Search Visibility Harm

The ranking usually holds; the page around it changes

The first assumption to drop is that this is a ranking problem. On a search for the company name, the company's own site almost always still comes first. What changes is the composition of the page: a complaint site, a review profile, a news piece or a forum thread lands at position two, three or four, in the same screenful as the brand's own result. Nobody lost a ranking. The page now contains an accusation.

Three separate effects follow, and they behave differently enough that lumping them together produces nonsense:

• Click diversion. Some proportion of people searching the brand click the negative result instead of, or as well as, the company's site.
• Click suppression. Some proportion read the page, click nothing and leave. No analytics package on the company's own site sees this at all; it exists only as an impression with no matching click.
• Downstream conversion damage. People who arrive anyway, having read it, and buy at a lower rate. This one is measurable on property you control, and it is frequently the largest of the three.

Two further mechanisms sit outside all of it. Autocomplete and related-question boxes can surface the allegation as a suggested search, spreading it to people who never went looking. And an AI answer can repeat the accusation with no click to count and no impression attributable to the brand. Both are real harm; neither appears in any console.

Four numbers, and what each one is actually worth

Search Console reports impressions, clicks, click-through rate and average position for a property you have verified. Their value is not equal.

Click-through rate on a single held-constant query is the most informative number in this subject. Holding the query fixed removes the biggest confounder — the mix of searches changing underneath you — and asks a narrow question: of the people who searched this company by name, did the same share still click through? A branded CTR that falls while impressions hold steady is an unusual signal and much harder to explain away than a traffic total.

Impressions measure demand for the query rather than performance on it. A rise in branded impressions after an attack is common and is not good news; it usually means more people are searching the name because they read something.

Clicks are the headline number and the most contaminated, because they aggregate every query, device and location at once.

Average position is the weakest and the most over-read: an average across many different result pages, devices, locations and personalizations. A move from 1.2 to 1.6 can be a genuine slip or a shift in query mix. It also cannot see what actually happened here, because your position did not change — position three did.

Separating the searches that carry your name

Branded search is the demand signal a reputation event actually moves. Somebody who types a company name has already decided to consider it; whether they then click is a question about confidence rather than interest. Non-branded search behaves differently and mixing the two hides everything.

Since March 2025 this no longer requires regular expressions and guesswork. Google's Search Console documentation describes a built-in branded and non-branded query filter, and states that it carries “a 16-month history of data, starting from when it was first introduced in March 2025.” That is Google describing that filter specifically. For how long the Performance report retains its underlying data overall, I am not putting a figure on this page: I could not confirm one in Google's own documentation, and an invented retention number is exactly what gets quoted back at somebody in a meeting. Check Google's current help pages, which are revised without announcement.

Applying a filter also changes what the numbers mean, which Google says plainly:

“Some queries are omitted from the report to protect user privacy. These are called anonymized queries. They're included in chart totals, unless a query filter is applied (for example, ‘queries containing' or ‘queries not containing' a given string).”

— Google Search Console Help, “Performance report (Search results): Dimensions and data groupings”, read 12 August 2026

So the moment a brand filter goes on, the totals change basis. Comparing a filtered period against an unfiltered one produces a difference that is an artifact of the filter — an easy mistake to make with two tabs open.

What the console samples, withholds and gets wrong

Search Console is the best instrument available here and it is not a ledger. Google is unusually candid about its limits, and each one matters when a chart is being used to describe harm.

“Due to internal limitations, Search Console stores and shows only the most important data rows. Not all queries beyond anonymized queries are shown in the table.”

— Google Search Console Help, “Performance report (Search results): Dimensions and data groupings,” read 12 August 2026

Alongside that: most reports “only cover a representative sample of URLs to assess quality of data, rather than a comprehensive listing of all known URLs”; chart and table totals can disagree, which Google attributes to aggregation differences between property and page; and data becomes available in two to three days, so the last few days of any comparison are incomplete rather than bad.

Then the failure mode nobody checks. Google maintains a public page logging its own Search Console reporting errors, and the entries are neither rare nor small. In 2026 alone: an announcement on 3 April 2026 of a logging error that affected impression reporting from 13 May 2025 through 27 April 2026; FAQ rich results ceasing to appear from 7 May 2026, dropping impression counts; and logging errors reducing Discover clicks and impressions on 21 May 2026 and 24 June 2026. Before attributing a step change to a reputation event, read that page for the date on your chart. It is free, it is Google's own, and almost nobody looks at it.

