Google Spam Update & GSC Image Data: SEO Pulse

Google’s September 2026 Spam Update Unfolds

Google initiated the rollout of its September 2026 spam update on September 24, establishing a deployment window that extends up to two weeks. This timeline marks a significant departure from previous enforcement actions this year. According to the Search Status Dashboard, the update began at 9:15 a.m. PDT and applies globally across all languages. It serves as the fourth spam-related adjustment of 2026, following iterations in March, June, and August. Previous updates completed their rollouts within two days and sixteen hours, making this extended period notable for its duration. Google has not publicly specified the exact categories of spam content this update targets.

The prolonged rollout suggests that immediate data may not reflect the full impact of the changes. Webmasters should monitor Search Console metrics closely, correlating any fluctuations in rankings or traffic with the specific dates of the rollout rather than assuming instant results. Google advises sites potentially impacted by this update to review its spam policies to ensure compliance.

New Multimodal Filtering in Search Console

Google is introducing a multimodal filter within the Performance report of Search Console, allowing users to isolate data related to image-based searches. This feature encompasses queries originating from Google Lens, the Circle to Search function on Android devices, direct image uploads to Google Search, and the right-click “Search this image” capability in Chrome. The filter is available under the Web search type and is being deployed globally.

Because multimodal interactions rely heavily on visual input rather than text, query-level data is unavailable. Google’s dimensions help page notes that these searches “mostly use images rather than text.” Additionally, this reporting capability is expanding to the generative AI performance report.

For e-commerce platforms and sites featuring product imagery, this filter provides a granular, page-level view of user engagement via image searches. Users can compare these interactions against traditional text-search traffic, segmented by country and device. However, the data does not reveal which specific image prompted the search, focusing instead on the destination pages.

Addressing Canonical Errors and De-Indexing

John Mueller, a Search Advocate at Google, recently addressed a complex de-indexing issue reported on Reddit involving a casino site. A website owner noted that Google incorrectly treated a betting page as the canonical URL for unrelated content regarding companies and suppliers. Another user observed similar behavior where a third-party canonical was indexed under a generic JavaScript error title, leading to suspicions that Googlebot had crawled an error page instead of valid content.

Mueller confirmed that crawling an error page could indeed cause such indexing issues and directed users to utilize Search Console’s live URL test for verification. He emphasized that regardless of the technical root cause, the result remains the same: the affected pages are “not showing in search for normal content.”

The core takeaway from Mueller’s response is proactive monitoring. He recommended running automated tests before deploying new site versions and fetching critical pages on a regular schedule. This approach allows webmasters to identify and rectify broken pages before they become entrenched errors in the search index.

Publisher Traffic Trends and AI Mode Impact

Data from Chartbeat’s 2026 Publisher Playbook highlights a sharp decline in organic visibility for media publishers. Between July 2025 and July 2026, Google Search traffic dropped by 40.2%, accelerating from a 21.9% decline the previous year. Similarly, Discover traffic fell from 6.6% to 34.3%. While the playbook does not clarify whether these figures represent absolute pageviews or share of total traffic, the trend indicates a significant shift. Conversely, direct, internal, and dark social traffic increased their share over the same period, channels that publishers often describe as having greater control over.

Separate research from the University of Pennsylvania and Northeastern University examined the effects of AI Mode in Google Search. In a field experiment involving 1,100 U. S. participants, researchers used a browser extension to force full adoption of AI Mode for seven days. Participants using AI Mode clicked through to external websites 18.8 percentage points less frequently than those using standard search. They also reported lower trust in information accuracy and reduced satisfaction with the search experience.

This experimental data contrasts with statements from Google’s Liz Reid, who noted in August 2025 that organic click volume remained relatively stable year-over-year. The study reflects a forced adoption scenario rather than natural user behavior, offering insight into potential risks of AI-driven search interfaces but not necessarily predicting outcomes for individual sites.

Analyzing Your Own Data

Each recent development underscores the importance of relying on proprietary data rather than industry averages. The ambiguous target of the spam update requires checking specific ranking changes against rollout timelines. The new multimodal filter reveals page-level trends without query details, while Mueller’s guidance prioritizes independent testing over reactive fixes. Chartbeat’s network statistics and the AI Mode experiment’s sample group provide context but do not dictate individual site performance. As Search Console offers more segmented views, maintaining detailed records becomes essential for accurate diagnosis and strategy refinement.