For one week in October, ChatGPT visits quietly vanished from website reports and showed up as "direct" traffic. Here is what that glitch teaches every small business owner about tracking AI search, and the seven checks I would set up this week.
By Warren Schuitema, AI Systems Consultant, Matchless Marketing / Matchless AI
Published: October 18, 2026 | Last Updated: October 2026
Last week, a lot of business owners opened their website reports and saw something strange.
Visits from ChatGPT dropped off a cliff. At the same time, "direct" visits jumped. Same pages. Same week. Nobody had changed a thing.
The customers never left. The label fell off.
So how do you know if ChatGPT and Google AI are sending customers to your small business? You can't trust a single number in a single report. You need three simple signals working together: a clean "AI sources" view in your analytics, a habit of asking new customers how they found you, and a monthly check where you ask the AI tools yourself. Set those up once, and a glitch like last week's becomes a footnote instead of a panic.
Here is what happened, what the research says, and how to build a scoreboard you can trust.
Key Takeaways
- From about October 8 to October 15, 2026, links from ChatGPT stopped carrying the tag that tells your analytics where a visit came from. Many of those visits got filed as "direct."
- One Adobe analyst tracked about 500 pages and found ChatGPT visits fell by about 31,000 a day while direct visits rose by about 26,000 a day on the same pages.
- A February 2026 BrightLocal survey found 45 percent of consumers now use ChatGPT or other AI tools for local business recommendations, up from 6 percent the year before.
- AI answers change from one search to the next, so one check tells you very little. Track the trend, not the single reading.
- The fix takes about an hour: build an AI sources view, add one question to your intake form, and run a short monthly prompt check.
Your Website Report Got the Story Wrong for a Week
Here is what broke.
When someone clicks a link inside ChatGPT, that link usually carries a small tag at the end of the web address. It reads utm_source=chatgpt.com. That tag is how Google Analytics and similar tools know the visit came from ChatGPT. OpenAI started adding it in June, according to Search Engine Roundtable.
Then OpenAI released GPT-6 on October 7 for paid plans and October 8 for free users. Within a day or two, the tag disappeared.
Cedric Huesler, who works at Adobe, spotted the cause. He wrote on LinkedIn that the new version opened links in a way that dropped both the tag and the referrer, the other clue analytics tools use. "There's no tag and no referrer, so your analytics files the visit under direct," he wrote.
He compared October 9 to 13 with the week before, across about 500 pages where ChatGPT drove at least 5 percent of visits. His numbers: ChatGPT visits down about 31,000 a day. Direct visits up about 26,000 a day on the same pages.
On October 15, Huesler posted an update: "Relieved. It was a bug. ChatGPT is tagging its links with utm_source=chatgpt.com again." OpenAI has not made a public statement about it.
So the problem is fixed. Why write a whole article about it?
Because most small business owners had no idea it happened. If you checked your report that week, you might have decided ChatGPT stopped sending you anyone. You might have dropped a project that was working. Or you might never have known AI was sending you customers in the first place, because those visits were already hiding in "direct."
I spent 16 years in manufacturing and supply chain. One lesson from those years has stuck with me. A bad gauge is worse than no gauge. With no gauge, you know you're guessing. With a bad gauge, you make confident decisions on wrong numbers.
Last week, a lot of businesses had a bad gauge and didn't know it.
Customers Already Ask AI Who to Hire
This matters more every month, because the share of customers who ask an AI for help is no longer small.
Nearly half of consumers use AI for local recommendations. BrightLocal's Local Consumer Review Survey 2026, published in February, polled 1,002 U.S. adults. It found that use of ChatGPT and other AI tools for local recommendations rose "from 6% last year to 45%." That is a giant jump in one year. Think plumbers, dentists, bakeries, tutors, and photographers.
Google keeps adding links inside its AI answers. On October 15, 2026, Google's Hema Budaraju announced "continued and substantial increases to the presence of inline links" in AI Mode and AI Overviews. Google says it will show those links higher and more often. It is also testing a section at the top of some answers that points to the one or two official websites for a task. More links means more chances for a customer to click through to you, and more visits you'll want to count correctly.
Bing now shows site owners where they get cited. On October 13, Search Engine Land reported a new Topics Portfolio view inside the AI Performance report in Bing Webmaster Tools. It groups the searches that cite your site into topics. It also lists "Top Opportunity Queries," where competitors get cited and you don't. The catch: it shows citations, not clicks, and smaller sites may see few topics or none.
The big platforms are building for a world where customers ask an AI first. Your reports need to keep up.
AI Visitors Are Worth Counting, and Easy to Miss
Here is the research that tells me this is worth an hour of your time.
AI visitors tend to buy. Adobe tracks traffic to U.S. retail websites. According to Digital Commerce 360's coverage of Adobe's July 2026 data, AI referral traffic to retail sites rose 62 percent from a year earlier. Adobe also found that AI-referred visits convert at a rate 60 percent higher than other traffic, and that those shoppers bring in 53 percent more revenue per visit. Those numbers come from retail, so a service business will see different results. But the direction is clear. People who arrive from an AI answer often show up ready to act.
