At a glance
- One strategy covers SEO, AI search and the goal behind both, turning a visitor into a customer. It comes straight from Google's published guidance: write your website for people, not for search engines or AI.
- Nothing else matters if search engines and AI tools are not allowed to read your website. Check that first.
- Say plainly what you do and where you do it, and answer the questions customers actually ask in enough detail to be useful.
- Write from your own experience. Generic articles cost nothing to produce now and are worth about that much.
- Keep your business details the same everywhere online, earn real reviews, and skip anything sold to you as an AI trick.
If people are using ChatGPT, Google AI Mode, Copilot, Claude, Perplexity and the rest to find businesses, what should you actually do? Less than the marketing around "AI optimization" suggests. Google's guide to generative AI search says established SEO practice still applies and that you do not need special AI files, AI markup or a separate content strategy. The better plan is to make the website you have clear, useful, open to crawlers and backed by real expertise, rather than to run a second optimization program for AI. Here is where we would start.
1. Make sure search systems can reach the site
Nothing else matters if the crawler cannot get in. Google's technical requirements are that Google's crawler, the automated program that reads websites, can open the page, the page loads without an error, and it contains text Google is allowed to index. Meeting them does not guarantee indexing. Failing them rules it out. AI search adds crawlers to the list. OpenAI's OAI-SearchBot is one, and sites that block it do not appear as sources in ChatGPT search answers. Checking who can get in now means checking more than Google's and Bing's crawlers. For a typical business site we would allow the legitimate search crawlers and decide separately about crawlers used for AI model training, as covered in part three of this series.
2. Say plainly what the business does
One homepage opens with "Quality. Integrity. Service." Another opens with "Commercial and residential roofing serving Boise and the Treasure Valley, including roof replacement, storm repair and metal roofing." The first sounds polished. The second tells a search system something it can use. Every search system has to work out what you are, what you offer and when you are the right answer, and that should not take detective work. State your products or services, your service area, your specialties, the industries you serve, your qualifications, what sets you apart and how to engage you. That means replacing vague marketing language with information, which is a different thing from keyword stuffing.
3. Answer the questions customers actually ask
People search conversationally now. Instead of "wool rug cleaner Boise" they ask whether anyone near Boise specializes in cleaning a handmade wool rug with pet urine damage. A site that discusses wool rugs, pet contamination, delicate fibers and its service area gives an AI far more to work with than a generic "rug cleaning" page. That does not mean a thin page for every imaginable question. Google warns against producing large amounts of content mainly to attract traffic, automated content included, and asks for content made to help people. Aim for coverage, not volume: the questions customers repeatedly ask, the problems you actually solve, and enough detail that a reader learns something.
4. Publish what you know from experience
This may be the biggest opening generative AI has created. Generic information now costs nothing. Anyone can generate "10 tips for maintaining your roof" and publish it in five minutes, which is exactly why it is worth so little. Experience is different. A roofer knows which materials fail early under the local mix of snow, sun and temperature swings. A CPA knows the three bookkeeping problems that show up every time it takes over a small construction company's books. A web company knows which accessibility failures it keeps finding on sites built on one particular platform. Google's guide calls this non-commodity content and gives it particular weight, alongside its longstanding advice to write original, people-first material. Write about what you actually know.
5. Give your important subjects real depth
You do not need to answer every question, but you do need to cover the subjects you want to be known for. An HVAC company that calls geothermal a specialty and gives it one sentence has left a search system almost nothing to retrieve. A company that explains what geothermal systems are, which properties suit them, installation and maintenance, common problems, expected life, local climate factors and the questions customers ask has documented its expertise. That is simply writing down what you know.
6. Use clear page structure
Give pages descriptive titles, meaningful headings and a logical order, so a reader can tell at a glance what a page covers. The same structure serves accessibility, usability and search at once. "Can wool rugs be steam cleaned?" tells an AI far more than "More information." Clean page code lets an AI read the relationships between headings, paragraphs, lists and links. There is no need to break paragraphs into "AI-friendly chunks." Google says its AI features do not require it. Organize for people. AI uses the same structure.
7. Use structured data where it genuinely applies
Structured data, also called schema markup, is code added to a page that describes certain kinds of content explicitly: an organization, a local business, an article, a product, an event. Google explains that it gives search engines clear clues about meaning and can make pages eligible for enhanced results. Google is equally clear that markup guarantees nothing and must describe content that is actually on the page, and that no special AI schema exists for its generative features. Add structured data because it accurately describes your content, not because someone says AI requires it.
8. Keep business information consistent across the web
Your website is one source among many. Business profiles, directories, review platforms, industry bodies and local publications all describe you, and when five sources describe you five ways, an AI has to guess which one is right. Check the basics regularly: legal business name, address, phone, website, service area, category, hours and main services. Google Business Profile matters most for local businesses because Google's AI features draw on Google's search and local data. The same principle applies everywhere else.
