Google has now openly acknowledged what practitioners have been observing for months: AI Mode attracts queries earlier in the buying journey. The implication is that it pays less - at least in direct conversion terms. That framing comes from a paid search angle. But it carries a more important signal for AEO, one that most teams haven't yet worked through.
If AI Mode is capturing research-stage intent, the content that earns citations there is structurally different from content built to win at the bottom of the funnel. And the way you measure success has to change with it.
What 'Earlier in the Journey' Actually Means for Citation Signals
Upper-funnel queries in AI Mode look different. They are broader, more exploratory, and often involve comparison or category-level questions rather than specific product intent. A user asking AI Mode to explain the difference between two types of pension structure is not the same user who is ready to open an account. The query fan-out from that kind of prompt pulls in sources that explain well, not sources that sell hard.
That shifts the citation advantage toward content with genuine educational depth - well-structured explainers, comparison frameworks, and content that answers the 'why' before the 'which'. Thin category pages optimised for bottom-funnel conversion copy are unlikely to earn those citations. The source material and the query intent simply do not match.
This is not a new idea in SEO terms, but the mechanism is different in AI search. A human Google user clicking a comparison article might bounce to a product page. An AI Mode or Perplexity response citing your explainer content may never send that user to your site at all. The citation is the exposure. That changes what 'content performance' means entirely.
The Attribution Problem Gets Harder at the Top of the Funnel
Bottom-funnel AI visibility is already difficult to attribute. Upper-funnel AI visibility is harder still. If a user encounters your brand in an AI Mode answer during a research session, closes the tab, and returns three weeks later via a branded search to convert, the AI touchpoint is invisible in most attribution models. Google Search Console AI Overview data only covers what it covers. Direct referral traffic from AI Mode is inconsistently tagged. The gap between influence and measurement is wide.
The practical response is to treat AI Mode citations at the top of the funnel the same way you treat earned media - as brand exposure that feeds pipeline indirectly. That means looking at branded search volume trends alongside AI visibility changes, rather than trying to draw a straight line from an AI Mode citation to a closed deal. Where CRM data allows first-touch or multi-touch attribution across longer windows, that is where the evidence of AI-influenced journeys will eventually surface.
For teams running AEO programmes, this is an argument for establishing visibility baselines now. If you are not tracking which queries trigger AI Mode citations for your brand or category, and you are not monitoring that consistently over time, you will have no way of knowing when that exposure increases - or what it correlates with downstream.
Content Structure for Discovery-Stage AI Queries
Upper-funnel AI queries tend to be conversational and multi-part. A user asking an AI search engine to help them understand a complex buying decision will often issue a prompt that covers several sub-questions at once. AI Mode, ChatGPT, and Perplexity all use query fan-out to pull in relevant sources for each part of the response. Content that is structured in clear, self-contained sections - each addressing a specific question - is better positioned to be extracted for individual sub-answers.
This is where heading structure, schema markup, and FAQ-style content architecture do genuine work. Not because schema is a magic citation signal - it is not - but because it makes the extraction task easier for the model. If your H2 is a question and the following two paragraphs answer it directly, that chunk of content is straightforwardly usable. If your content flows as one long narrative without clear structural breaks, it is harder to cite accurately.
For category and comparison content specifically, tables and structured lists outperform prose paragraphs for AI extraction. If you are covering a topic where users want to compare options - pricing tiers, feature sets, different approaches to a problem - format that content so a model can extract the comparison cleanly. Prose explanations below the table still matter for depth and authority, but the structured element is what gets pulled into the answer.
What Google Admitting This Changes for Organic Strategy
Google acknowledging that AI Mode occupies an earlier part of the funnel is significant not just for PPC planning but for how organic and AEO teams position their work internally. The traditional organic search model assumed that ranking well meant driving traffic that could convert. AI Mode breaks that assumption at the discovery stage. Visibility there may deliver brand exposure with no direct traffic at all.
That does not make it valueless. Brand recall, category association, and trust-building during research all influence downstream purchasing behaviour. But it does mean the ROI conversation changes. Teams running AEO programmes need to be explicit with stakeholders about what upper-funnel AI citations are expected to deliver - and what metrics will capture that over a realistic timeframe.
One concrete implication: if AI Mode is pulling research queries away from organic click-through results, and those queries were previously driving traffic to informational content, that traffic will decline. The question is not whether that decline is happening - it almost certainly is for many categories - but whether the AI citation is substituting for it in brand exposure terms. Tracking both the click volume fall and the AI citation presence together gives you a more honest picture than either metric alone.
What to Actually Do With This
Start by auditing which of your content assets are currently targeting discovery-stage intent. These are typically your 'what is', 'how does', and 'which is better' pages. Run those queries manually in Google AI Mode and in Perplexity. Note whether your content is being cited, which competitors are cited instead, and what the structure of their cited content looks like.
Where you are absent, the gap is usually one of two things: your content lacks the structural clarity needed for AI extraction, or your domain does not carry enough topical authority in that specific sub-area for the model to surface you. Both are fixable, but they require different interventions. Structural issues are addressed through content reworking. Authority gaps require a longer programme of original content, digital PR, and consistent citation-building in the relevant topic cluster.
Finally, make sure your bot access is clean. AI crawlers from Google, OpenAI, and Perplexity all need to be able to index your discovery-stage content. Check your robots.txt against GPTBot, Google-Extended, and PerplexityBot. If those crawlers are blocked on your informational content - even accidentally through a blanket rule - your AEO programme is running blind. The visibility you are trying to build simply cannot happen.