Architecting Authority

Groew / Insights / AI Clicks and Impressions in GSC
SEO Data 13 min read May 2026

What AI Clicks and Impressions in Google Search Console Mean for B2B

Google Search Console is the primary data source most B2B companies use to track organic search performance. Between May 2025 and April 2026, two changes made that data misleading in ways most companies have not accounted for. This article explains exactly what happened and what your data actually means.

Updated May 2026 Google disclosed the impressions bug in a 47-word Search Console changelog entry on April 3, 2026. The note confirmed that impression counts were over-reported from approximately May 2025 onwards. Clicks were confirmed unaffected. No announcement was made to the broader SEO community — the disclosure appeared only in the GSC changelog. Historical data from the affected period has not been corrected and cannot be recovered.
Direct answer

Google Search Console over-reported impressions from May 2025 to April 2026 due to a bug related to Google AI Mode. Clicks were not affected. From June 17, 2025, AI Mode results were also blended into Web search totals with no filter to separate them. This means impression data and CTR from mid-2025 onwards are unreliable for year-over-year comparison. Clicks remain the most accurate metric in Search Console for this period.

What Happened in Google Search Console Between 2025 and 2026

Two separate changes affected Search Console data in the same 12-month window, and together they make year-over-year performance comparisons unreliable for most B2B websites.

The first change was the impressions bug. When Google AI Mode was introduced in mid-2025, a counting error caused some impressions from AI Mode panels to be recorded in Search Console even when the user had not scrolled to, or interacted with, the standard search results where your website appeared. The result was inflated impression counts — impressions being recorded for pages that were technically "visible" in the browser but not actually seen by the user in any meaningful sense.

Plain English — what an impression is

In Google Search Console, an impression is counted every time one of your web pages appears in a Google search result. You do not need to click it — just appearing on the page counts as one impression. Normally this is a useful signal: high impressions on a keyword means Google is showing your page for that search. The bug caused impressions to be recorded even for positions that were far down the page and unlikely to be seen, inflating the counts artificially.

The second change was the blending of AI Mode results into standard Web search totals. From June 17, 2025, Google folded AI Mode performance data into the same reporting bucket as regular organic search. Source. This means there is no way, within Search Console, to separate "traffic from standard Google results" from "traffic from Google AI Mode results." They are counted together.

The practical effect: a B2B company comparing its January 2026 Search Console data to January 2025 data is comparing two fundamentally different measurement systems. The numbers look like they measure the same thing. They do not.

Google Search Console data reliability timeline
Pre-May 2025 Clean data Clicks + impressions both reliable May 2025 Impressions bug Impressions over-reported Clicks: still accurate June 17 2025 AI Mode blended AI Mode + Web merged, no filter Apr 3 2026 Bug disclosed 47-word changelog History not corrected

The impressions bug ran from approximately May 2025 to April 2026. AI Mode blending into Web totals began June 17, 2025. Both changes affect year-over-year data comparisons for B2B websites.

What Google AI Mode Is and Why It Changes How Results Are Counted

Plain English — what Google AI Mode is

Google AI Mode is a feature that shows an AI-generated answer at the top of a Google search results page, before the usual list of links. It works similarly to asking ChatGPT a question — Google reads multiple web pages and writes a summary answer for you. Instead of ten blue links, you see one AI-written response. The web pages Google used to write the answer appear as small citation links underneath. If your site is one of those citations, your URL appears — but most users do not click through. They read the AI answer and stop.

Google AI Mode was progressively rolled out from mid-2025 in the US before expanding more broadly. It is different from AI Overviews — the shorter AI summaries that appeared above some standard results from 2024 onwards. AI Mode is a full-page AI answer experience. When it appears, it substantially changes the click behaviour of the person searching.

Research tracking zero-click rates across different Google surfaces found that queries resolved by AI Mode have a 93% zero-click rate. Source. That means 93 out of every 100 people who see an AI Mode answer do not click through to any website. Standard Google results have approximately a 60% zero-click rate (people who search but do not click). AI Mode is dramatically worse for driving website traffic.

