What are the top ten things to check on your website if you want AI tools to recommend your business?

If you want your business to be part of conversations with AI, here are the 10 most important things to check on your current website.

Get rid of page builders, check your site can actually be crawled, state very clearly what you do, answer real questions, make sure there are people behind the content, use schema (although that’s in debate), make sure your headings are all structured correctly, get mentioned off-site, use case studies and make sure you are mentioned elsewhere consistently. Phew, that’s a lot.

Get some help with fixing these

• Check 1: Are AI crawlers actually allowed in?
• Check 2: Is your site built on a page builder?
• Check 3: Does your site clearly state what you do and who you do it for?
• Check 4: Does your content answer real questions?
• Check 5: Is there a real person behind the content?
• Check 6: Do you have schema markup?
• Check 7: Do your pages have a logical heading structure?
• Check 8: Are you mentioned anywhere else on the web?
• Check 9: Do you have any real case studies or outcomes?
• Check 10: Is your NAP consistent across the web?

This is not a general SEO checklist. There are plenty of those. This is specifically about what AI tools – ChatGPT, Perplexity, Google’s AI Overviews, Claude – need from your website before they will confidently mention you in a response.

The distinction matters because the two things are related but not the same. You can have a site that ranks reasonably well in traditional search and still be completely absent from AI recommendations. The reason is that AI tools are not just matching keywords to queries. They are trying to build a confident, well-evidenced picture of who you are, what you do, and whether you are worth mentioning. If your site makes that picture hard to build, you will not get mentioned – regardless of how long you have been in business or how good you actually are.

Work through these checks. Some will take five minutes. Some will need your developer. All of them directly affect whether AI tools include you when recommending businesses in your field.

Check 1: Are AI crawlers actually allowed in.

Before anything else – check whether AI crawlers can actually reach your site. If they cannot get in, nothing else on this list matters.

AI tools index the web using their own crawlers. OpenAI uses GPTBot. Anthropic uses ClaudeBot. Google uses a range of crawlers for its various AI products. All of them check your robots.txt file before they start crawling, and if your robots.txt tells them to stay out, they will.

A surprising number of sites block AI crawlers accidentally. Some security plugins add broad disallow rules without the site owner realising. Some developers set up robots.txt during a staging build and never revisit it. Some site owners have deliberately blocked AI crawlers for data privacy reasons but have not thought through the commercial implication of doing so: if you block the tools that feed AI recommendations, you are invisible to those recommendations.

How to check: Visit yourdomain.com/robots.txt in your browser. Look for any lines that say User-agent: GPTBot, User-agent: ClaudeBot, or User-agent: * followed by Disallow: /. If you see those, AI crawlers are being turned away at the door.

Check 2: Is your site built on a page builder.

This one is specific to AI visibility rather than search in general, and it is worth understanding why.

When an AI crawler visits your site, it reads the raw HTML. It is looking for your content – what you do, who you serve, what makes you credible. On a clean, hand-coded site, that content is easy to find. On a page builder site – Elementor, Divi, WPBakery and their various relatives – your content is buried inside layer upon layer of structural markup that exists purely to make the drag-and-drop editor work.

To put some numbers on it: I pulled the source code from a basic page built in a popular page builder. The page had around 10,000 readable characters of content. The HTML that wrapped it was nearly a quarter of a million characters. Your actual content represented less than 5% of what the crawler had to process to find it.

AI crawlers work with token budgets – they allocate a finite amount of processing to each page. A page builder site spends most of that budget on structural noise before the crawler gets anywhere near your content. The signal-to-noise ratio is genuinely poor, and it directly reduces how clearly an AI tool understands what your page is about.

A hand-coded, semantically structured site puts your content front and centre. An H1 is the page subject. A paragraph contains what it says it contains. There are no redundant wrapper divs with 40-character class names in between. The crawler gets to your content faster and extracts more meaning from less effort.

How to check: Right-click any page on your site and select “View page source.” Search for “elementor”, “et-pb” (Divi), “wpb-wrapper” (WPBakery), or “fl-builder” (Beaver Builder). Alternatively, install the free Wappalyzer browser extension – it identifies the technology stack behind any site in one click.

Check 3: Does your site clearly state what you do and who you do it for.