The baseline a business either has or does not

Here is the uncomfortable part. Almost everything above assumes there is a before. A business that was not recording branded query performance until the week it found the problem has lost that comparison permanently, and no money spent afterward reconstructs it. Search Console does not backfill to before the property was verified. Rank trackers hold what they were asked to track. Screenshots of a result page exist only if somebody took them.

That is the strongest practical argument for monitoring, and it is worth being precise about what kind of argument it is. Watching your rankings does not protect you, any more than a camera prevents anything. It is an argument about the record: on the day something happens, the twelve months before it already exist in a form somebody else can inspect, with dates nobody assigned retrospectively.

Four things are worth capturing from today:

• Verify the property and export the Performance report monthly, to a dated file that lives outside the tool.
• Record what the result page itself looks like for the company name, logged out, as a dated full-page capture. Search Console reports your clicks, not what else was on the page — and that is the whole subject.
• Track the queries that carry the brand, including misspellings and the name paired with words like reviews, complaints or scam.
• Log every change you make to the site, so a traffic step six months from now can be checked against a redesign before it is blamed on an article.

Why a click-through curve is not a damages figure

The standard vendor estimate takes a published CTR-by-position curve, applies it to an assumed search volume, multiplies by a conversion rate and an order value, and prints a number. Every step of that is unsound, and the studies themselves say so.

SISTRIX published the most-cited breakdown — Why (almost) everything you knew about Google CTR is no longer valid, by Johannes Beus, published 14 July 2020 and modified 7 July 2025 — from Search Console data combined with its own result-page layout data across more than 80 million keywords, mobile only. The headline is 28.5% for position one. The caveat matters more:

“Global CTRs across all types aren't very useful because depending on the search intention, and therefore SERP layout, the CTR for position #1 will vary between 13.7% and 46.9%.”

— Johannes Beus, SISTRIX, published 14 July 2020, modified 7 July 2025, read 12 August 2026

Position one with sitelinks: 46.9%. Pure organic: 34.2%. With a featured snippet above it: 23.3%. With a knowledge panel: 16.7%. Note where that top figure lives — sitelinks appear overwhelmingly on navigational, branded searches, which is exactly the query type this subject is about. SISTRIX is a commercial search-analytics company publishing its own dataset. The other continuously updated public curve, the Google Organic CTR Study from Advanced Web Ranking, is built from Search Console data and updated monthly since September 2015; its publisher sells rank-tracking software and gives the study away. It can at least be segmented to branded queries, which is the only way it should be read.

The page itself has also moved under everyone. Pew Research Center, publishing 22 July 2025, tracked 900 US adults who installed browsing-monitoring software, across 68,879 Google searches in March 2025 of which 12,593 produced AI summaries: users clicked a traditional result on 8% of visits where a summary appeared against 15% where none did, clicked a link inside the summary on 1% of visits, and ended the session entirely on 26% of summary pages against 16% without. That is a general-population finding rather than a brand-query one, and it is enough to show that curves built before 2024 describe a page that no longer exists.

And the fatal objection, which applies even to a perfect curve: no published study measures what happens to a company's clicks when a complaint page ranks third for that company's name. That research does not exist. Modeling it from a general curve is not estimation, it is invention — and multiplying four uncertainties together does not average them out, it compounds them.

Everything a ranking change can be besides the attack

A search change is consistent with a reputation event. It is also consistent with a long list of other things, and an honest report says which were checked:

• A core algorithm update, a layout change on the query, or an AI summary starting to appear where it did not before.
• A migration, redesign, template change, or a canonical, redirect or robots directive that quietly removed pages from eligibility.
• A competitor publishing something that outranks you, unrelated to any attack.
• Pages dropping out for reasons of their own, or a reporting error at Google's end logged on its own anomalies page.
• The possibility that the search data is fine and the problem is elsewhere in the funnel.

The correct phrasing for a conclusion here is consistent with, not caused by, unless something in the record connects the two beyond timing. That is a smaller claim than most reports make and the one that holds up when somebody competent reads it.

None of this reaches the result itself. A console is an instrument, not a lever: it will tell you with real precision that fewer people clicked through on your own name after a particular week, and it has no mechanism whatsoever for doing anything about the page that caused it. Removal and suppression are separate work with separate odds. What this produces is a dated, sourced account of what happened to the business's visibility — useful when the next conversation is with counsel or an insurer, and beside the point if the hope was that the article would go away.

Frequently Asked Questions

Published 2026-08-12