Small businesses are being sold a new pile of SEO work: special AI schema, llms.txt files, “citation-ready” paragraph chopping, and packages promising placement in Google AI Mode.
Google just made the sales pitch much harder.
In July 2026, Google published an official guide to optimizing for generative AI features. Its message is unusually direct. You don’t need secret technical tricks to appear in AI Overviews or AI Mode. You need a crawlable, useful website with clear business information and content worth citing.
That doesn’t mean AI search can be ignored. Google says AI Mode uses a query fan-out technique that runs multiple related searches to answer complex questions. A page can now compete across parts of a buyer’s question that were once separate searches.
The practical response isn’t buying every new acronym. It’s knowing which work helps and which work is theater.
Myth 1: You need special AI schema
There is no “AI Overview schema” or “AI Mode schema.” Google says structured data isn’t required for generative AI search, and there is no special Schema.org markup you need to add.
That doesn’t make structured data useless. Google still recommends using supported markup when it accurately describes visible page content. Product, local business, organization, event, and other supported types can help Google understand a page and make it eligible for specific search appearances. Google’s structured data guidelines also say the markup must represent the page’s main content and cannot mislead people.
Here’s the business-owner test: ask your vendor which Google-supported search feature the proposed markup qualifies your page for. Then ask to see it pass Google’s Rich Results Test. If the answer is vague language about “feeding the AI,” don’t pay extra for it.
Put the budget toward accurate service details, prices where appropriate, product availability, locations, authorship, and customer proof. Mark up that real information with a supported type when it fits. Schema should describe substance, not replace it.
Myth 2: Every small business needs an llms.txt file
An llms.txt file is a proposed text file intended to help language models find selected site content. It has attracted attention because it sounds like robots.txt, a standard website owners already recognize.
Google’s new guidance specifically says businesses can ignore unnecessary AI text files such as llms.txt when optimizing for Google Search. Google also says pages must be indexed and eligible to appear with a snippet to show in its generative features. A separate AI file doesn’t override those requirements.
If adding the file takes 15 minutes and you have a clear non-Google use for it, fine. But don’t let an agency package it as a major SEO deliverable. It will not repair pages blocked by noindex, weak internal links, duplicate service copy, or a site Google cannot render.
A better use of the same hour is checking your five highest-value service URLs in Search Console’s URL Inspection tool. Google’s documentation says that tool can show index status, rendered pages, and requested resources. Fix the pages that aren’t accessible before creating files nobody requires.
Myth 3: AI search requires “chunking” every page
Some consultants tell owners to rewrite every page into short, isolated blocks so an AI system can extract them. Google now lists content “chunking” among the AEO and GEO hacks businesses can ignore.
Readable structure still matters. A buyer comparing two commercial HVAC systems should be able to find cost, lead time, warranty, installation requirements, and service coverage without digging through a wall of text. Headings and short sections help that person.
The mistake is writing for extraction instead of understanding. Twenty disconnected 40-word answers can leave out the conditions that make advice correct. A roofing price depends on roof size, pitch, material, tear-off requirements, decking condition, and market. Removing that context may make a sentence quotable while making it less useful.
Organize pages around actual buying questions. Answer each question directly, then include the limits, evidence, and next step a customer needs. That’s good web writing whether the visitor comes from a blue link, an AI answer, or a referral.
Myth 4: More AI-written pages create more AI visibility
Volume is tempting. A tool can produce pages for every town, service variation, and question in a spreadsheet before lunch.
Google’s guidance on generative AI content says automation can be useful for research and structure, but generating many pages without adding value for users may violate its scaled content abuse policy. Google’s spam policy defines scaled content abuse as producing many pages primarily to manipulate rankings, regardless of whether humans, automation, or both created them.
For a local company, 60 city pages that swap only the place name aren’t 60 assets. They’re 60 pages to maintain, fact-check, internally link, and keep out of each other’s way.
Build one strong service page first. Add location pages only where you can include material differences: service boundaries, local regulations, staff, project photos, response times, customer stories, or office details. Google’s people-first content guidance asks whether a page demonstrates firsthand expertise and gives visitors a satisfying answer. A page assembled from generic paragraphs rarely clears that bar.
Use AI to reduce blank-page time, not responsibility. A person who understands the service should verify every claim before publication.
Myth 5: Buying mentions will manufacture citations
AI citation services sometimes promise to place your company name across forums, list sites, and low-quality articles so models will treat the brand as authoritative.
