- ChatGPT Search expanded its citation pool in Q2 but became more selective — it now prefers sources that answer a specific question directly rather than covering a topic broadly.
- Google Gemini's AI Overviews began pulling from Google Business Profile content more aggressively, making GBP updates a direct search-ranking lever for local businesses.
- Perplexity shifted toward rewarding 'answer density' — posts that define terms, include numbered steps, and use structured headers get cited more than long narrative articles.
- Schema markup (especially FAQPage, HowTo, and DefinedTerm) now functions as a citation signal in AI engines, not just a traditional SEO tool.
- First-person expertise signals — real data, named authors, specific examples — are increasingly how AI engines distinguish trustworthy sources from filler content.
- Owner-operators who publish consistent, specific content on a narrow topic outperform generalist sites in AI citation rates, even with far lower domain authority.
The Quarter AI Search Grew Up
For two years, the advice on AI search optimization was mostly "wait and see." Q2 2026 ended that. Three platforms — ChatGPT Search, Google Gemini, and Perplexity — each made structural changes that are now producing measurable differences in which content gets cited and which disappears. If you publish content for your business, these changes affect you directly.
This isn't a roundup of press releases. It's a practical breakdown of what actually shifted, what the evidence shows, and what to do before the next quarter compounds the gap.
ChatGPT Search: Wider Pool, Stricter Filter
OpenAI expanded the crawl footprint of ChatGPT Search in Q2, indexing a broader range of domains — including smaller business sites that were previously invisible. That sounds like good news, and it is, but the citation behavior tightened at the same time.
What changed: ChatGPT Search now shows a strong preference for content that answers a specific question in the first 150 words. Pages that bury the answer inside a long preamble — or that treat a topic broadly without a clear thesis — are being crawled but not cited. The model appears to be running a relevance pass that rewards directness.
What this means for you: Your blog posts and FAQ pages need to lead with the answer, not the context. If your post title is "How to Handle a Late Delivery Complaint," the first paragraph should state exactly how to handle it. The explanation and nuance can follow. Posts that open with "In today's fast-paced e-commerce environment..." are being skipped.
There's also a growing preference for named authorship and specific credentials. Generic "Koira Team" or "Staff Writer" bylines are weaker citation signals than posts attributed to a named person with a short bio that establishes why they know this. If you're a salon owner writing about managing no-shows, say that.
Google Gemini: GBP Is Now a Search Signal
The most operationally significant change of the quarter came from Google. Gemini's AI Overviews — the AI-generated summaries that appear above traditional results for local and informational queries — started pulling content directly from Google Business Profile posts, Q&A sections, and service descriptions at a measurably higher rate than before.
This is a big deal for any business with a physical location or a local service area. It means the content you post to your GBP isn't just for people who find your listing — it's now being fed into the AI layer that answers questions like "best [service] near me" or "does [business type] in [city] offer [thing]."
What changed specifically:
- GBP service descriptions are being used to populate Gemini answers about local service availability.
- GBP Q&A content — the questions and answers that appear on your listing — is being cited in AI Overviews when the question matches a user query.
- Businesses that post to GBP weekly are appearing in AI Overviews for informational queries at higher rates than dormant listings.
What this means for you: Treat your GBP like a mini-blog. Write service descriptions that answer real questions, not just list features. Populate the Q&A section with questions your customers actually ask — and answer them in plain, specific language. Post updates weekly, even short ones. This is one of the highest-leverage content moves a local business can make right now, and most aren't doing it.
For owner-operators managing GBP updates manually, this is the kind of repetitive weekly task that self-driving software handles cleanly — write the post once, set the cadence, and stop thinking about it.
Perplexity: Answer Density Wins
Perplexity made the most technically visible changes this quarter. Its citation algorithm now strongly rewards what can be called answer density — the ratio of directly useful information to total word count.
Earlier analysis of Perplexity's indexing behavior covered how the platform crawls and caches content. What's new in Q2 is the citation weighting. Perplexity is now measurably preferring:
- Numbered step sequences (HowTo-style content)
- Defined terms with clear one-sentence definitions
- Specific data points — percentages, timeframes, named examples
- Short headers that match natural-language questions
Long narrative posts — the 2,000-word "ultimate guide" format — are getting crawled but cited at lower rates unless they contain dense answer sections. A 900-word post with five clear definitions, a numbered process, and a FAQ block is outperforming a 2,500-word essay on the same topic.
What this means for you: Structure is now substance. Adding a FAQ section, a HowTo block, and a few defined terms to your existing posts isn't just good schema hygiene — it's directly increasing the probability that Perplexity surfaces your content in an answer. This is a retrofit you can do to existing posts, not just new ones.
The Schema Signal Is Real Now
For years, schema markup was treated as a nice-to-have — something that might help with rich snippets but wasn't worth obsessing over. That changed this quarter.
FAQPage, HowTo, and DefinedTerm schema are now functioning as explicit quality signals for AI engines, not just formatting hints for traditional search. The mechanism appears to be that structured markup makes it easier for AI systems to extract and verify specific claims — which increases the confidence score for citing that content.
The practical implication: if you're publishing content without schema, you're leaving citation probability on the table. The markup doesn't need to be elaborate. A simple FAQPage block on a post that already has a FAQ section, or a HowTo block on a post that already lists steps, takes minutes to add and measurably affects how AI engines process the page.
