are small numbers, and small numbers behave badly. Position two reads 19.00 percent against position one's 21.90 percent, which is not a real gap so much as two thin samples landing near each other. Position six reads 0.80 percent while position seven reads 1.44 percent, which is the curve going backwards. A pooled observational table can be non-monotonic because the samples are small and the queries, devices, and search layouts differ. That does not isolate a position effect.

So the top of this table is a snapshot, not a benchmark. Do not quote position one at 21.90 percent as though it were a measured constant. It is what a young portfolio happened to record on 1,671 first-place-bucket impressions.

The bottom of the table is a different story. Positions 21 to 50 carry 95,041 impressions and positions 51 and beyond carry 233,178. Those buckets have more observations, but they still describe this portfolio and window. Their 0.19 percent and 0.03 percent CTRs are useful context, not universal probabilities for a different site.

The reason the deep end is so well populated and the top is so thin is the same reason the curve exists at all. These are predominantly young, niche informational and B2B properties, most of them built or rebuilt recently. New pages start at depth. Which means we have an enormous amount of data about what happens at depth and very little about what happens at the top, and it would be dishonest to present those two halves as equally reliable.

What the Shape Actually Tells You

Look at the bottom three rows again. Positions 11 to 20 earn 0.62 percent. Positions 21 to 50 earn 0.19 percent. Position 51 and beyond earns 0.03 percent.

Now consider what that means for a site sitting at an average position in the fifties with a few thousand monthly impressions. At 0.03 percent, three thousand impressions is expected to produce about one click. That is a rough expectation under this particular pooled model, with substantial uncertainty for a different site.

This is the most common misreading of Search Console data I run into. Someone sees impressions climbing and clicks stuck near zero and concludes they have a click-through problem, so they go rewrite title tags. Ranking distance can be the larger constraint. An average position of 55 also blends observations, so inspect the actual query-page pairs before ruling out an intent or presentation problem.

Rising impressions with flat clicks is not a failure signal. On a young site it is usually the expected middle stage: Google has started surfacing you, and you are surfacing in the part of the curve where clicks do not exist yet. The number to watch in that stage is average position, not clicks. Movement toward more prominent positions may create better click opportunities. There is no universal position threshold where clicks switch on.

Where the Classic Definition Runs Out

Everything above describes one explicit ranking-based model. The assumption underneath it is the part that has aged badly: that there is one results page, ranked in one order, and a click is the only outcome worth measuring.

Search visibility now spans several surfaces that do not move together.

• Traditional organic rankings. The original surface, and the only one the classic formula measures. Still real, still worth tracking, still not the whole picture.
• Featured snippets and knowledge panels. Elevated placements pulled from a page's content, often independent of that page's normal ranking position.
• AI-generated overviews. Summaries layered on top of traditional results, drawing from a different and considerably less transparent selection process than organic ranking.
• AI-native answer engines. Chat systems and assistants that skip the results page entirely and answer directly, citing sources or not. See What Is AI Search Visibility? for the retrieval mechanics of this surface.
• Zero-click and entity surfaces. Knowledge graph entries, brand panels, and any surface where the answer arrives without a click. That still counts as visibility even though it produces no traffic and no CTR.

A blended vendor score is weighted toward whichever surfaces that tool measures well, which in practice means traditional rankings, because that is the oldest and cheapest data to collect. You can post a healthy score while being nearly absent from AI overviews and answer engines, and the score will not tell you, because it is not measuring that.

This matters practically. If you are trying to work out why traffic feels disconnected from your apparent authority, or why competitors get cited by AI tools more than their rankings would predict, a single blended number hides the exact gap you are looking for.

How to Actually Measure It in 2026

Treat search visibility as five readings, not one.

1. Classic weighted visibility, using the formula above against your own CTR curve rather than a vendor's. Keep the keyword basket, search-volume source, and weighting method documented.
2. Featured snippet and knowledge panel presence, checked directly against the queries where those elements actually appear.
3. AI overview presence, checked by running your target queries and recording whether you are cited, summarized, or absent. Record the date, query, location, and observed response because the presentation can vary.
4. Answer engine citation frequency, checked by querying AI assistants about your topic and brand directly, not inferred from search-engine data.
5. Server-log signal, specifically separately classified browser requests and successful good-bot or AI-crawler requests to the relevant landing pages. This tells you whether machines are reaching the content at all before you worry about whether they are citing it.

The fifth is the one most sites skip and often the fastest diagnostic. Healthy crawler access establishes that those requests reached the content. It does not establish indexing, retrieval, or citation. Page clarity is one useful thing to inspect next, alongside relevance, source competition, and the actual answers observed; AI semantics explains that part of the inspection.

What to Fix First

Work out which surface is weak before choosing a remedy, because the evidence should determine the next task.

If classic visibility is low and you are buried past position 20, inspect intent, content usefulness, and internal linking, then evaluate titles in that context. A better snippet alone is unlikely to resolve a large ranking gap.

If rankings are fine but AI overview and answer-engine presence are weak, the work shifts to entity clarity, structured data, and stating claims in a form a machine can extract without guessing. Those mechanics are covered in AI Search Visibility vs. Search Visibility.

If good-bot requests are low across the board, nothing downstream matters. Fix crawlability and discovery first.

The free AI Search Visibility Checker inspects retrieval clarity on one page. It does not measure all five surfaces or calculate your ranking-based visibility score. Use Search Console for observed clicks and impressions, and direct observations for AI citations.

The Honest Caveat

The classic formula is well defined and you can calculate it today. The cross-surface picture is not, and anyone selling you a single definitive number that spans organic, AI overviews, and answer engines is rounding off real uncertainty, because AI overview and answer-engine presence still are not exposed consistently by any standardized reporting tool as of 2026.

Calculate the classic number, because it is real and it will tell you honestly where you sit on the curve. Track the other four surfaces separately, be direct about which ones you can measure well right now, and treat the rest as directional until the reporting catches up.

And measure your own CTR curve rather than trusting anyone's, this one included. Ours is published precisely so you can see what a real one looks like, thin spots and all, instead of a smoothed vendor line with no sample sizes attached.

From a Visibility Score to a Useful Visit

The Searchmetrics example shows how a fixed keyword basket and a click model become a score. GSC impressions are observations of your own visibility, not the full search demand for those terms. Substituting impressions for search volume creates a different indicator, so label that choice and keep it consistent.

Our visibility-to-inquiry study follows a real 28-day baseline: 8,148 page impressions and three Google clicks. It separates ranking work from the offer a reader finds after clicking. For a review of your own pages, data, and next steps, request an AI Symantix Audit.