AI answers change a lot from one try to the next. In a study published in September 2026, BrightLocal ran more than 200,000 local, non-branded prompts across ChatGPT, Google AI Mode, and AI Overviews for over 1,300 business locations. When the same search ran again, the list of businesses overlapped only 20 to 33 percent on average. Google Maps mentioned a tracked business in 66 percent of searches. The AI tools mentioned it in 32 to 38 percent. And a business that showed up once appeared again in only about half of later tries.
That second finding is why one quick check tells you almost nothing. You could ask ChatGPT about your business on Monday, see your name, and feel great. Ask again Tuesday and you might be gone.
Getting named and getting linked are two different things. Semrush released its 2026 AI Visibility Index on October 13, built from millions of prompts per month across ChatGPT, Gemini, and Google AI Mode. In the UK data, the overlap between the 100 most-mentioned brands and the 100 most-cited websites ran from about 37 to 45 out of 100, depending on the month. In plain words, an AI can recommend your business by name and still send the click somewhere else, like a review site or a directory.
ChatGPT is leaning on original sources. Search Engine Journal reported on October 12 that, according to Ahrefs data, ChatGPT's most-cited sites shifted toward primary sources between March and September 2026. Reddit's share of cited U.S. answers fell from 19.1 percent to 2.3 percent. Wikipedia fell from 14.1 percent to 2.9 percent. Ahrefs notes that citation rank is not the same as traffic, but the trend favors sites that publish their own real information.
Put those together. Customers are asking AI. AI visitors often arrive ready to buy. And the tracking is shaky, the answers shift, and the label can fall off without warning.
Fix the Gauge Before You Trust the Reading
Most advice about AI search jumps straight to "how do I show up more?" I think that skips a step.
In a factory, before you change a process, you check the measuring tool. If the gauge reads wrong, every fix you make after that is built on bad information. You might "fix" something that was never broken.
I call this the Gauge Check. Before you spend a dollar or an hour trying to win more AI traffic, make sure you can see the AI traffic you already have.
The Gauge Check asks three questions:
- Can my analytics tell AI visits apart from everything else? If all your AI visits sit inside "referral" or "direct," you can't see the trend.
- Do I have a second signal that doesn't depend on a tracking tag? Asking customers how they found you still works when a tag breaks.
- Am I reading the trend, or one data point? AI answers move around too much to judge on a single day.
If the answer to all three is yes, a bug like last week's barely matters. You'd see the AI number dip, see "direct" rise by about the same amount, and see your "how did you hear about us" answers hold steady. Three signals. One clear story.
That is the shift I want small business owners to make. Stop asking "Is AI sending me customers?" once. Start keeping score in a way that survives a glitch.
How I Keep Score on My Own AI Systems
I hold my own systems to this same rule.
I run my business day to day with a workforce of AI agents. Each one has a clear job, and each one writes a short recap at the end of every work session. Ten lines at most. If nothing happened, the recap says so. Nothing sends, posts, or spends money without my approval.
Why the recaps? Because manufacturing taught me that a system you can't see is a system you can't trust. The recaps are my gauge. When something looks off, I don't guess. I go read what happened.
Your website traffic deserves the same treatment. A few honest signals, checked on a regular rhythm, will do the job. Skip the fancy dashboard.
I test every system I build at home the same way too. My Weekly Meal Planner replaced HelloFresh in our house, and it earned that spot because it ended the Sunday planning spiral. My Digital Wall Calendar replaced a $50 a month subscription and ended the nightly schedule conversation. I know those systems work because I can see the results every week.
If you want a deeper look at keeping score on an AI tool once it's running, read my guide on how to know if an AI customer service agent is actually working. Same idea, different tool.
What Does This Look Like for a Local Service Business?
Here is a made-up example, just to show how it plays out.
Say you run a small house-cleaning business with a home office and two part-time cleaners. You check Google Analytics once a month.
In September, you see about 40 visits a month from chatgpt.com. Nice. In mid-October, you check again and ChatGPT shows almost nothing. Your stomach drops. Did something break? Should you pay someone to "fix your AI ranking"?
Now picture the same owner with the Gauge Check in place.
You open your AI sources view and see the ChatGPT dip. Then you look at "direct" for the same week and see it rose by about the same amount. Next you check your booking form, where you added "ChatGPT or another AI tool" as an answer to "How did you hear about us?" Three new clients picked that answer in October, same as September.
Story solved in five minutes. Nothing broke. The label fell off for a week, and your customers kept coming.
That is the difference between a gut reaction and a calm decision. Same data. Better gauge.
If you sell products online, the stakes are even higher. My post on whether AI shopping agents will send customers to your small business or straight to Amazon covers what to fix on your product pages so AI can find you in the first place.
7 Steps to Track AI Search Traffic This Week
Most of these take less than 15 minutes each, and every tool here has a free option.
- Build an "AI sources" view in your analytics. In Google Analytics 4, create a custom channel group or a saved report that pulls visits whose source includes chatgpt, perplexity, gemini, copilot, or claude. Now AI visits sit in their own row instead of hiding inside "referral." Check it once a month.