For B2B companies, AI Mode creates a specific measurement problem: your Search Console data may show increasing impressions from queries where AI Mode is now appearing, but those impressions represent people who will not visit your site. Impressions are recorded. Clicks do not follow. CTR falls. On the surface it looks like your pages are becoming less effective at converting searchers. In reality, the search experience itself has changed — fewer people are clicking because they already got their answer from the AI panel.

The Blended Data Problem: Why Your Performance Reports Are Misleading

Before June 2025, if you wanted to understand how your B2B website performed in Google search, you could look at Search Console's Web results filter and see a relatively clean picture. Clicks, impressions, and CTR all reflected the same type of result — standard blue link results in the main SERP.

From June 17, 2025, Google started including AI Mode result data in those same Web totals. Source. There is no toggle to turn this off. There is no filter to separate "standard results" from "AI Mode results." The number you see is a blend of both.

Why blended data is a problem for B2B reporting

Imagine your accountant gave you a financial report that combined two different business units into a single total — but did not tell you which was which, and there was no way to separate them. That is what Search Console is now doing for traffic from standard results versus AI Mode results. If your "organic traffic" figure rises, you cannot tell if that is because your rankings improved or because AI Mode is now counting more interactions. If it falls, you cannot tell if your content is declining or if AI Mode is absorbing traffic that previously became clicks.

What is included in your Google Search Console Web totals as of June 2025
Before June 2025 Standard Organic Clean data GSC Web total After June 2025 Standard Organic AI Overviews AI Mode GSC Web total No filter available

Before June 2025, Search Console Web totals showed only standard organic results. After June 2025, AI Overviews and AI Mode data are blended in with no way to separate them.

The result of all this is that year-over-year comparisons in Search Console for any period containing mid-2025 onwards are comparing fundamentally different things. A B2B company whose impressions doubled between January 2025 and January 2026 may have simply had more AI Mode appearances — not better rankings. A company whose CTR halved may be suffering from the structural click decline of AI Mode, not from content quality problems.

Which Metrics to Trust and Which to Treat With Caution

Reliable

Clicks

The impressions bug did not affect click counts. Every click recorded in Search Console represents a person who actually navigated from Google to your website. Use clicks as your primary KPI for the May 2025 to April 2026 period. Clicks per query tell you which keywords are driving real visitors. Click trends over time tell you whether your organic presence is growing or shrinking in real terms, unaffected by the measurement changes.

Use with caution

Impressions

Impressions from May 2025 to April 2026 are likely over-reported due to the bug. Google confirmed the bug in April 2026 but did not correct historical data. Impressions are also affected by AI Mode blending from June 2025. Treat any impression spike during this period with scepticism before attributing it to genuine ranking improvement. Post-April 2026, impressions may become more reliable as the bug is resolved — but AI Mode blending continues.

Use with caution

CTR (Click-Through Rate)

CTR is calculated by dividing clicks by impressions. Because impressions are unreliable, CTR is also unreliable for this period. A falling CTR may mean your content is less compelling — or it may mean impressions inflated while clicks held steady, arithmetically reducing the ratio. Do not use CTR as a performance indicator for the affected period. Compare raw click counts instead.

Reliable

Average Position

Position data in Search Console — showing where your pages rank for specific queries — was not directly affected by the impressions bug. However, AI Mode blending means some "position" data may reflect placement in AI Mode panels rather than standard result rankings. Use position data directionally to understand which keywords you rank for and whether rankings are moving up or down, but treat absolute position numbers for competitive analysis with some caution.

The simplest test to apply to your own data

In your Search Console account, filter by date to the period May 2025 to April 2026. Look at your impressions and clicks side by side. If impressions grew substantially but clicks did not grow at the same rate, the divergence is likely the bug and AI Mode blending — not a genuine performance change. Impressions growing 40% while clicks grew 5% is a classic pattern from this period. Trust the clicks line. Ignore the impressions inflation.

What to Do Now: Rebuilding Reliable B2B Search Measurement

The measurement problem created by these changes does not resolve itself automatically. Without a deliberate response, B2B teams will continue making strategic decisions based on data that conflates four different things: standard organic rankings, AI Overviews appearances, AI Mode appearances, and over-counted impressions from the bug period.

