How Customers Actually Find Businesses in 2026
- AI engines now decide who gets found, not just Google rankings.
- ChatGPT, Claude, and Perplexity fetch your pages live when someone asks a question.
- You can see exactly which AI crawlers visit, what they read, and whether they send clicks.
- The data is free to capture with a 14-day trial, or start with a free one-off visibility check.
- If you aren't measuring AI discovery, you're guessing.
Most people find a business in 2026 by asking an AI assistant and taking the first answer it gives. That answer is pulled from a small set of sources the AI trusts, and if your site isn't among them, you don't get the visit. This is how customers find businesses now, and it runs alongside search engines, not instead of them. SEO still matters, but it's no longer the only gatekeeper.
The old map is gone
You used to know the game. Rank well on Google, get the clicks, answer the phone. That path still works, but it's narrower. Every year, more product searches, local questions, and "best X for Y" queries start in a chat box, not a search bar. The AI composes an answer, cites a few sources, and the customer never scrolls through a results page.
That changes what you should measure. Clicks and rankings on Google are lagging indicators. AI visibility is a leading one. If ChatGPT or Perplexity routinely cites your competitor's page on the questions your customers ask, you're losing discovery before the click ever happens.
What AI engines actually do with your site
AI assistants don't browse the web like people do. They send automated crawlers to read pages, and they do it in three distinct ways. Kymo classifies them into a bot taxonomy so you can tell them apart:
| Crawler class | What it does | Example user agents |
|---|---|---|
ai_answer | Live fetch made for one person's question, in the moment | ChatGPT-User, Claude-User, Perplexity-User |
indexing | Builds a search index the AI draws from later | OAI-SearchBot, Claude-SearchBot, PerplexityBot |
training | Bulk crawl for model training data | GPTBot, ClaudeBot |
Each one matters for a different reason. A live ai_answer fetch means someone asked a question and your page got pulled in to answer it. That's the closest thing to a qualified lead you'll ever see from a machine. An indexing crawler visit means the AI engine is building a profile of your site for future answers. A training crawl is less actionable but still tells you your content is being read at scale.
Take OAI-SearchBot (OpenAI's search index crawler) as an example. When it visits, OpenAI is storing your content for its search and answer features. That's the crawl that leads to citations. If it never shows up, ChatGPT can't cite you, no matter how good your content is.
The pattern that kills small business traffic
The frustrating part is that AI engines don't tell you when they've read you. You have to look at server logs or install something that catches the requests. Most small business owners don't do either, because until recently there was no easy way.
So the typical story goes like this. Your Google traffic holds steady for a while, then slowly erodes. Your competitor launches a PDF guide answering the exact question your customers ask. ChatGPT starts citing it. You don't know because you only check Google Analytics, which shows AI bots as noise or filters them out entirely.
Kymo shows the whole picture in one dashboard. Human pageviews and unique visitors on one side. AI crawler activity on the other. You see the exact pages GPTBot (OpenAI's training crawler) reads, which pages the ai_answer crawlers fetch live, and how many confirmed click-throughs come back from AI referrals.
That last number matters. It's the difference between "AI knows I exist" and "AI sent me a customer." Some AI engines send referral traffic directly to your site when they cite you. Kymo's AI visibility, explained page walks through how that shows up in your reports.
What to look at first
You don't need an analyst to read this. Start with the AI Visibility report, which shows you three things:
- Which of your pages AI assistants fetch most often.
- The likely prompts behind those fetches.
- Confirmed click-throughs from AI referrals.
The prompts are the useful part. If you see your pricing page being fetched for "how much does a plumber cost in Austin" but your blog post on maintenance tips is being ignored, you know where to focus. Write more answers to the questions people actually type into ChatGPt.
The Reading the AI report guide covers how to interpret each section without drowning in data. It's a 15 minute read and it's the difference between "I installed analytics" and "I understand my traffic."
Why normal analytics tools miss this
Google Analytics never told you about AI bots because it wasn't built for them. Traditional analytics track humans. Cookies, sessions, user agents that look like browsers. AI crawlers don't behave that way, so they get bucketed as "bots" and filtered out, or they pollute your numbers.
Kymo does something different. It uses cookieless visitor counting, where the visitor identifier is a salted server-side hash that rotates at UTC midnight. Raw IPs are never stored. No human visitor's user-agent string is stored either. But AI crawler user agents are kept on the crawler's own event row, so the classification stays auditable. A crawler is not a person, and treating it like one was the old mistake.
The practical side
Pricing is simple. Solo is $9 a month, or $90 a year, for up to 10,000 events a month. Studio is $29 a month, or $290 a year, for up to 100,000 events a month. Both plans include every feature, and AI crawler tracking does not count against your event limit. If you exceed the limit, the dashboard pauses but your data is not deleted. Both plans start with 14 days free, no card required.
You also get standard analytics on top of AI tracking. Pageviews, unique visitors, referrers, top pages, countries, devices, browsers, sessions, bounce rate, and session duration. There's a real-time map of active visitors, outbound click tracking, goal tracking with revenue on a goal, and UTM campaign tracking. If you're migrating from GA4, you'll find everything you actually used and nothing you didn't.
Make this decision cheaply
If you're still not sure whether AI engines can even see you, run the free AI visibility check. It reads a URL from public signals only, so your site doesn't need Kymo installed to test it. The report arrives by email. It's the fastest way to find out if ChatGPT has ever cited you.
For a deeper look at what's worth paying for, the AI SEO Tools for Small Business: What's Worth Paying For post covers how to separate useful tools from hype. And if you're wondering whether your bounce rate is actually an issue, Bounce Rate for Small Business Sites: What's Normal gives you the baseline. The broader picture of how discovery is shifting is in the AI search for small business category.
This is all part of a bigger shift. Search is being replaced by answers, and that shift is documented in The Small Business Guide to AEO. Read it once and you'll understand why being "answerable" matters more than being "rankable."
The recommendation
Start with the free check if you're early. See what AI engines already know about you. Then sign up for Kymo if you're serious about tracking this continuously, because a one-off audit is a snapshot and these systems change weekly. Do the free check if you're curious, and start tracking it if you're competing.
What you get when you start
Run the free AI visibility check on your own site to see which AI engines already crawl you, or start tracking continuously to watch the crawlers on a live dashboard. The free check works on any URL, no Kymo install needed, and the report lands in your inbox. The continuous tracking shows you every AI crawler visit, the pages they read, and the clicks they send back, alongside your normal human analytics.