September 22, 2026
Author: Jason Faber
Why rankings and traffic stopped telling the story, why page-level attribution can be a fool's errand, and how I'd measure and report SEO and AEO to a CMO in 2026.
Almost every client asks me some version of this. Usually it's the CMO. Sometimes it's a CFO looking at a budget line. Lately it comes with a chart showing organic traffic flat or down while the SEO spend stays the same.
For most of my career, this question was easy to answer. Rankings went up, traffic went up, and leads followed. You put a line chart in the deck and everyone nodded.
That answer doesn't hold anymore. SEO didn't stop working. The measurement model it was built on did, and most dashboards haven't caught up.
For two decades, SEO reporting ran on a simple chain:
→ Rank for a keyword→ Earn the click→ Convert the visitor
It was linear and you could attribute every step. Rankings were a fair stand-in for traffic, and traffic was a fair stand-in for revenue. You could point at a page, name the keyword it ranked for, show the sessions it drove, and trace the demo requests that came out the other end.
Every SEO dashboard built in the last 15 years assumes the same thing: the user eventually visits your website.
That assumption is now the problem.
Zero-click became the default. SparkToro's 2024 study with Datos found that 58.5% of US Google searches end without a click. For every 1,000 searches, only 360 clicks go to the open web. That was before AI Overviews were fully rolled out.
AI Overviews made it worse. Seer Interactive's tracking found that organic CTR on queries showing an AI Overview fell 61% between June 2024 and September 2025, from 1.76% to 0.61%. CTR on queries without an AI Overview fell 41% over the same period. It's not just the AIO. Searchers have changed how they behave everywhere.
Impressions and clicks moved in opposite directions. Adam Heitzman described one client whose impressions were up 32% year over year while clicks fell 17%. More visibility, fewer visits. If your dashboard only shows clicks, that looks like failure. It isn't.
LLMs became the research layer. According to Wynter's research, 84% of B2B buyers use AI for vendor discovery and 68% start their research in AI tools before they ever touch Google. By the time they search, they often already know who they're considering.
The data got noisier. Bots now make up 51% of all web traffic, according to Imperva's 2025 Bad Bot report. In the same Search Engine Land piece, one ecommerce site found 23% of its recent traffic came from AI crawlers that never converted. If you aren't filtering, you're trending the wrong numbers.
Discovery spread out. ChatGPT, Perplexity, Gemini, Reddit, YouTube and review sites all shape buying decisions now. None of them show up in Google Search Console.
The natural response to all of this is to go looking for the traffic somewhere else. "Organic is down, but AI referrals are up, so let's track those instead."
I'd push back on that. So does Gaetano DiNardi, who makes the case bluntly in AI platforms are not going to "recoup" your lost Google search traffic. For most B2B SaaS sites, AI referral traffic is under 5% of the total, roughly what you'd expect from social media. And it's levelling off, not taking off.
AI platforms and social platforms share an incentive: keep the user on the platform. They don't want you to click out. As Gaetano quotes John-Henry: "AI search was not designed for traffic generation. It was designed for displaying answers."
So if you swap "organic sessions" for "AI referral sessions" as your headline KPI, you haven't fixed anything, you've actually made the problem worse.
Gaetano has eloquently coined this the Dark SEO Funnel, which has three stages:
1. Ingestion: LLMs take in and understand your content, your brand and what others say about you
2. Recommendation: AI suggests your brand when a buyer asks about a problem you solve
3. Verification: The buyer goes to Google, searches your brand name, and converts
Your analytics only sees the third step. It credits the conversion to Direct or Branded Search. The work that actually earned it (the content, the citations, the third-party mentions, the reviews) happened upstream, where nothing can attribute it.
The attribution breaks and the funnel becomes very, very dark.
Visibility in AI search also comes in two forms, and they're driven by different things:
And citations aren't just vanity. Seer found that brands cited in an AI Overview earned 35% higher organic CTR than brands that weren't cited. Visibility and clicks still connect. They just don't connect in a straight line anymore.
Before you change what you measure, you have to reset what "good" looks like. Otherwise, every report becomes a debate about a traffic line going the wrong way.
Look at where the traffic loss came from. For most sites, it's overwhelmingly top-of-funnel informational content: definitions, how-tos, "what is X" queries. These are consensus questions AI can now answer perfectly well on its own.
And many of those clicks were never worth much. They were topically loose, rarely converted, and inflated traffic numbers for years. As Gaetano puts it, most websites "never deserved those clicks in the first place."
So here's what you should do:
Rankings and traffic don't disappear in this model. They get demoted from scoreboard to diagnostic. They still explain why things move, show your baseline reach, and flag technical problems or content decay early. They just belong in the SEO team's workspace, not the board deck.
Here's the framework I use with clients. Every metric belongs to a tier, and every tier has a different audience.
A few notes on making each tier work.
This is the only tier the CEO and board should really care about, and it's where conversions and revenue live.
→ Report branded and non-branded organic conversions. Branded search is where the Dark SEO Funnel surfaces: the buyer discovered you in ChatGPT or a community, then searched your name to verify and convert. Count only non-branded and you leave out most of what SEO now contributes. Fold branded in, but note that SEO is not the only channel driving this – brand, paid, social, PR and sales all play a role, too.
