The SEO-GEO Paradox

AI Overviews cite your content and keep the user on their platform. Traffic is falling while visibility rises. This is the SEO/GEO paradox.

in Strategy

2026 in context

Here is the central tension of search in 2026: Google processes more queries than ever before. AI systems cite more content than ever before. For well-structured sites, brand mentions across AI surfaces are at an all-time high. And yet, organic traffic is declining. For many publishers it’s down 20–40%. For some, over 70%. Your content is being read and synthesised and shared. You are just not getting credit for it in the metric that still shows up in most dashboards: the visit. The evidence is hard to dispute: SparkToro/Datos research placed zero-click searches at ~60% of all Google queries, while AI Overviews now appear in nearly half of all searches. Ranking first means little.

The industry responded with a vocabulary explosion: GEO (Generative Engine Optimization), AEO (Answer Engine Optimization), LLMO, AI SEO, and a lot of debate about which term would stick. An analysis of 75 thought leaders found GEO winning: 59% use it, with 82% positive framing. The naming wars distracted from the harder question: How do you actually measure any of this?

Here are some other realities for us to reconcile with:

Engagement numbers are down

The paradox sharpens further when you look at engagement data. Users spend an average of 49 seconds in AI Mode, double the time spent with AI Overviews, yet 93% of those sessions still end without a single click. The users get their answers, they are well-informed. They just never leave Google. And the content that informed them came from someone’s website.

Keyword rankings are no longer the north star

Since the formative years of SEO, a page-one ranking felt like winning. But when over 60% of searches now end without a click, a #1 position can mean very little in practice. The scoreboard has changed, and the metrics need to follow. A more honest set of KPIs for 2026 looks less like a rank tracker and more like a presence audit. Some metrics that could be worth thinking about are:

  • AI Citation Rate/Frequency: Are you being referenced inside ChatGPT, Perplexity, Gemini, and Google AI Mode responses for your target queries over time.
    Branded Search Volume: Are more people searching for you by name? Rising branded queries signal that AI recommendations are driving awareness downstream.
  • Engagement Quality: When people DO click through, are they staying? Time on page and conversion rate matter more than raw traffic volume now.

Brand mentions are the new backlinks in AI search

Research shows that one of the strongest predictors of ChatGPT visibility is the volume of web sources that include your brand in a relevant context – listicles, expert citations, positive reviews, etc. Meanwhile 61% of the signals shaping AI brand reputation come from editorial media sources, not owned content. Digital PR and a solid brand social media presence is no longer a nice-to-have; it’s core infrastructure for AI search visibility.

Factual density in content boosts AI citation by ~40%

An analysis of multiple GEO studies, found that adding citations and statistics to content improved AI Overviews citation rates by roughly 40%. A newer e-commerce study found persuasive, longer descriptions outperformed citation-heavy copy in AI product recommendations, suggesting the signals vary by intent and platform. However, the problem is that AI recommendations are highly inconsistent, and personalized for individual user. Running the same prompt rarely returns the same list or order of results, according to a recent study.  No universal GEO playbook exists yet (and possibly won’t).

AI bots are now the first readers of your content

A recent experiment tracked GPTBot’s behavior on a brand-new, zero-backlink website: it appeared within minutes of launch and made over 29,000 requests in the first 12 hours, before a single human had visited. The bot didn’t wait for a sitemap submission, a Google index ping, or a social share. It just showed up.

The implication few are discussing: we are crossing a threshold where the majority of “reads” of any new web content will be by AI crawlers, not humans. Content is increasingly produced to feed machines that summarize it for people who never visit your site. The entire business model of content marketing, driving qualified humans to your owned properties, is quietly inverting. Nobody has a clear answer for what replaces it, but the brands that build strong off-site mention volume and structured, machine-readable content architecture now will own the next era of discovery.

Bing is now the silent kingmaker for AI search, and few are optimizing for it.

The entire industry is focused on Google. But recent research makes the case clearly: Bing indexing is critical for ChatGPT visibility. ChatGPT, by far the dominant LLM with nearly 5 billion monthly visits, does not scrape Google. It routes web lookups through Bing’s index. If Bing hasn’t crawled and indexed your content, you are invisible in ChatGPT’s web-augmented answers, full stop.

Yet virtually every GEO guide, every “how to rank in AI search” tutorial, focuses exclusively on Google signals. Bing Webmaster Tools, IndexNow API submission, and Bing-specific crawl optimization are mentioned almost nowhere in the current conversation.

A glimpse into how AI search systems actually cite your website

Microsoft recently launched AI Performance inside Bing Webmaster Tools. This is the first data tool from any major search engine showing how content is cited inside AI-generated answers. It tracks citations across Copilot, Bing AI summaries, and partner integrations, and surfaces something called grounding queries: the reformulated search phrases AI systems generate internally when retrieving your content. Bing WMT is free to set up, imports directly from Google Search Console, and is the only place you can currently pull first-party AI citation data at the URL level.

What this means for your search strategy

  • Audit your Bing presence, not just Google. Check Bing Webmaster Tools for crawl errors, submit your sitemap, and consider using the IndexNow API to speed up Bing indexing for new content.
  • Shift budget toward digital PR, earned and social media. Brand mentions on third-party sites are the primary signal shaping how AI systems understand your brand’s reputation.
  • Focus on “Information Gain” and building authority signals. Think about what your brand could do or say differently about a particular topic, and with authority. Add author-bios, stats and quotes wherever applicable.
  • Restructure content for passage-level retrieval. AI systems don’t read pages, they extract chunks. Content with clear H2s, direct answers in the first couple of sentences, and specific data points is far more likely to be cited.
  • Define alternate KPIs. Move beyond rankings.
  • Diversify beyond Google. Visibility in ChatGPT, Perplexity, Reddit, YouTube, and vertical social media platforms is now part of the same discovery funnel.