What I'd Do for AI Visibility If I Only Had One Hour

Most advice on AI visibility assumes you have a quarter, a content team and a budget. Most people have an hour.

If that were me, starting from zero, this is exactly what I would do. Five steps, roughly 10 to 15 minutes each, in order. Each one builds on the last.

1. Make sure AI agents and crawlers can read your site

Nothing else matters if AI systems cannot access your content. So I start here.

There are two things to check:

robots.txt. Open yourdomain.com/robots.txt and look for rules that block AI crawlers such as GPTBot, OAI-SearchBot, ChatGPT-User, ClaudeBot, PerplexityBot or Google-Extended. Many sites block them by accident through a blanket rule or a CMS default. If you want to be cited, the search and user-triggered bots need access. Our list of AI search crawlers and their user agents covers which is which.
Your CDN and firewall. Cloudflare, Akamai and similar tools can block bots even when robots.txt allows them. Check that AI crawlers are not being challenged or returned a 403.

Then check how your pages render. Most AI crawlers do not run JavaScript well. If your key content only appears after client-side rendering, they see an empty page. View the page source, or fetch it with curl, and make sure the actual text is in the HTML.

If you want to confirm what is really happening, AI crawler analytics shows which bots hit your site and which pages they fetch.

2. Check your visibility in ChatGPT

Next I want a baseline. How often does ChatGPT mention my brand today?

You can do this by hand, but a tool like Airefs does it properly. Add your brand, add a set of prompts your customers would actually ask (category questions such as “best CRM for small agencies”, comparison questions, “alternatives to” questions) and run them in ChatGPT.

What I look at:

Mention rate. In what share of answers does my brand appear?
Position. Am I listed first, or buried at the end?
Competitors. Who shows up instead of me?

AI answers vary from run to run, so a single manual check is noisy. That is why I prefer tracking prompts repeatedly and looking at the rate rather than one answer. If you want the manual route first, how to track visibility in ChatGPT walks through it.

You are not trying to fix anything yet. You are getting a number to improve.

3. Look at which sources ChatGPT cites

This is the step that tells you what to do. ChatGPT does not make up its recommendations from nowhere. It leans on specific pages, and those pages are visible.

In Airefs, I open the Sources view for the prompts I just tracked and look at:

Top domains. Which websites get cited most often for my prompts?
Top URLs. Which exact pages?
Citation type. Are they listicles, review sites, comparison pages, forums, or competitors’ own blogs?

Then I learn from it. If the top sources are “best X tools” listicles, I need to be in those lists. If they are review sites, I need reviews there. If a competitor’s comparison page dominates, I need a better one. The sources show you the shape of the content that wins in your category.

Within those top sources, I look for discussions: Reddit threads, LinkedIn posts, Quora questions, forum posts.

These are the easiest wins in the whole hour. A listicle on a third-party site takes weeks of outreach to get into. A Reddit thread that ChatGPT already cites can be commented on today.

For each thread I check:

Is it recent enough that a comment will be read?
Is the question relevant to what my brand does?
Can I add something useful, not just a pitch?

Then I write a genuine comment that answers the question and mentions my brand where it fits, with context on why it is a good option. Be honest about who you are. Comments that read as ads get downvoted or removed, and they do not help. The Reddit marketing guide for AI search goes deeper on tone and what works.

Airefs shows the discussion threads that are already cited in AI answers, so you can skip the searching and go straight to commenting.

5. Create an AI instructions page

The last step is the one most brands skip. I create a page on my own site written for AI, not for humans.

An AI instructions page is a plain, raw page with no design, no scripts and no marketing copy. Just facts and statements about your brand, written so a model can lift them directly:

What the company is and does, in one or two sentences
Who it is for
Key features and pricing
How it compares to alternatives
The exact way you want the brand described
Links to your most important pages

Keep it factual, specific and consistent with the rest of your site. Publish it at a clear URL such as /ai-instructions/ and link to it from your footer so crawlers can find it. You can see how we did it on our own AI instructions page.

A word of caution. This is not a magic switch. The evidence that files like llms.txt change what models recommend is weak, as we covered in llms.txt does not influence LLM recommendations. I treat the AI instructions page as a clean, canonical source of truth. It gives crawlers and retrieval systems accurate facts about you, and it costs almost nothing. It will not replace steps 3 and 4.

The order matters

Crawlability comes first because it is the gate. The baseline and source analysis come next because they tell you where to spend effort. Social threads follow because they are the fastest way to act on what you learned. The instructions page closes the loop by giving AI a clear description of your brand.

An hour will not make you the top recommendation in your category. But it will tell you where you stand, why, and what to do next, which is more than most brands know today.