9. Earn real reviews and third-party mentions
Your website is what you say about yourself. Independent sources say something different. Reviews show recurring customer experience. Industry organizations establish credentials. Local publications add context. Links from related websites show that you are part of a real industry and a real place. None of this excuses bought links or fake reviews, which were bad SEO before and do not improve by being renamed GEO. The goal is a real reputation reflected online. If twenty customers independently say you do excellent emergency furnace repair, that is meaningful information about what you are known for.
10. Make expertise attributable
Who wrote this, and why should anyone trust it? Substantive content benefits from its source being visible: author names where appropriate, professional experience, credentials, company history, firsthand examples, project experience and references to authoritative sources. Google's people-first guidance asks site owners to consider the who, how and why behind content and whether visitors can tell who made it. As the web fills with anonymous AI-generated articles, showing who is behind yours counts for more.
11. Keep important information current
Stale information misleads people and AI alike. When service areas, staff qualifications, published prices or the rules an article describes change, update the page, and keep structured data in step with what is visible. Google's guidelines require that it be accurate and current. Do not change publication dates to make old content look fresh. Google warns against it. Update when something meaningful has changed.
12. Do not hide information behind unnecessary technology
A visually impressive site can still make its information hard for automated systems to retrieve. Important content belongs in a technically sound, crawlable page, not behind a complicated interaction, a login, an inaccessible script or a third-party widget when it could simply be on the page. Here technical SEO, usability and accessibility all ask for the same thing.
13. Maintain an XML sitemap and sensible internal links
Search systems need paths to your pages. An XML sitemap, a file that lists your pages for search engines, helps them find important pages, especially on larger or frequently updated sites. Internal links matter as much. A good resource that no other page links to is an island. Link service pages to the resources that answer common questions, link those resources back to the service, and organize related resources under a topic page. Visitors get a useful path from one page to the next, and an AI can see which pages belong together.
14. Don't rush to create an llms.txt file because somebody says you need one
The proposed llms.txt format has had a lot of attention as a way to describe a website to AI systems. It may prove useful to some service someday. It is not a requirement of AI search today, and Google clarified in June 2026 that it does not use the file and that having one neither helps nor harms visibility in Google Search. The same test applies to every new "AI optimization" feature. Which AI system uses it? Where does that provider document it? What does it actually affect? Is there evidence it improves visibility? Those four questions settle most sales pitches.
15. Measure what you can
AI visibility is becoming measurable. Bing Webmaster Tools added an AI Performance report in February 2026 showing when your content is cited in AI answers across Copilot, Bing and partner experiences, and Microsoft expanded it in June with topics, user intents and citation share. Google Search Console reports impressions in Google's generative AI features and added multimodal search data in September 2026. Your analytics may also show referrals from AI platforms, with one caveat: an AI system can answer a question without anyone clicking through. Visibility now includes being cited as a source as well as being visited, and the tools for measuring the first are still catching up.
16. Don't build around one company's behavior
ChatGPT matters today. So do Google, Microsoft, Claude and Perplexity, and their relative importance will shift. A content strategy built on a theory about what one version of one AI system prefers is fragile. Information that is accessible, accurate, specific, useful, well organized and backed by real expertise keeps its value whichever AI system wins.
A practical priority list
Google has given site owners the same advice for as long as it has published guidelines for them: write pages for people, not for search engines. Its guidance on helpful content still rests on it, and its 2026 guide to generative AI search says the same thing about AI. On a technically sound site, that is the whole strategy. Everything on this list is that advice applied.
For most small and midsize business websites, we would do the work in this order.
- Make the site technically crawlable and indexable.
- State clearly what the business does, where it operates and what makes it relevant.
- Build out thin service and product pages.
- Publish useful content based on real customer questions and your own expertise.
- Keep business profiles, reviews and other off-site references accurate.
- Use appropriate structured data and sound technical SEO.
- Allow legitimate AI search crawlers where AI discovery fits your goals.
- Start watching AI citations and referral traffic as the tools mature.
Not on the list: creating an llms.txt file, rewriting paragraphs into AI-sized fragments, publishing hundreds of generated FAQs, buying an "AI authority" package or making a separate page for every way a question can be phrased. Those are easier to sell because they sound new. The fundamentals solve the actual problem.
The larger lesson
Search has changed many times since businesses started optimizing for Google: mobile, local, voice, featured snippets, machine-learned ranking, and now generative AI. Each shift brought real technical details worth learning, and each brought claims that everything you knew had become obsolete. That has rarely been true. AI search changes how information is retrieved, combined and presented, and it adds crawlers, interfaces and reports. It has not changed the objective. Your website has to offer information worth finding. If your business is clearly described, open to crawlers, genuinely useful, well regarded and backed by real expertise, you are already doing most of what AI search systems need in order to recognize you as a good answer. That is not a new trick, just good SEO.
In this series
- SEO and AI: what's real and what's hype
- How does an AI assistant decide which businesses to recommend?
- Should you block AI bots from your website?
- How to make your website visible in AI search without chasing AI SEO tricks (this article)