Step 1

Rebaseline your performance benchmarks using clicks only

For any B2B keyword or page that you track over time, set a new baseline using click data from April 2026 onwards — the first month after Google disclosed the impressions bug. All historical comparisons involving May 2025 to March 2026 data should be treated as potentially compromised for impressions and CTR. When reporting to stakeholders on organic search performance for this period, use clicks as the headline metric and note that impressions data is unreliable.

Step 2

Cross-reference Search Console clicks against your analytics sessions

A reliable way to spot-check Search Console data accuracy is to compare GSC click counts against the organic sessions count in your main analytics platform (Google Analytics 4, Plausible, or equivalent). These two numbers will never match exactly — tracking gaps and bot traffic cause small differences — but they should be in the same order of magnitude. A large divergence (GSC showing 2x the sessions your analytics records) suggests a data quality issue worth investigating before using either number in reports.

Step 3

Track AI Mode appearance separately using manual spot-checks

Since Search Console does not offer a filter to separate AI Mode from standard results, the only way to understand your AI Mode presence is manual spot-checks. Set up a spreadsheet. Once a week, run your top 10 target queries in Google. Record whether AI Mode appears, whether your site is cited in it, and whether your standard link also appears below. Track this over 12 weeks. This is not a scalable system — but it is the only way to build a directional picture of your AI Mode presence while Google develops better tooling.

Step 4

Focus B2B content strategy on clicks, not rankings

Given that AI Mode has a 93% zero-click rate, ranking in an AI Mode answer is commercially near-worthless compared to ranking in standard results where people actually click. B2B content should be optimised for the queries where searchers are still clicking through — typically transactional, high-intent, vendor-evaluation queries. Use the SEO Audit Tool to identify which of your pages are receiving clicks versus pure impressions. Pages with high impressions but near-zero clicks are likely appearing in AI Mode or AI Overviews — and may benefit more from being cited there than from ranking well in the standard list, given the changed click behaviour.

The structural shift happening in B2B search is that the traditional SEO metric pyramid — ranking position at the top, then clicks, then conversions — no longer accurately represents how search value flows. For many queries, being cited in AI Mode generates brand recognition without traffic. Being cited in standard results still generates traffic. B2B companies that optimise for citations and clicks simultaneously — rather than just rankings — are building a more durable search presence for the AI era. That is the foundation the B2B organic search infrastructure is built on.

Alokk's perspective
Alokk, Founder at Groew
Alokk Founder and Lead Growth Architect, Groew
When I reviewed client Search Console accounts after the April 2026 disclosure, almost every account showed the same pattern: a large impressions spike from mid-2025 that did not correspond to any genuine ranking improvement, and a CTR decline that looked alarming but was entirely explained by the denominator problem. One client had been considering a complete content overhaul because their CTR had dropped from 4.2% to 1.8%. When we stripped out the impressions inflation and looked at raw clicks, their actual visitor numbers had grown 14% year-over-year. They were performing better than ever. The panic was based on bad data. This is exactly why the first thing we do when auditing a B2B search presence is separate the measurement problem from the genuine performance problem — they require completely different responses.

Common Questions

Google Search Console had a bug from approximately May 2025 to April 2026 that caused impressions to be over-reported. It happened because AI Mode results were counted as impressions even when users had not engaged with the standard search results where your page appeared. Clicks were not affected. If your impressions jumped significantly in mid-2025 without a matching click increase, the bug is the most likely explanation. Historical data from this period has not been corrected by Google.
No. As of May 2026, there is no filter in Search Console to separate AI Mode results from standard Web results. From June 17, 2025, they were blended into the same Web totals. This means you cannot tell from Search Console alone whether your traffic comes from standard results or AI Mode panels. The only way to check your AI Mode presence is manual testing — searching your target keywords in Google and recording whether you appear in AI Mode answers.
No. The bug only affected impressions. Clicks were confirmed unaffected by Google's April 2026 disclosure. CTR was also affected — because CTR is calculated by dividing clicks by impressions. Inflated impressions produce an artificially low CTR even when your actual visitor numbers are fine. Use clicks as your primary metric when reviewing data from May 2025 to April 2026.
Google AI Mode is a feature that generates an AI-written answer to a search query, displayed above the standard list of links. Research tracking B2B websites found that queries answered by AI Mode have a 93% zero-click rate. This means 93 out of 100 people who see an AI Mode answer do not click through to any website. A B2B site can appear in AI Mode citations and still see declining click traffic, because the AI answer satisfies the query without requiring a visit.
Use clicks as the primary metric for the May 2025 to April 2026 period. Clicks are unaffected by the bug and represent actual site visits. Treat impressions and CTR from this period with caution — both are affected by the bug and by AI Mode blending. For historical comparisons, use click data only, and note the data quality issue in any stakeholder reports. For ongoing tracking from April 2026 onwards, monitor whether impressions and CTR stabilise now that the bug is resolved.
From Groew's Search Authority Team

