Google's moving slower on some updates but faster on others. This week, we're looking at what that means for how you work.

TL;DR: Google's September spam update is rolling out over two weeks instead of days—giving you time to watch, not just react. New tools in Search Console show image-based searches. One experiment shows forced AI Mode cuts clicks to websites by 18.8%. Your data is your most reliable compass right now.

🚨 The Spam Update Slowdown

Google's September 2026 spam update started rolling out on September 24—but here's what caught attention: the rollout window is up to two weeks. Compare that to March, June, and August updates this year, which all finished in under three days.

Why does this matter? A longer rollout means your data won't tell the full story for days. If your traffic drops on day one, you can't assume that's the final impact—the update's still rolling out.

The practical move: don't panic early. Cross-reference any ranking or traffic shifts against the full two-week window before deciding what to change. Google hasn't specified what this update targets, so review your spam fundamentals—automated content, user-generated spam, repetitive pages, site hacking. The basics still win.

🔍 Image-Based Search Data Now Shows in Search Console

Google Search Console's Performance report now includes a filter for multimodal searches—searches where someone used an image. This covers Google Lens, Circle to Search, and Chrome's "Search this image" feature.

What you get: page-level data showing which of your pages are found through image queries. What you don't get: the actual image queries themselves, because they don't work the same way as text.

The value is real if you're e-commerce, fashion, or photography-heavy. You can now compare how pages perform in image searches versus text searches, segmented by country and device. Look for pages where visual search brings surprising traffic—or where you're missing opportunities.

Hit reply

Are you seeing traffic come through image-based searches? How significant is it compared to your text search volume?

🛠️ The Canonicalization Disaster Risk

A Reddit user shared a nightmare: their site's pages got de-indexed because Googlebot crawled an error page and treated it as canonical, then indexed that error page under a generic JavaScript error title instead of the real content.

John Mueller acknowledged that Googlebot can indeed pick up and canonicalize the wrong page, and when it does, your actual content stays hidden. The key insight: catching this before Google makes it permanent is what matters.

His core advice: run automated tests on critical pages before you push anything live. Schedule fetch checks to catch errors before search engines turn them into permanent problems. This isn't a one-time audit—it's ongoing monitoring of the pages that matter most.

📉 Publisher Traffic from Google Search Keeps Declining

Chartbeat's 2026 Publisher Playbook found that Google Search traffic to major publishers fell 40.2% year-over-year from July 2025 to July 2026. That's steeper than last year's 21.9% decline.

Google Discover traffic dropped even more steeply—from a 6.6% decline to 34.3% in the same period. Meanwhile, direct traffic, internal cross-promotions, and "dark social" (messages, apps, untrackable referrals) all grew in share.

One caveat: this data comes from Chartbeat's publisher network, which leans toward news and media. Your site may see different trends. But the signal is clear: betting everything on Google is riskier than it used to be. Owning your audience through direct relationships matters.

⚡ AI Mode Cuts Website Clicks by 18.8%

Researchers from the University of Pennsylvania and Northeastern University ran a field experiment where participants were forced into Google's AI Mode for their searches. The result: they clicked through to external websites 18.8 percentage points less often than usual.

Normally, these users clicked through sites 0.6% of the time they used AI Mode. In the forced scenario, they almost never left Google.

Participants also reported lower trust in information and lower satisfaction with their search experience. This is a window into the future: if AI answers live inside the search engine and users rarely need to click out, the SEO game changes. You'll need strategies for getting visibility within AI-generated summaries, not just traditional rankings.

"Your data is your most reliable compass right now."

One thing I'm thinking about: The theme running through all of this is self-reliance. Google's rollouts are slower, so you need your own monitoring. AI answers eat clicks, so you need to diversify. Publisher traffic is declining, so you can't assume Google will sustain your growth.

The narrative—especially in the Chartbeat numbers and the AI Mode experiment—is that relying entirely on Google is increasingly risky. The search landscape is less transparent (longer updates, incomplete reporting, answers consumed in-engine) and more competitive (AI looking to answer instead of directing to sites).

This doesn't mean SEO is dead. It means SEO as a standalone strategy is incomplete. Build your search fundamentals. Monitor your data obsessively. Then diversify—own your audience, build email lists, create reasons for people to visit you directly. Google will still drive significant traffic for most sites, but the days of treating it as your only traffic engine should be long gone.

Till next time,

John Sukowaty