→ Count high-value conversions from organic. Demos, trials, sign ups, qualified form fills, purchases. If you can report on revenue, even better.
This is where most of the "new" measurement lives, and where the most important leading indicators sit.
→ Track money prompts, not vanity keywords. Build a set of 30 to 50 high-intent prompts that match how your buyers actually ask AI for help: "best [category] for [use case]," "[competitor] alternatives," "how do I solve [problem]." Track whether you're mentioned, whether you're cited, and who shows up instead.
→ Use branded search as a leading indicator. Rising branded query volume is one of the clearest signs that upstream visibility is working, even while non-branded clicks are flat or falling. The challenge here is that SEO can't solely claim responsibility for this – you will battle with brand, performance, customer support, sales, and social for this.
I am seeing branded search grow across my client portfolio. In the last 6 months, branded search traffic is up 17%, 15%, 27% and 18% across my four core clients' websites.
→ Watch homepage traffic. It has been widely reported that rising homepage traffic is another signal of SEO working in a zero-click world. People who discovered you somewhere else come to your front door.
I am also seeing homepage traffic grow across my core clients - up 8%, 15%, 12% and 11% over the last 6 months.
→ Monitor sentiment and accuracy. It's not enough to be mentioned. You need to be described correctly. Track how AI describes your product, pricing and positioning, and flag hallucinations.
Keep all the old metrics here, and keep them mostly for the SEO team. Rankings explain movement. Impressions show reach. Indexing and Core Web Vitals catch technical problems. Crawler activity tells you whether AI bots can get to your content.
Organic traffic lives here too, with one condition: segment it. Total organic sessions is the least useful number on your dashboard, because it blends branded with non-branded, top of funnel with bottom, and humans with bots. Non-branded organic clicks is a much better metric to track and analyze, but look at it at the page or content grouping level (blog, product pages, docs, locations, etc.)
The following are no longer top of mind:
Self-reported attribution. Add a "How did you hear about us?" field to your high-intent forms, and make it free text. When buyers type "ChatGPT," "Google," or "searched online," that's the dark funnel showing up in your data. This can be extremely effective at helping you measure and track your success.
Correlation over attribution. You won't get a clean line from one citation to one deal. What you can get is directional confidence: AI visibility rising, branded demand rising, and branded or direct conversions rising together over the same period. Those are all signals that point in the right direction for your business.
SEO used to report to the head of Growth, but today the C-suite and board are far more invested. This is a good thing, but it also changes how we communicate our impact.
Set expectations. Before the first report goes out, agree with leadership that SEO will be measured on contribution, not click-level attribution. Walk them through the new baseline and why. If you skip this step, you'll spend every review defending a traffic line.
Answer three questions, in this order:
Structure the dashboard to match:
→ Top: hero scorecards. Search-influenced pipeline and revenue and high-intent conversions, each with a period-over-period or year-over-year change.
→ Middle: market presence trends. Share of voice across your commercial topic clusters against your top three competitors, AI citation and mention share on your money prompts, and branded search demand over time.
→ Bottom: a health check. A simple green/yellow/red status for indexing, Core Web Vitals and overall traffic direction. It's there to show the engine is running, nothing more.
Leave out keyword ranking tables, backlink counts, domain authority scores, crawl errors, schema warnings and page-by-page traffic breakdowns, unless a specific revenue page needs attention. This is all important stuff, but none of it really belongs in front of a CMO.
Match the cadence to the audience. Monthly for marketing leadership, quarterly for the exec team and board. SEO moves over quarters, not weeks. Reporting it weekly to executives invites overreaction to noise. I can't tell you how many times I've heard "Our traffic is down 11% WoW" or "why did we lose 18 top ten keywords last week?".
I'd rather be upfront about the limits than oversell the framework.
AI visibility data is directional. Every AI tracking tool works from a sample of prompts that you provide, and LLM answers vary from run to run. This is – and I cannot emphasize this enough – NOT real data. Prompt tracking is a simulation, not real users. Treat the numbers as trends, not precise measurements, and be wary of anyone claiming exact AI rankings. Make sure you're tracking enough prompts (and the right prompts), you're pulling data across multiple models, and you're pulling it frequently. Tracking 5 prompts, once a month on ChatGPT is useless.
Branded demand has many drivers. PR, paid media, events, product launches and word of mouth all move branded search. SEO contributes to it. It doesn't own it. Claim your share, not the whole thing.
You won't get certainty. Every CMO wants a perfect attribution model, but it doesn't exist. What you get is a set of signals that, taken together, show whether the channel is building the business. That's less satisfying than a clean attribution report. It's also closer to the truth, and a CFO will trust an honest "here's what we can and can't see" far more than an inflated model that falls apart under one follow-up question.
SEO measurement used to be about proving that a page earned a ranking and a click. Now it's about showing that your brand is present, recommended and chosen across every place buyers look, most of which you'll never get a referral from.
The brands handling this well aren't the ones with the most sophisticated attribution. They reset their baseline, stopped arguing about individual keywords, and measure the portfolio against the outcomes the business cares about.
Your SEO may be working better than your dashboard suggests. The job now is to build a dashboard that can see it.