- Write down the October glitch. Add a note to your reports, or a line in a simple log, that says: "October 8 to 15, 2026: ChatGPT links lost their tracking tag. Some AI visits counted as direct." Six months from now, when you compare months, you'll know why October looks odd.
- Ask every new customer how they found you. Add one question to your booking form, intake form, or first phone call: "How did you hear about us?" Include "ChatGPT or another AI tool" as an option. This signal doesn't break when a tag disappears.
- Run a monthly prompt check, three times over. Write down five questions a real customer would ask, like "best house cleaner near downtown Phoenix for a small home." Ask each one in ChatGPT, Gemini, and Google's AI Mode. Run the set three times and record how often you appear. Because AI answers shift so much, the count across three tries tells you more than any single answer.
- Open Bing Webmaster Tools. Bing Webmaster Tools is free. Add your site, then look at the AI Performance report and the new Topics Portfolio view. If you see "Top Opportunity Queries," those are questions where competitors get cited and you don't. That's your content to-do list.
- Compare AI visits with "direct" every month. Put the two numbers side by side. If AI drops and direct rises by about the same amount, suspect a tracking problem before you suspect a customer problem. That one habit would have saved a lot of worry last week.
- Give AI something worth citing. The research shows ChatGPT leaning toward original sources. So publish your own real information in plain text on your site: your services, your prices or price ranges, your service area, your hours, and answers to the questions customers ask you every week. You are the primary source on your own business. Act like it.
Do steps 1, 3, and 4 this week. Add the rest by the end of the month.
Frequently Asked Questions
How do I see ChatGPT traffic in Google Analytics?
In Google Analytics 4, open your traffic acquisition report and look for the source chatgpt.com, usually under referral. For a cleaner view, create a custom channel group that collects AI sources like ChatGPT, Perplexity, Gemini, Copilot, and Claude into one row. Then check that row monthly and compare it with direct traffic.
Why did my ChatGPT traffic suddenly drop in October 2026?
From about October 8 to October 15, 2026, ChatGPT links stopped carrying the utm_source=chatgpt.com tag after the GPT-6 release. Many analytics tools then counted those visits as direct. Analysts reported the tags returning on October 15. If direct traffic rose while ChatGPT fell, that bug is the likely cause.
Can I trust what ChatGPT says about my business when I search for it myself?
Only as one data point. BrightLocal found that when the same local search runs again, the list of businesses overlaps only 20 to 33 percent on average. Run the same questions several times, across ChatGPT, Gemini, and Google AI Mode, and track how often you appear over months.
Do I need to pay for an AI visibility tool?
Not to start. Google Analytics, Bing Webmaster Tools, and a simple spreadsheet cover the basics for a small business. Paid tools from companies like Semrush and BrightLocal track AI mentions at scale, which helps once you have many locations or compete in a crowded market. Start free, then upgrade if you outgrow it.
What kind of website content gets cited by AI search tools?
AI tools favor clear, original, factual information. Ahrefs data reported by Search Engine Journal shows ChatGPT citing more primary sources in 2026. For a small business, that means plain-text pages listing your services, prices or ranges, service area, hours, and honest answers to the questions customers ask you most.
Count What Counts
Back to that strange week in October.
The customers never left. The label fell off. And the owners who noticed were the ones with more than one way to see the truth.
So run your tracking through the same test I use on every system I build.
Time. Can you answer "Is AI sending me customers?" in five minutes, without guessing? An AI sources view and a monthly prompt check get you there.
Money. Are you about to spend on a fix, or cut a project, based on one number from one week? Check the gauge first. It might save you from paying to repair something that was never broken.
Energy. Does a weird dip in your report send you spiraling? Three signals that agree with each other let you shrug, write a note, and move on.
AI search is going to keep changing. Tags will break again. Answers will shift. That's fine. Aim for a gauge you trust and the habit of reading it.
Fix the gauge, then read the trend. That's A Better Way.
Keep Learning
- How Do You Know If an AI Customer Service Agent Is Actually Working? How to keep score on an AI tool after it goes live.
- Will AI Shopping Agents Send Customers to Your Small Business, or Straight to Amazon? What to fix on your product pages so AI shopping tools can find you.
- How Do You Protect Your Small Business From AI-Powered Scams in 2026? The scams that come with new AI tools, and how to stop them.
- What Does Google Replacing Gemini Gems With Skills Mean for Your Small Business? Another Google AI change that touches how small businesses work.
About the Author
Warren Schuitema is an AI Systems Consultant with Matchless Marketing and Matchless AI. He spent 16 years in manufacturing and supply chain, where he learned to check the gauge before trusting the reading. He runs his business day to day with a workforce of AI agents that each report back in a short recap, and none of them can send, post, or spend without his approval. At home, his Digital Wall Calendar replaced a $50 a month subscription and his Weekly Meal Planner replaced HelloFresh. He helps small business owners and parent entrepreneurs build practical AI systems that give back Time, Money, or Energy, and he is a partner in the Goal Achievers Summit.