If your site is on Shopify, WordPress, or a GoDaddy Airo site, there are tools that handle this automatically. The manual version is still worth doing if you're not using one.
First-Person Expertise: The Signal That's Hardest to Fake
Across all three platforms, one signal emerged as the most durable differentiator: genuine first-person expertise. Real data from your own business. Specific examples with named context. Opinions that could only come from someone who has actually done the work.
AI engines are getting better at identifying content that could have been written by anyone about anything versus content that reflects actual experience. The former is being deprioritized. The latter — a salon owner writing about their specific no-show policy and what changed when they implemented it, with real numbers — is being cited at higher rates even from low-authority domains.
This is counterintuitive if you've been optimizing for SEO in the traditional sense, where domain authority and backlinks dominated. In AI search, a small site with genuine expertise on a narrow topic is now competitive with large sites publishing generic content at scale.
The actionable version: Write from what you actually know. If you run a pet grooming business, write specifically about what you've learned about managing seasonal demand, not a generic post about "tips for pet groomers." Specificity is now a ranking signal.
What to Do Before Q3
Here's the prioritized action list based on what actually changed this quarter:
Audit your top 10 posts for answer-first structure. Does each post answer its core question in the first 150 words? If not, rewrite the opening.
Add FAQ and HowTo schema to existing posts. You don't need new content — retrofit the structure onto what you already have.
Update your GBP weekly. Write service descriptions that answer questions. Populate the Q&A section. Post short updates. This is now a direct Gemini signal.
Add named authorship to posts. A real name and a one-sentence bio that establishes expertise improves citation rates across all three platforms.
Replace generic intros with specific ones. Cut the preamble. Start with the answer, the data point, or the specific claim.
Identify your three narrowest expertise areas and publish specifically in those. Broad coverage at low depth is losing to narrow coverage at high depth in AI citation behavior.
The gap between businesses that adapt to these changes and those that don't is compounding quarter by quarter. The good news is that most of these changes reward the kind of content that's genuinely useful — which means the work you put in has value beyond any single algorithm update.
The Bigger Pattern
Looking across all three platforms, the direction is consistent: AI search engines are converging on a model that rewards specific, structured, expert-attributed content and deprioritizes generic, broad, anonymous content. This isn't a temporary quirk — it's the logical outcome of training models to distinguish useful answers from filler.
For owner-operators, this is actually an advantage. You have genuine expertise that large content farms can't replicate. The question is whether you're publishing in a way that lets AI engines recognize and surface it. Based on Q2's changes, the answer for most small businesses is: not yet, but it's fixable.
“A small site with genuine expertise on a narrow topic is now competitive with large sites publishing generic content at scale — AI search changed that this quarter.”
| Area | Pre-Q2 approach | Post-Q2 approach |
|---|---|---|
| Post structure | Long preamble, broad coverage, answer buried mid-page | Answer in first 150 words, specific thesis, explanation follows |
| Schema markup | Optional nice-to-have for rich snippets | Active citation signal — FAQPage, HowTo, DefinedTerm directly affect AI sourcing |
| Google Business Profile | Static listing updated occasionally for NAP accuracy | Weekly-updated content source feeding Gemini AI Overviews directly |
| Authorship | Generic byline ('Staff Writer', 'Team') acceptable | Named author with expertise bio improves citation rates across all AI engines |
| Content depth strategy | Broad topical coverage at moderate depth to capture keyword range | Narrow expertise depth on specific topics outperforms broad coverage in AI citations |
| Post length | Longer = more authoritative; 2,000+ words standard target | Answer density beats word count; 900-word structured post outperforms 2,500-word narrative |
How to adapt your content for Q2 2026 AI search changes
- 01Audit your top posts for answer-first structure. Open your ten highest-traffic posts and check whether each one answers its core question within the first 150 words. Rewrite any opening that starts with context-setting or preamble — move the answer to the top, then explain it.
- 02Add FAQPage and HowTo schema to existing content. You don't need to write new posts — retrofit structured markup onto content you already have. If a post has a FAQ section, add FAQPage schema. If it lists steps, add HowTo schema. This directly increases AI citation probability without changing the visible content.
- 03Update your Google Business Profile weekly. Write service descriptions that answer specific customer questions, populate the Q&A section with real queries and plain-language answers, and post a short update at least once a week. GBP content is now a direct input to Gemini AI Overviews for local queries.
- 04Add named authorship with an expertise bio. Replace generic bylines with a real name and a one-sentence bio that establishes why that person knows the topic — 'Sarah runs a 3-location dog grooming business in Austin' is more effective than 'Content Team.' This applies across ChatGPT Search, Gemini, and Perplexity citation behavior.
- 05Identify your three narrowest expertise areas and publish specifically. Pick the three topics where you have genuine first-hand experience and publish content that could only come from someone who has actually done the work — real numbers, specific examples, named outcomes. Narrow depth now beats broad coverage in AI search citation rates.
- 06Add DefinedTerm schema to posts that define concepts. Any post that explains what something is — a process, a term, a framework — should include DefinedTerm schema on that definition. This makes the definition extractable by AI engines and increases the chance your page is cited when someone asks what that term means.
- 07Retrofit high-traffic posts before writing new ones. The Q2 changes reward structure and expertise signals, both of which can be added to existing content. Prioritize updating your best-performing posts before investing in new long-form pieces — the ROI on retrofitting is higher right now than publishing fresh generic content.