AI tools need to categorise you. When someone asks “who are the best employment law solicitors in the East Midlands?”, the AI is not running a directory lookup – it is drawing on everything it has absorbed about businesses in that category. To include you in that answer, it needs to have clearly understood, from your website, that you are an employment law solicitor, that you operate in the East Midlands, and that you are worth recommending.

If your homepage says “we help businesses achieve their potential through people-focused solutions”, the AI cannot categorise you confidently. That sentence could describe an HR consultancy, a coaching business, a training company, a recruitment firm, or a dozen other things. Vague positioning makes for vague categorisation – and vague categorisation means you do not get cited when a specific question is asked.

Your homepage, your about page, and your service pages should all state plainly what you do, where you do it, and who your clients are. Not as a tagline. As a clear, literal description that a machine could extract and use.

How to check: Read your homepage with fresh eyes and ask: if a machine read only this page and nothing else, could it accurately describe my business in one sentence? If the answer is no, your positioning is too vague to be cited confidently.

Check 4: Does your content answer real questions.

AI tools are trained to surface content that answers questions. This is not metaphorical – the entire architecture of these tools is built around matching questions to useful answers. If your site does not contain answers to real questions your buyers are asking, there is nothing there to pull from.

Most B2B websites are written for the business rather than the buyer. They are full of statements about how experienced and client-focused the company is, and almost entirely empty of anything useful. An AI tool cannot cite “we are passionate about delivering exceptional results.” It can cite a clear explanation of when a business needs an HR consultant, what the process looks like, and what it typically costs.

Think about what your prospects are actually trying to understand before they contact you. Write content that answers those questions directly. That content becomes citable. A brochure does not.

How to check: Read through your service pages and blog. For each page, ask: does this answer a specific question a buyer would have? If every page is a variation on “here is what we offer and why we are great”, you have a content problem that will keep you invisible to AI recommendations regardless of everything else.

Check 5: Is there a real person behind the content.

Google’s E-E-A-T framework – experience, expertise, authoritativeness, trustworthiness – has been central to how Google evaluates content quality for years. AI tools care about the same signals, because they are built in part on the same foundations.

One of the clearest signals of genuine expertise is attribution. Content written by a named person with verifiable credentials, professional history, and a visible presence elsewhere on the web carries far more weight than unattributed content that could have been written by anyone or generated by a tool.

If your website has no named authors, no team page with real people on it, and no content that demonstrates first-hand experience, you look like a generic entity. AI tools have no basis to recommend a generic entity with confidence – especially when there are clearly-documented alternatives in the same category.

How to check: Look at your blog posts and key service pages. Are they attributed to a named person? Does that person have a bio, a LinkedIn profile, and other visible signs of real expertise? Is there a team page with actual names and roles? If your site is entirely faceless, that is a credibility gap that AI tools will notice.

Check 6: Do you have schema markup.

Schema markup is structured data you embed in your HTML to tell crawlers explicitly what type of content they are looking at. Without it, a crawler has to infer everything from the text. With it, you are giving it a direct, unambiguous description of your business, your services, your people, and your location.

For AI mentions specifically, the most useful schema types are Organisation (your name, what you do, where you are, how to contact you), Service (defining individual service offerings clearly), Person (for your team and authors), and FAQ where you have clear question-and-answer content. LocalBusiness schema is worth adding if you serve a geographic area.

Schema does not guarantee you will be cited – but it removes ambiguity about who you are and what you do. On a site where the content is already solid, well-implemented schema can be the difference between being clearly understood and being a probable-but-uncertain entity that an AI tool hedges around.

How to check: Go to search.google.com/test/rich-results and run your homepage through it. If it finds nothing, you have no schema. You can also view the page source and search for application/ld+json – this is the most common format for schema markup. If it is not there, ask your developer to add it.

Check 7: Do your pages have a logical heading structure.

Heading tags – H1, H2, H3 – are not just visual formatting. They are the signposting structure that tells a crawler what each section of your page is about. A well-structured page is far easier for an AI tool to extract specific answers from, because the structure tells it exactly where one topic ends and another begins.

