What Actually Changed for Ecommerce SEO When AI Overviews Arrived

In January 2026, Google flagged structured data errors across several of our client sites. Not vague warnings. Specific errors, in Search Console, with item counts and page-level detail. The errors had existed before. What changed was that Google was now actively surfacing them as a priority issue rather than a secondary notice.

That timing was not coincidental.

AI Overviews pull heavily from structured data. When Google builds an AI-generated answer about a product, a service, or a how-to, it’s largely assembling that answer from structured content: schema markup, clearly labeled entities, well-organized page sections. Sites that give Google clean, parseable, well-attributed data are more likely to be cited. Sites with structured data errors are giving Google a reason to look elsewhere.

The structured data errors we fixed in January were things we’d have gotten to eventually anyway. The urgency was new.

What AI Overviews actually changed, and what they didn’t

There’s been a lot of panic content about AI Overviews replacing organic search entirely. That’s not what’s happening. Traditional blue-link results still exist. Organic rankings still drive traffic. The structure of the search results page changed (AI Overviews appear above organic results for many queries), but organic traffic didn’t disappear.

What did change:

Click-through rates dropped for informational queries. When an AI Overview directly answers a question, users sometimes get the answer without clicking. For “what is” and “how to” queries, this is real. Traffic to purely informational content has declined for some sites.

The definition of “ranking” expanded. Being cited in an AI Overview is now a meaningful result, arguably more prominent than a #1 organic ranking for some queries. This requires a different optimization target: not just ranking your page but making your content the source Google reaches for when constructing its answer.

Transactional and navigational queries are mostly unaffected. Users searching “best running shoes under $100” or “Shopify migration service” are still getting traditional results. They want options, not a summarized answer. AI Overviews are less prevalent in these results, and where they do appear, they typically include product or service recommendations with links.

Brand mentions without links now matter. LLMs, including the models behind Google’s AI Overviews and ChatGPT, are trained on web content. A brand mentioned repeatedly in authoritative sources, forums, Reddit threads, and industry publications develops a presence in that training data even without a direct link. This is the “unlinked mention” concept that SEO practitioners have discussed for years, now directly relevant because the models powering AI search have internalized those mentions.

What we’re doing differently

Across our client SEO programs, three things changed in how we approach the work.

Structured data is no longer optional. It was always a good practice. Now it directly affects whether your content is eligible to be cited in AI Overviews. We audit and implement schema markup (Product, Organization, FAQPage, BreadcrumbList, Article) as a standard part of every engagement, not an optional add-on. We also monitor Search Console’s structured data report on a monthly basis rather than quarterly.

FAQPage schema on every key landing page. AI Overviews frequently pull from FAQ schema because it maps directly to the question-answer format of AI-generated responses. Adding properly structured FAQ sections to service and solution pages, with questions drawn from real user queries, creates extraction opportunities for AI answers. We’re implementing this across service pages and high-value landing pages as a standard element.

Content structure has shifted toward direct answering. The format that gets cited in AI Overviews is the format that directly answers a question at the top of a section, then provides supporting detail. We’ve moved away from longer discursive introductions toward a structure where the first sentence of each major section is a complete, extractable answer. This helps with AI citation and also tends to improve regular search snippet selection.

E-E-A-T: experience is now the hardest to fake

Google’s E-E-A-T framework (Experience, Expertise, Authoritativeness, Trustworthiness) has been part of their quality guidelines for years. The first E, Experience, was added in 2022 and is now the one that matters most in practice.

Experience means demonstrating that the content author has direct, first-hand experience with the topic. Not that they researched it, not that they’ve written about it before. That they’ve done it.

This is the hardest signal to fake and the easiest to demonstrate if you actually have it. A blog post about Shopify migration that includes specific problems we encountered on a real project, specific data from real performance benchmarks, and specific decisions we’d make differently. That content reads differently from a post assembled from research. The signals that it’s real experience are distributed throughout the piece: details that couldn’t have come from a Google search, specific numbers, things that went wrong.

It’s also the signal that generic content farms can’t produce at scale. Which is presumably the point.

We’ve pushed our content program explicitly in this direction: more posts grounded in real client work, specific numbers where we can share them, honest assessments of what worked and what didn’t. Not just because it performs better, but because it’s more useful content.

Reddit and forums are now part of the SEO landscape

One of the more interesting developments of the last 18 months: Reddit results appearing in AI Overviews and LLM training data at a meaningful rate. Google’s content licensing deal with Reddit and the general trend of forums appearing in AI answers has made community-driven content more visible than it’s been in years.

The practical implication for brands: what people say about you in forums and community spaces now has a path into AI-generated answers that it didn’t have before. This isn’t something you directly control, but it is something to be aware of, and to factor into how you approach brand reputation and community engagement.

For ecommerce brands with genuine customer communities or strong product followings, this is an opportunity. Authentic user-generated discussion about your products creates the kind of third-party context that LLMs draw on when constructing answers.

What hasn’t changed

The fundamentals still work. Fast sites with clean technical SEO, quality backlinks from relevant sources, and well-structured content that directly answers user queries still rank. They also get cited in AI Overviews more than slow, messy, thin-content sites. The core investment in technical health, content quality, and link authority is as important as it’s ever been.

Transactional intent is still about traditional results. For the keywords that actually drive ecommerce revenue (product category searches, brand comparisons, service-intent queries), AI Overviews haven’t fundamentally changed the game. Users in buying mode want options and specifics, not a summarized answer. The traditional organic result remains the primary driver.

Quality beats quantity. The Helpful Content Update (2023) already shifted the calculus away from high-volume thin content. AI Overviews reinforce this. A well-researched, experience-grounded post that clearly answers a specific question in detail is worth more than ten generic posts targeting similar keywords.

If you’re running a Shopify store and want to understand where your SEO program stands relative to these changes (structured data health, content structure, E-E-A-T signals), a site audit will give you a specific picture. Or take a look at how we approach SEO for ecommerce if you want to understand what an ongoing program looks like.

Last Updated: June 2026