Google tells businesses not to pursue inauthentic mentions. Its spam policies also prohibit link spam such as paid links created for ranking purposes.
Real mentions are valuable because they represent real activity. A chamber directory listing, trade association profile, local news story, supplier page, conference presentation, or customer case study can confirm who you are and what you’ve done. The distinction is simple: would the mention exist if search engines and AI systems could not see it?
Small businesses already have raw material for legitimate visibility. Sponsorships, certifications, partnerships, community work, completed projects, and expert interviews are publishable facts. Document those, keep your company name and contact details consistent, and earn coverage from organizations with an actual audience.
Myth 6: You need a separate version of your site for AI crawlers
Google says its generative search features use publicly accessible, crawlable content. The page also needs to meet Google’s normal technical requirements and be eligible to appear in Search with a snippet.
Serving simplified or altered content only to a crawler creates risk without fixing the real website. Google’s spam policies define cloaking as showing different content to search engines and users to manipulate rankings. Even when intent isn’t malicious, maintaining two versions creates more places for prices, policies, and availability to fall out of sync.
Make the normal page easy to access and understand. Use server-rendered HTML or make sure important JavaScript content renders reliably. Keep primary information visible to customers. Return a working HTTP status. Don’t hide the service description behind an interaction a crawler may never complete.
You can control how much Google may use in previews. Google documents nosnippet, data-nosnippet, max-snippet, and noindex as controls that also apply to AI features. Use those controls carefully. Blocking snippets can also remove a page’s eligibility for generative features, so this is a policy decision, not a casual plugin setting.
Myth 7: An agency can guarantee AI Overview placement
No honest provider can guarantee an organic placement that Google generates query by query. Google’s own advice for hiring an SEO says Google never accepts money to include or rank sites in organic results.
AI Mode makes guarantees even shakier because one user can ask follow-up questions and receive different supporting links as the topic changes. Google’s explanation says query fan-out searches across multiple subtopics and data sources. There isn’t one fixed result to “own.”
A useful agency can guarantee work it controls: correcting crawl problems, improving priority pages, adding measurement, interviewing your experts, documenting customer proof, and reporting what changed. It can run repeatable visibility checks for important questions. It cannot promise that Google will cite your company on a specific date.
Ask for a list of deliverables, the business reason for each one, and the measurement plan. Walk away from screenshots of cherry-picked AI answers presented as permanent rankings.
What Google actually wants you to improve
Google’s guide is less exciting than the myths, which is probably why it’s more useful. The priorities are familiar:
- Access: Important pages must be crawlable, indexed, and eligible for snippets. Check canonical tags, robots directives, HTTP responses, and internal links.
- Useful information: Publish original service knowledge, clear answers, accurate business facts, and evidence that reflects real experience.
- Page experience: Google recommends pages that are easy to navigate and work well across devices. Remove intrusive clutter and make the next action obvious.
- Visible facts: Keep structured data consistent with the text customers can see. Update old prices, hours, team information, and availability everywhere they appear.
- Measurement: Google says generative AI performance can be monitored in Search Console. It also reports traffic from AI features within Search Console’s performance reporting, so connect search trends to calls, forms, booked appointments, and sales.
Start with the pages closest to revenue. For most service businesses, that’s the homepage, core service pages, top location pages, pricing or process page, and contact page. Fix broken fundamentals there before commissioning another batch of AI-focused blog posts.
A 30-day plan that doesn’t waste money
During week one, inspect your ten most valuable URLs. Confirm they return a 200 status, can be indexed, have self-consistent canonical tags, and appear in the XML sitemap. Make sure every page is reachable through a normal internal link.
In week two, interview the person who sells or delivers the work. Add the questions prospects ask, honest qualification criteria, common complications, process details, and proof. Link supporting claims to original sources.
Use week three to reconcile your public facts. Compare the website, Google Business Profile, major directories, proposals, and social profiles. Correct conflicting hours, phone numbers, service areas, names, and offer details.
In week four, establish a baseline. Record Search Console clicks and impressions for priority pages, then measure qualified forms, calls, bookings, and sales. Google’s documentation notes that Search Console clicks and Analytics sessions will not match exactly, so watch the direction of the data instead of forcing identical totals.
AI search has changed the result page. It hasn’t changed the need to run a credible business, explain your work clearly, and maintain a website customers can use.
If you’re being pitched an AI SEO package and want a straight answer on what’s useful, get started with Your Web Team. We’ll find the work that can improve leads and cut the deliverables that only sound new.
