When a small business is missing from AI answers, the cause is rarely something exotic. It is usually a handful of ordinary mistakes that nobody noticed, because the website looks fine in a browser. A blocked crawler here, a page that renders blank without scripts there, a homepage that never says what the company does.
By the iSuggest.ai Team · Updated for 2026
This post lists twelve of the most common ones, roughly in the order we would check them. For each we explain why it matters, how to tell if you have it, and the quickest sensible fix. None of these fixes guarantees that an assistant will name you, and we say so wherever the evidence is thin. If you are new to the topic, what GEO is gives the background.
Why small sites are especially exposed
Larger companies have teams, tooling and agencies. A small site usually has a template, a plugin or two, and an owner who has other things to do. That means defaults decide outcomes: whatever the theme, the host or the security plugin does is what AI crawlers get. Meanwhile the stakes are rising. Industry compilations report that roughly 68% of US Google queries are zero-click (Digital Applied) and that AI referral traffic grew about 16x between 2024 and 2026. Those are compilations of third-party data with their own methods, but the direction is clear: being named in an answer is becoming one of the ways people find a business.
Mistake 1: Blocking AI crawlers without meaning to
Why it hurts. If a crawler is told not to fetch your pages, it cannot read them. One report from LLM Pulse found that over 35% of the top 1,000 sites block at least one major AI crawler, and small sites often do it by accident through a copied template or a plugin setting.
How to tell. Open yourdomain.com/robots.txt and read it. Look for broad rules such as a blanket disallow, or named rules for AI user agents you did not intend.
The fix. Decide deliberately which crawlers you want to allow, and edit the file to match. Blocking can be a fair business choice; doing it by accident is the mistake. See robots.txt syntax mistakes and how to check if GPTBot can crawl your website.
Mistake 2: A firewall that blocks the crawler anyway
Why it hurts. robots.txt is only a request. A security plugin, host-level firewall or content delivery network may challenge or reject automated traffic before it reaches your page, whatever robots.txt says.
How to tell. Check your security or bot-protection settings for rules about non-browser traffic. If you can, look at server logs for requests from AI user agents and see what response they received.
The fix. Allow the crawlers you want on purpose, and confirm them by their published details rather than trusting a name in the request alone.
Mistake 3: Content that only exists after JavaScript runs
Why it hurts. Many AI crawlers read the raw HTML and do not execute scripts. A page that is an empty shell until scripts load may look blank to them.
How to tell. Fetch your page with a plain tool and read the response, or view the page source (not the inspected, live version). If your main text is missing, you have this problem.
The fix. Use server-side rendering or pre-rendering for important pages, or choose a theme or builder that outputs text in the HTML. Our guide on what client-side rendering hides shows the test.
Mistake 4: A homepage that never says what you do
Why it hurts. A homepage full of slogans such as "Elevating your experience" leaves a machine, and a human, guessing. An assistant that cannot work out what you are will not confidently recommend you for anything.
How to tell. Cover the design and read only the text. Could a stranger say in one sentence what you sell, to whom and where?
The fix. Add a plain sentence near the top: who you are, what you offer, who it is for and where you operate. Slogans can stay, but the sentence must exist.
Mistake 5: Facts hidden in images and PDFs
Why it hurts. Prices, menus, opening hours and service lists pasted as images or buried in PDFs are hard or impossible for text-reading systems to use, and they are often exactly what a user asks about.
How to tell. Try to select the text on the page. If you cannot, a machine probably cannot read it either.
The fix. Put the facts in the page body as text, using a table or list. Keep the PDF if you like, as an extra.
Mistake 6: No structured data, or structured data that lies
Why it hurts. Structured data gives machines a precise version of facts like your name, address and hours. Missing markup makes systems work harder; markup that contradicts the page can reduce trust.
How to tell. Run your pages through a markup validator and compare the output with what is on the page.
The fix. Add Organization or LocalBusiness markup, and Article, Product or FAQ markup where it fits, and make sure every value matches visible content. No markup guarantees a result, but accurate markup removes ambiguity. See structured data for AI search engines.
Mistake 7: Inconsistent business details
Why it hurts. If your name is "Green Leaf Cafe" on your site, "Greenleaf Café Ltd" on a directory and "The Green Leaf" on social media, systems may treat them as different things, splitting whatever signal you have built.
How to tell. Search for your business and list every variant of its name, address and phone number you find.
The fix. Pick one canonical version and correct every profile you control. This is the heart of entity clarity.
Mistake 8: Thin, generic pages
Why it hurts. A service page of two vague sentences gives an assistant nothing specific to quote. Systems favour passages that state concrete, checkable facts.