The Complete Guide to Reading B2B Search Data in the AI Era

Google Search Console was designed for a world where organic results were a clean, single-source measurement. AI Overviews, AI Mode, and the impressions bug have changed that. This guide explains how to extract meaningful B2B performance signals from data that has become structurally more complex to read.

Why Traditional SEO Metrics Are No Longer Sufficient

The organic search measurement model built over the past 15 years — track rankings, watch impressions grow, monitor CTR — was designed for a search environment where every result was a link and every appearance was an opportunity for a click. That model does not map accurately to a search environment where 60% of standard queries end without a click, 83% of AI Overview queries end without a click, and 93% of AI Mode queries end without a click. The old metrics still exist in Search Console. They just no longer mean what they used to mean for B2B businesses trying to track genuine commercial performance.

Read the complete guide

The Three-Layer Search Measurement Framework

B2B search performance now needs to be measured across three distinct layers. Layer one is standard organic performance: clicks from standard blue link results, measured in Search Console and cross-referenced in GA4. This is still the most commercially valuable layer — clicks from standard results convert to site visits. Layer two is AI appearance performance: whether your brand, content, or domain appears in AI Overviews and AI Mode answers. This builds brand recognition and citation authority but generates fewer clicks. Layer three is AI platform citations: whether your content is cited by ChatGPT, Perplexity, Google AI Mode, and other AI tools when buyers research your market. Each layer requires different measurement tools and different optimisation strategies.

How to Build a Clean B2B Search Reporting Template

A practical B2B search report for 2026 onwards should include five data points. First: click trend from Search Console, month-over-month and year-over-year, filtered to the domain and most important landing pages. Second: organic session count from GA4, cross-referenced against GSC clicks to check tracking consistency. Third: conversion actions from organic traffic — form fills, demo requests, content downloads — tracked in GA4 via goal completions or GA4 events. Fourth: AI Mode appearance rate for top 10 keywords, measured via weekly manual spot-checks and recorded in a shared document. Fifth: referral traffic from perplexity.ai, claude.ai, and chatgpt.com in GA4 referral reports — this is the early signal of AI citation performance before native tooling improves.

The GSC Historical Data Problem and What Cannot Be Recovered

Google confirmed in its April 2026 disclosure that historical impression data from the bug period will not be corrected. The numbers in your account from May 2025 to approximately April 2026 remain as-reported — including the over-inflated impression counts. This creates a permanent gap in historical benchmarking. For B2B businesses that need to present year-over-year growth data to boards or investors, the practical approach is to document the data quality issue in any historical report covering this period, present click-based trends as the headline, and use April 2026 as the start of a new clean benchmark period. While imperfect, this approach is more defensible than presenting impression or CTR data that is known to be unreliable.

Connecting Accurate Measurement to Revenue Infrastructure

The reason accurate search measurement matters for B2B companies is not reporting hygiene — it is resource allocation. A B2B marketing director who believes their CTR is declining because their content is getting worse will make different investment decisions than one who understands the CTR decline is structural, caused by AI Mode absorbing query resolution. The first person might commission a content overhaul that solves the wrong problem. The second person invests in content architecture that performs across both standard results and AI platforms simultaneously. Groew's B2B SEO system starts with a measurement audit precisely for this reason: fixing the signal before optimising the system.

Get a Free Audit of Your B2B Organic Search Performance

The SEO Audit Tool analyses your site against current Google quality signals, identifies which pages are at risk, and flags measurement gaps so you are making decisions on accurate data — not inflated impressions from a bug period.

Run My Free SEO Audit
ESC