Google published a page listing the referrer values Search traffic sends. Discover and Images send the same https://t.co/0JA5V1ByPO referrer, and GSC counts a click on select while analytics waits for the tag to fire.
Why GSC and GA4 never match: https://t.co/qKOEoOWrkx
AI made shipping pages cheap for everyone. So the edge moves from how much you ship to knowing which changes actually worked.
Cheap habit: annotate the change the day it ships, note what you expect, set a review date.
Better: compare against similar pages you didn't touch.
Branded vs. non-branded is now in the Search Console API, not just the UI.
Also new: webMultimodal data (Lens, Circle to Search) and multi-country/device filters.
Catch: eligible properties only. Not eligible = no rows.
How we split brand traffic: https://t.co/qR99ZwKTg7
Visits to the top million websites are 12.2% below their Q1 2022 peak (Similarweb via SparkToro, per SEJ).
Before blaming your content for a long slide, check your trend against the market.
Then ask: sudden sitewide drop or slow decay on older URLs?
https://t.co/JbBVQhRKla
Seeing an impressions dip in the GSC Generative AI performance report for Nov 9–11?
Google confirmed it's a logging bug, not a real drop.
Catch: the November core update started Nov 10, inside that same window.
Annotate both before you compare weeks.
OpenAI's new Business Hub in ChatGPT shows businesses their mentions, profile views and clicks to their website.
That's a first-party number for AI referrals, something most of us have only been able to estimate.
Next test: does it line up with what your analytics records?
Google's November 2026 core update started Nov 10 and may take up to 2 weeks.
Retail heads-up: a full rollout ends right before Black Friday week, so a naive before/after compares holiday traffic to a normal week.
Annotate the date now. Compare later.
https://t.co/zzyFwpXVZy
Update on those early-November Search Console click spikes:
Marie Haynes reviewed site data in Search Engine Journal and concluded it isn't human traffic. Check the mobile trendline, and look at pages ranking below position 10.
https://t.co/qNO9KpE0Qt
OpenAI is building a Search Console-style tool so businesses can claim their listing and track mentions, impressions, and clicks in ChatGPT.
Kong Metrics will cover this data when they release the API.
https://t.co/aPTIxWzQQR
@glenngabe Great natural experiment. One tip before anyone draws conclusions: split by query type. Informational and transactional queries in the same property probably tell very different stories.
Doing November SEO reports? 👀
Several SEOs are flagging sudden 2x to 6.5x click spikes in Search Console around Oct 31 to Nov 2, while impressions and rankings stayed flat.
No confirmed cause yet. Annotate it before a client reads it as a win 📝
@CrystalontheWeb The device split is the part most AI visibility reports skip 👀
Same check works on your own data: if GA4 shows ChatGPT referrals, break them down by device before deciding how much budget AI search really deserves.
@aleyda The "same 6 organic results" part is what fools most reports. Rank says nothing changed, pixels say result #1 now sits below the fold. Worth tracking above-the-fold share next to rank for exactly this.
Want the cleanest read yet on what AI Overviews do to clicks?
Glenn Gabe's tip: in GSC, pick the last 6 months, filter to France, and look at July 22, 2026, when AIOs launched there.
Some sites lost clicks and CTR. Others barely moved. Same feature, very different outcomes.
If you're looking to find the true impact of AIOs on clicks and click-through rate, then head to GSC, select the past 6 months as the timeframe, filter by France, and then check the Performance reporting. Then check Gen-AI impressions surging on that date, then check clicks, click-through rate, impressions, and position. You might find some very interesting data.
Since Google rolled out AIOs on that date, you can get a clear view of the impact... During my testing, not all sites dropped... Some lost a significant number of clicks, CTR dropped heavily, etc., but some remained stable throughout the rollout. Check my post for more details.
@ekuzevska One more for the checklist: log the date and model version of every run. Answers drift week to week, so a visibility score without a timestamp is hard to compare with anything.
@ConnorGillivan Solid list. One addition to the ICP prompt: paste real customer phrasing from calls or reviews first. The output gets much sharper when the model works from buyers' own words instead of your description of them.
Cyrus Shepard opted Zyppy out of Google's AI features for 28 days. AI was ~10% of impressions but an estimated ~21% of clicks (AI CTR ~0.9% vs ~0.4% sitewide). Regular organic CTR dipped too while out.
One site, one test. Worth replicating before you opt out.
Google updated its AI content guidance on Oct 1. It didn't add a penalty for AI content.
The part people skip: the manual fact-check and review it asks for also covers metadata. Titles, meta descriptions, structured data, image alt text.
https://t.co/QM7051JBT2