How to tell. Ask of each key page: what could someone quote from this that is specific to us?
The fix. Add the details only you can give: what is included, typical timelines, price ranges where you can share them, who it suits and who it does not. Write passages that make sense on their own. See how to write content AI models want to quote.
Mistake 9: No answers to real questions
Why it hurts. People ask assistants questions in natural language. If your site never states the answer to "how much does it cost?" or "do you cover my area?", there is nothing to retrieve.
How to tell. Write down the ten questions customers ask you most often, then check whether your site answers each in plain text.
The fix. Add a short FAQ and answer-first sections on key pages. Our guide to writing an FAQ section for AI search covers how, and FAQ schema covers the markup. Only mark up questions that are really on the page.
Mistake 10: Relying only on your own website
Why it hurts. What others say about you appears to carry weight. One analysis reports brand mentions correlate with AI search visibility about three times more strongly than backlinks, 0.664 versus 0.218 (Media Officers), and a Meltwater study of more than 8 million citations across eight models reports earned and news media at 37.6% of citations versus 0.2% for press releases (Meltwater). Both are single studies, but they point the same way.
How to tell. Search your name. Is anyone other than you describing what you do?
The fix. Earn independent descriptions: customer reviews, local coverage, partner pages, community answers. See brand mentions versus backlinks and earned media versus press releases.
Mistake 11: Stale content
Why it hurts. Out-of-date prices, closed services and old dates erode trust and can lead an assistant to repeat something wrong about you.
How to tell. Review your most visited pages for anything that is no longer true.
The fix. Put real dates on content, update when facts change and remove what is obsolete. Set a recurring reminder; see how often to re-audit your site.
Mistake 12: Never measuring anything
Why it hurts. Without a baseline you cannot tell whether changes help, so effort drifts toward whatever feels productive.
How to tell. If you cannot say whether any assistant has mentioned you this quarter, you are not measuring.
The fix. Ask several assistants the questions your customers ask, monthly, and record whether you are named. Answers vary between runs, so look at patterns over weeks. Watch referral traffic too; see how to measure AI search traffic in Google Analytics.
A worked example
Imagine a three-person plumbing company. This is an illustration, not a case study. Its site was built by a freelancer years ago on a popular platform. Run through the list above and a typical picture emerges.
- A security plugin challenges unfamiliar automated traffic (mistake 2).
- The price list is a photographed flyer (mistake 5).
- The homepage says "Quality you can trust" but never lists the towns it serves (mistake 4).
- The business appears under two names on two directories (mistake 7).
- No page answers "do you do emergency call-outs?" (mistake 9).
The fixes take a few hours: adjust the firewall rule, retype the prices into a table, add a sentence listing the service area, align the directory names, and add a short FAQ with honest answers. The owner also asks a few recent customers for written reviews. Nothing here is clever, and no outcome is promised. But each change removes a reason for a system to skip the company, which is the realistic goal.
How to prioritise
If you cannot do everything, work in this order, because each step is a precondition for the next.
- Access. Mistakes 1 and 2. If crawlers cannot reach you, nothing else matters.
- Readability. Mistakes 3 and 5. If the text is not in the HTML, they cannot read it.
- Clarity. Mistakes 4, 6, 7 and 8. Make it obvious what you are.
- Answers. Mistake 9. Give them something to quote.
- Reputation and upkeep. Mistakes 10, 11 and 12. These are ongoing habits, not one-off fixes.
Our AI SEO checklist turns this into 25 concrete checks, and the first 90 days plan for startups shows how to sequence the work over a quarter.
What not to do instead
Do not buy guarantees
Be wary of anyone promising that an assistant will cite you. No one controls that outcome.
Do not stuff keywords
Repeating phrases unnaturally helps neither readers nor systems. Clear writing wins.
Do not fake reviews or mentions
They break platform rules, can be detected and damage the trust you are trying to build.
Do not abandon classic SEO
Conventional search still drives much of the traffic for most small sites. GEO work should sit on top of good SEO basics, not replace them. See GEO versus SEO: ranking is not enough.
Find your own mistakes in minutes
Going through twelve checks by hand is possible but slow. The free iSuggest.ai audit examines crawler access, rendering, structured data, headings and identity signals in one pass, and shows you which of these mistakes your site actually has. You can read what an audit measures for the full list. Audits are free and not metered, so you can run one before and after each fix to see whether it did anything.
Once the basics are right, you can publish a factual snapshot of your business to the iSuggest.ai AI directory, with pages for ChatGPT, Claude and Gemini. Publishing spends credits, which are non-refundable, and improves findability without guaranteeing a citation; see how credits work. See how it works, run an audit on your homepage, then sign in at your account page when you want to publish.