How to get recommended by ChatGPT, Google AI and Perplexity

There's a new way customers find you, and it isn't Google's ten blue links. It's the answer ChatGPT, Google's AI or Perplexity gives when someone asks "who's a good electrician near me". Here's how to actually get recommended, minus the snake oil.

First, the honest framing. AI search is a real, growing channel, and it's still young enough that most of the advice about it is either guesswork or someone selling a shortcut. This guide is the version without the hype. What genuinely moves the needle, what doesn't, and the groundwork worth doing whatever the tools do next.

The businesses that win AI search aren't the loudest. They're the easiest for a machine to understand and trust.

They're three different machines

The biggest mistake is treating ChatGPT, Perplexity and Google's AI as one thing. They source their answers in genuinely different ways, and a page one of them loves, another ignores.

ChatGPT leans heavily on encyclopedic sources and on what other websites say about you. Being mentioned elsewhere, in reviews, directories and local write-ups, can matter as much as your own site does.

Perplexity runs a live web search for every single question, so it rewards fresh, clearly-dated content. New pages can be quoted within hours of being indexed, and content published recently gets cited far more often than an undated page from three years ago.

Google AI Overviews sit on top of the normal Google index, so good SEO is the entry ticket. But the pages it actually quotes are the ones that answer the question in the first line, back it with real figures, and carry clean structured data.

The takeaway isn't "do three different things". It's that all three reward the same underlying qualities: be clear, be current, be trustworthy, be easy for a machine to read.

What actually gets you recommended

Answer the question first

Machines quote the sentence that answers the question, not the one that clears its throat first. If a page is about EICR costs, the opening line should say what an EICR costs. Lead with the answer, then explain. It reads better for humans too.

Clean structured data

Schema markup is code that tells a machine exactly what you are, with no guessing. For a local or service business, the types that earn their keep are Organization, LocalBusiness, Service, FAQPage and BreadcrumbList. Most small-business sites have none of it, or have it broken. Getting it right removes the ambiguity that makes an AI skip you.

Date your work

Perplexity in particular favours fresh content, and a visible "updated 2026" with a real date beats an undated wall of text. It's a small thing that quietly tells every engine your information is current.

Be talked about elsewhere

Especially for ChatGPT, what other sites say about you feeds the picture. Real Google reviews, a listing in the right local directories, a mention in the local paper: all of it builds the machine's sense that you exist and can be trusted. Your own website can't do this bit alone.

Let the right robots in

AI engines send crawlers to read the web. If your site quietly blocks them, and plenty do, either by default or because a nervous developer added a rule, you're invisible to AI before you've started. The fix is a robots.txt file that welcomes the AI search crawlers. More on exactly which ones below.

Which robots to let in

The big AI companies each run named crawlers. The ones worth knowing: OpenAI runs GPTBot, OAI-SearchBot and ChatGPT-User; Anthropic runs ClaudeBot and Claude-SearchBot; Perplexity runs PerplexityBot; Google runs Google-Extended.

There's a useful distinction hiding in that list. The search crawlers (OAI-SearchBot, Claude-SearchBot, PerplexityBot) are the ones that let you show up in AI answers. The training crawlers (GPTBot, ClaudeBot, Google-Extended) are the ones that use your content to train the models themselves. You can allow the first group so you get recommended, and separately decide whether you're comfortable with the second. Blocking training doesn't stop you being cited.

One honest caveat: the rules only work with crawlers that respect them. Some, notably Bytespider, are known to ignore robots.txt entirely. So the file is guidance, not a lock, and if you genuinely need to keep a bot out, that takes server-level blocking, not a polite request.

llms.txt, the honest take

You'll hear a lot about llms.txt. It's a simple text file that sits at the root of your site and tells AI tools, in plain language, what your site is and what matters on it. Think of it as a friendly summary written for machines.

Here's the part the people selling it leave out. As of 2026, no major AI company has publicly committed to using llms.txt in their live systems, with one exception: Anthropic has confirmed its Claude tools read it. Independent analysis of hundreds of thousands of sites found no clear link between having an llms.txt file and getting cited more often.

Anyone selling llms.txt as the secret to AI search is selling you something.

So why ship one at all? Because it's free, it can't hurt, Claude uses it today, and the direction of travel is toward more of this, not less. It's cheap insurance, not a magic bullet. That distinction is the whole point: do the cheap sensible things, and don't pay a premium for a shortcut nobody can prove works.

The "near me" search is changing shape

For years the goal was to rank for "electrician near me". That search is quietly turning into a question asked to an AI: "who's the best electrician in Basingstoke, with good reviews?" Same intent, new front door.

The good news is that the things that win the old search also feed the new one. Your Google Business Profile and your real reviews do double duty: they power the local map pack and they shape the AI's sense of who's trustworthy. Keep your name, address and phone number identical everywhere they appear, because machines trust consistency and get confused by contradictions. Get more reviews, reply to them, keep the profile current. It's the single highest-payoff thing a local business can do, for normal search and AI search alike.

How to check where you stand right now

You don't need a tool for the first check. Open ChatGPT, Perplexity and Google, and ask each one "who's a good [your trade] in [your town]?" See whether you appear, and take note of who does. That's your competition in this channel, and the gap you're closing.

If you'd rather be walked through it properly, we built a free AI Visibility Test that checks whether the main AI tools can find you, step by step, with no signup.

How Zenlio builds for this

Full disclosure: this is our job, so I'm biased. But here's the honest version of what goes into every site we build, because it's the same list above, done properly.

Clean schema on every page. Copy that answers the question in the first line. Content that's dated and kept current. Real Google reviews pulled in and marked up. A robots.txt that welcomes the AI search crawlers instead of accidentally blocking them. And llms.txt shipped as standard, on the honest understanding that it's cheap insurance rather than a miracle. If you want to see it in the wild, our own site and our live client sites are all built this way; go and ask an AI about them.

What we won't do is promise you'll be ChatGPT's top pick by Friday. Nobody can honestly promise that. This is groundwork, the kind that makes you findable and understandable, so that when a machine is deciding who to recommend, you're actually in the running.

AI search rewards the same thing good businesses have always wanted: to be clear about what you do, easy to trust, and simple to find. The tools changed. The job didn't.