Next →
Edition #3 coming soon
"What's your heading?"
I looked at my gauges. Thirty degrees off course. Two hundred feet above my altitude. My instructor's way of saying I wasn't looking at my instruments.
When you're flying in the clouds, you have a lot of things to manage at once. Charts. Air traffic control. Checklists. The full modern workload of flight. But there's a core group of gauges you have to come back to every few seconds. The same information a pilot in World War II would have used. The fundamentals. And if you neglect those core instruments for even 30 seconds, you drift. Thirty degrees off course is the optimistic outcome.
That's where we are in digital marketing right now. The clouds have settled in. New information is coming from every direction, and it's hard to know what deserves your attention. AI Overviews changed how Google distributes clicks. The March Core Update is still reshaping rankings. Meta rewrote how it counts conversions. GEO — generative engine optimization, meaning how you get cited in AI-generated answers rather than just ranked in traditional search results — is now a real discipline with a growing body of data behind it.
But the core gauges are the same metrics marketers have tracked for decades. Revenue. Pipeline. Cost per acquisition. Conversion rate. If you want to stay on course in uncertainty, you come back to those on a fixed scan. Let the new information process in the background. Check the fundamentals first.
Here's what I'm watching this week — and how I'd prioritize it.
This Week's Finds
-
Story 1
The Overlap Between Google Rankings and AI Citations Dropped From 70% to Under 20%
LLMrefs ↗
GEO — the practice of getting your content cited inside AI-generated search answers — has become its own discipline. Research from multiple GEO tracking firms shows the overlap between URLs that rank in Google's top 10 and URLs that get cited in AI Overviews has dropped from 70% to below 20%. Being number 1 on Google no longer means you're showing up in the AI answer above number 1.
Only 16% of brands are systematically tracking AI search performance right now. That gap is an opportunity. The brands building a GEO baseline today will have months of data on brands that start in Q3 or Q4.
-
Story 2
Three Weeks In: What the March 2026 Core Update Data Is Actually Showing
SEO Analysis ↗ALM Corp ↗
The March 2026 Core Update (rolling out since February 24th) is showing a clear pattern. Original, experience-driven content bounced. AI-generated thin content slipped. Finance affiliates and coupon/deals sites got hit hardest. SEMrush Sensor hit 9.5 — one of the highest readings it has ever logged.
The recovery pattern is consistent with past updates: consolidate thin pages, add verifiable expert perspective, and stop publishing content that could have been generated in seconds. The teams recovering fastest removed their weakest 20% of content rather than trying to improve it.
-
Story 3
Meta Changed How It Counts Clicks This Week. Your March Numbers Will Look Different.
SocialBee ↗Swipe Insight ↗
Meta updated click attribution for paid social. Click-through attribution now counts only link clicks — non-link interactions (reactions, comments, shares) move to a separate "engage-through" bucket. The video engaged-view threshold also dropped from 10 seconds to 5 seconds, which is actually good news for video advertisers running shorter content.
If your March performance report looks different from February, check attribution settings before drawing any conclusions. Also this week: Meta is rolling out ads on Threads globally, and a new image-to-video tool lets you upload up to 20 product photos and get multi-scene video ads back. The organic-to-ad ratio on Threads is still heavily weighted toward organic. That window does not stay open.
-
Story 4
Cognitive Bias in Digital Decision-Making Is Getting Worse as Information Overload Increases
Advances in Consumer Research ↗
Research published in Advances in Consumer Research finds that as digital information loads increase, confirmation bias and anchoring bias in decision-making get worse, not better. The first data point you see anchors everything that follows. If you open your campaign dashboard and CTR is down, your brain starts filtering the rest of the data through that frame.
The practical fix is low-effort: write your success criteria and your hypotheses before you look at results. Even a two-sentence note. Teams that define what "winning" looks like before reading the data make better decisions than teams that reverse-engineer a story from the numbers. This directly applies to how most of us are reading Core Update data right now.
70% → 20%
The overlap between URLs ranking in Google's top 10 and URLs cited in AI Overviews — down from 70% a year ago to under 20% today. The two systems are diverging fast.
What GEO Actually Is
GEO (generative engine optimization) is the practice of getting your content cited inside AI-generated search results. Not just ranking for keywords — getting referenced as a source when an AI system answers a question about your topic.
The platforms where this matters right now: Google AI Overviews (now appearing in over 25% of all searches, up from 13% in March 2025), ChatGPT with web browsing, and Perplexity. Each has slightly different citation patterns, but the underlying principle is consistent: AI systems prefer primary sources, original data, and clearly structured content from established entities.
What Drives AI Citations
Original data and proprietary research. AI systems can't summarize data that only you have. Original surveys, community polls, testing results, first-party analysis. This content is structurally resistant to being displaced because there's nothing else to cite. Reddit content appears in 5.5% of all AI Overviews right now, in part because forum discussions contain real experience that isn't available elsewhere.
Named experts and verifiable quotes. AI systems are pattern-matching on credibility signals. A named expert, a direct quote with attribution, a specific claim tied to a specific person carries more citation weight than anonymous commentary.
Clear structure with specific definitions, numbered steps, and stated data points. This is how AI models decide what to pull into an answer. Content that reads as a clear, retrievable chunk beats dense flowing prose for citation purposes.
Publisher authority and entity recognition. Brand entities that Google and AI systems have strong signals for get cited more. Building topical authority is not just a Google SEO play anymore — it feeds GEO performance directly.
What Doesn't Work
Summary content that aggregates what's already widely known. Thin pages without original perspective. Content that scores well for keyword density but doesn't have a data point AI could pull out and cite. These pages may rank fine in traditional search for now. They're nearly invisible to AI systems.
The 3-Step Baseline to Run This Week
You don't need a GEO agency or a new tech stack. You need a baseline and a hypothesis. This takes 30 minutes.
-
1
Run your top 10 keywords in Google, ChatGPT, and Perplexity. Note whether your domain shows up in any AI-generated answers. Three search engines, ten keywords, thirty minutes. That's your baseline.
-
2
Check which sources do appear. Those are your GEO competitors right now — regardless of where they rank in traditional search. Study what those pages have in common: data points, named experts, structural clarity.
-
3
On your most important pages, audit what you have that's irreplaceable. What original data, named experts, or experience-driven perspective does that page have that an AI would have to cite you to reference? That's your GEO moat. Build more of it.
Key Takeaway
The teams I'd bet on for the next 18 months are the ones asking "am I getting cited?" right now, not just "am I ranking?" Both matter. The weighting is shifting — and the brands building a baseline now will have months of data on everyone who waits.
IMG's Take on GEO
Here's where we actually land on GEO right now: we think the category is real, the overlap data is real, and the brands that are going to struggle most in the next two years are the ones treating GEO as a future problem. It's a present problem for anyone whose content strategy depends on organic click volume.
That said, we're skeptical of the vendor circus that's forming around it. Most of what gets sold as "GEO strategy" is just good content marketing with a new label. Write better, more specific, more original content. Use real data. Make your pages easy to parse. This is not a new idea. It's the same idea, now with higher stakes.
The thing that is new is the tracking. If you're not building a baseline for how often you appear in AI-generated answers, you're making decisions with a gap in your data.
The IMG community has been tracking the March Core Update in real time since it started rolling out. Members are sharing traffic data, what moved, what held, and what the GEO picture looks like for their specific niches. When something shifts this fast, ground-level data from people actually running campaigns is worth more than any industry roundup.
Join the IMG Community →
