AI Search Changes In Q1 2026 [Recap]

SEJ Live’s opening panel covered three months of AI search changes from three angles. I covered the news, SEJ Founder Loren Baker covered the business case, and Managing Editor Shelley Walsh covered content strategy. The on-demand recording is available here.

The session was called “What’s Hot, What’s Not,” and our goal was to identify the Q1 changes worth acting on in Q2, and what steps you can start taking today.

AI Overviews Are Costing Clicks, But Not All Of Them

The headline number from Q1 is that clicks drop when AI Overviews appear, but the loss varies by query type. Google’s VP of Search, Robby Stein, said that when people scroll past an AI Overview without engaging, Google pulls it back for that query. The pages losing traffic are the ones answering simple questions. If someone searches for store hours or a return policy, the AI answers it, and nobody clicks through.

Shelley pointed to data from Amsive showing that branded queries with AI Overviews see an 18% increase in click-through rates. When people trust a source, they click through even when a summary is available.

She also pointed out that between half and three-quarters of all queries don’t trigger an AI Overview at all, depending on whose data you use. BrightEdge puts it at about half. Conductor puts it higher. Either way, there are entire categories of queries where you can still compete without an AI Overview in the way.

AI Mode And ChatGPT Are Both Selling Ads Now

AI Mode crossed 100 million monthly active users in the U.S. and India, with 75 million using it daily. During Q1, Google expanded how it monetizes AI-powered search, including Direct Offers in AI Mode, which lets businesses place promotions inside AI responses.

OpenAI began testing ads in ChatGPT for logged-in adult users on the Free and Go tiers. Industry reports put the early pricing at about $60 CPM with a $200,000 minimum commitment. OpenAI said the ads use the current conversation context for targeting.

Between Google and OpenAI, there are now multiple ways to place ads inside AI-generated answers. That wasn’t the case a few months ago.

Start tracking how often your brand gets mentioned in ChatGPT and AI Mode responses. You’ll want to know where you stand before deciding whether paid placement makes sense.

Replaceable Content Is What AI Threatens

Shelley’s segment drew a line between replaceable and valuable content. AI can summarize “what is SEO” or “how to change a bike chain” as well as any page that restates common knowledge. If your content is built on answering those kinds of questions, you’re competing directly with AI.

But content based on original research and firsthand experience is different. Shelley called this “golden knowledge,” borrowing a phrase from SEO veteran Grant Simmons. It’s your data and your experience. LLMs can’t generate it from training data.

Shelley said this looks like video interviews and original research, plus opinionated commentary from practitioners. She pointed to SEJ’s own changes as an example. SEJ has moved editorial toward experience-first formats and shifted revenue from programmatic to sponsorship and downloadable assets. Growing a direct audience is now the top priority.

The question to ask, she said, is why someone would click through from an AI summary to your site. If your content is a summary, there’s no reason. If it has depth, case studies, implementation detail, or nuance the summary can’t contain, that’s what drives the click.

Schema Markup Now Trains LLMs Across Platforms

Loren’s segment made the case that structured data has more value now than at any point in the last decade. Schema markup has always helped with rich snippets in Google. Now it also trains LLMs across platforms.

He shared an example of a client whose CEO shared a common name, and searching for that name plus “CEO” surfaced executives from other companies. Loren implemented organization and person schema. As soon as it went live, the correct CEO appeared in AI Overviews.

Loren ranked the structured data signals AI systems respond to. Schema markup was at the top, followed by clean heading hierarchy and semantic HTML. He put llms.txt as an emerging standard worth watching.

On markdown, Loren noted that Cloudflare had announced a new /crawl endpoint that same morning. The feature renders sites in clean HTML and markdown for LLMs, plus structured JSON. Loren’s point was that if Cloudflare is building this at the platform level, and LLMs learn from markdown, then the tooling to serve it is growing.

Getting Schema Off The Dev Backlog

Loren’s most relatable point was about internal buy-in. Anyone who’s worked with development teams knows schema tends to sit in the backlog behind other priorities. But the conversation changes when you tie technical SEO work to AI visibility.

Tell a client that AI answers depend on structured data, and that ticket moves up the sprint board. He connected this to broader executive buy-in. C-suite leaders are seeing AI Overviews and ChatGPT answers about their companies, and they’re asking questions. That attention creates an opening to secure funding for technical work that would have stalled in previous years.

For ecommerce specifically, Loren recommended the Shopify Knowledge Base App, which crawls product content and generates question-and-answer pairs.

Looking Ahead

During Q&A, the panel was asked about AI-generated content. Shelley confirmed that Search Engine Journal’s content is human-written, and we plan to keep it that way. All three of us agreed that AI works best as an augmentation tool for writers who already know their subject.

The full session, including the Q&A, is available on demand. The other two sessions from the event are also available. CallRail’s Emily Popson covered AI search KPIs in Session 2, and Forrester’s Nikhil Lai covered answer engine strategy in Session 3.

More Resources:

Featured Image: Search Engine Journal