The correct approach is simple: one H1 per page that describes the subject of that page. H2 for the major sections. H3 for subsections within those. Do not use heading tags to make text bigger. Do not skip levels. Do not put your company name as an H1 on every page of your site.

Page builder sites almost always get this wrong. Because heading levels are used to control visual sizing rather than document structure, it is common to find pages with multiple H1 tags, no H2 tags, and H3 tags scattered randomly throughout for decorative subheadings. An AI tool reading that page gets a genuinely confused picture of what it is about.

How to check: Install the free HeadingsMap browser extension for Chrome or Firefox. Run it on your key pages. What you want to see is a clean, logical hierarchy. What most page builder sites show you is a mess of repeated H1 tags and structurally meaningless headings that tell a crawler very little.

Check 8: Are you mentioned anywhere else on the web.

AI tools do not just read your own website. They have absorbed enormous amounts of content from across the web, and that includes everything written about your business from external sources. Multiple consistent mentions of your business, your services, and your location from credible third-party sources tell an AI model that you are a real, established business with a track record.

If the only place your business exists on the web is your own website, that is a thin evidence base. Your site might be technically excellent and full of great content, but from an AI’s perspective, a business with no external footprint is harder to recommend with confidence than one that is corroborated by trade press coverage, partner pages, directory listings, client testimonials on third-party platforms, and professional association profiles.

This is why PR, trade press, directory listings on credible platforms, LinkedIn activity, guest articles, and speaking engagements genuinely matter for AI visibility. They are not vanity metrics. They are corroborating citations that build the picture an AI tool needs to recommend you without hedging.

How to check: Google your business name in quotes. Look at what comes up beyond your own site. If the first two pages are almost entirely your own domain, you have a thin external footprint. Start building it: get listed on credible industry directories, ask for LinkedIn recommendations, look for trade press opportunities, and make sure your Google Business Profile is complete and accurate.

Check 9: Do you have any real case studies or outcomes.

Case studies are one of the most valuable pieces of content a B2B website can have, specifically for AI visibility. They demonstrate real-world experience, provide specific and citable information, and give AI tools concrete evidence that you have actually done what you claim to do.

A case study does not need to be long. It needs to be specific. Who was the client (even if anonymised by sector and size)? What was the problem? What did you do? What was the outcome? Those four things give an AI tool a clear, extractable story about your expertise in action.

“We work with businesses across a range of sectors to deliver transformational results” is not citable. “We helped a 40-person manufacturing business reduce staff turnover by 30% over 18 months by restructuring their line management training programme” is very citable. One of those tells an AI tool what you actually do. The other tells it nothing.

How to check: Go to your website and find your case studies or portfolio section. If it does not exist, that is the answer. If it does exist, read through them and ask whether each one contains a specific problem, a specific approach, and a specific outcome – or whether they are vague summaries that could apply to anyone in your field.

Check 10: Is your NAP consistent across the web.

NAP stands for name, address, and phone number. It is the basic factual data about your business, and consistency matters more than most people realise.

AI tools build their picture of a business from multiple sources. If your business name appears as “Smith & Co Ltd” on your website, “Smith and Co” on your Google Business Profile, “Smith & Co. Limited” on a trade directory, and “Smith Co” on an old citation site, the AI has to decide whether these are all the same business. That uncertainty makes it less likely to cite you confidently – because confidence requires consistency.

The same applies to your address and phone number. If you have moved premises, changed your phone number, or rebranded at any point, there is a reasonable chance that old, incorrect information still exists on directories and citation sites across the web. That inconsistency undermines the picture an AI tool is trying to build.

How to check: Search for your business name across Google, Bing, Yell, Yelp, FreeIndex, and any sector-specific directories relevant to your industry. Check that your name, address, and phone number are identical everywhere. Also check your Google Business Profile directly – this is one of the most heavily weighted data sources for AI tools covering local and regional business recommendations.

These ten checks are not about tricking AI tools into mentioning you. There is no trick. AI recommendations are built on a genuinely well-documented, clearly understood, externally corroborated business presence – and that takes real work to build.

What these checks do is identify where the gaps are. Most businesses that are invisible to AI recommendations have not done one specific thing wrong. They have a site that is structurally unclear, technically noisy, content-thin, and almost entirely self-referential. Fix those things and you give AI tools something to actually work with.