A growing share of the people looking for an organization like yours will never see a list of search results. They will ask an assistant, read a paragraph, and decide from that.
That raises an uncomfortable question: when someone asks ChatGPT, Claude, or Perplexity about your organization, what do they hear back? For many sites the answer is vague, outdated, about a different company with a similar name, or nothing at all.
Two things are worth saying before the checks.
First, nobody controls what these assistants say. They do not publish how they choose sources, their answers change from day to day, and anyone promising a guaranteed outcome is guessing with confidence. Be wary of that pitch.
Second, you do control whether your website is legible to the systems that feed them. That part is concrete, checkable, and mostly the same good practice that has always made sites easy to find. Here are the five checks we run first.
1. Can the Crawlers Get In?
Assistants learn about websites through crawlers, the automated visitors that read pages. Some crawl to train models. Some fetch pages live when a user asks a question. If they are blocked, your site does not exist for that assistant.
Blocks happen in two places:
- Your robots.txt file. Visit yourdomain.com/robots.txt. Look for rules that disallow everything, or that name AI crawlers specifically. Some sites added blanket blocks during the early debates about AI training and never revisited them.
- Your firewall or CDN. Security services increasingly block automated traffic by default, including crawlers you may want. This will not show up in robots.txt, so ask whoever manages your hosting.
Blocking some crawlers can be a legitimate decision. It should be a decision, though, not a leftover.
These rules also have a way of outliving their reasons. This year we found a single line in our own robots.txt, written years earlier for a sensible purpose, that was quietly stopping Facebook from loading the images on our blog posts. Nobody had looked at the file in a long time. That is typical.
2. Does Your Content Exist Without JavaScript?
Many modern websites build their pages in the visitor's browser using JavaScript. A person sees a complete page. A crawler that does not run JavaScript, and many do not, sees an empty shell.
The quick test: open a key page, right-click, and choose "View Page Source." This shows what the server actually sent. Search it for a sentence from the page, such as your description of what you do. If the sentence is there, crawlers can read it. If the source is mostly script tags and the text is missing, your content is invisible to anything that does not render the page like a browser.
This matters most for your homepage, your service or program pages, and your about page, which are exactly the pages an assistant would use to describe you.
3. Is Your Structured Data in Place?
Structured data is a layer of labeled information in your page code that tells machines what the page is about: this is an organization, here is its name, address, phone number, and logo; this is an event, with this date and location; these are frequently asked questions and their answers.
Search engines have used it for years. It gives assistants something unambiguous to work with instead of inferring facts from your prose.
To check, run a key page through a schema validator, such as Google's Rich Results Test or the Schema Markup Validator. Look for, at minimum:
- Organization (or a more specific type like LocalBusiness) on the homepage, with your name, URL, logo, and contact details.
- Types that match your content: events for events, articles for articles, FAQ markup for real FAQ sections.
- No errors. Broken structured data is sometimes worse than none, because it asserts something wrong.
Some sites also publish an llms.txt file, a plain-text summary of the site written for language models. It is a young convention and not every assistant reads it, but it is cheap to add and does no harm.
4. Does Your Site Say Plainly Who You Are?
Assistants summarize. If your own site does not state clearly who you are, what you do, who you serve, and where you operate, the summary will be vague or wrong.
Read your homepage and about page as a stranger would. Look for:
- A plain statement of what you do in the first few sentences, not a tagline that could belong to anyone.
- Your full name used consistently. If you are known by an acronym, state the full name and the acronym together. Associations with common acronyms are especially prone to being confused with other organizations.
- Specifics. Location, the members or clients you serve, your services or programs, named clearly.
- Consistency elsewhere. Your name, address, and description on directories, profiles, and listings should match your site. Conflicting information gives assistants conflicting answers.
This is the same clarity that helps a human visitor decide in ten seconds whether they are in the right place.
5. What Do the Assistants Actually Say?
The last check is the most direct: ask them.
Open ChatGPT, Claude, and Perplexity, and ask each the question a prospective member, customer, or donor would ask. "What does [your organization] do?" "Who provides [your service] in [your region]?" Read the answers in full.
Look for three things: whether the assistant knows your organization, whether it cites your website, and whether it describes what you actually do. It is common for one assistant to know you and another not to. Some search the live web as they answer and others rely mostly on training data, so a split result is normal rather than a contradiction.
Treat the result as a snapshot. Ask again next week and the wording will change, occasionally the verdict too. A poor result is a signal worth investigating, not a score to track daily.
Where to Start
If you run only one check this week, make it the first. A blocked crawler undoes every other improvement, and it takes five minutes to rule out.
After that, work in order. Content that exists without JavaScript and accurate structured data are foundational, and clear, consistent self-description is the part only you can write. The fifth check tells you whether the rest is working.
A Simple Rule of Thumb
You cannot control what an assistant says about you. You can control whether your site gives it anything accurate to say.
Focus on what you control, check the outcome occasionally, and ignore anyone selling certainty.
Final Thoughts
None of these checks require new technology or a new website. Most are the fundamentals of a well-built site, pointed at a new kind of reader. The organizations that show up well in assistant answers are rarely doing anything exotic. Their sites are open to crawlers, readable without scripts, clearly labeled, and plain about who they are.
If you would rather not run these by hand, our free AI Readiness audit checks crawler access, structured data, and on-page signals, and our AI Visibility check shows you what the assistants say about you, word for word.