AI-driven website traffic is becoming an important part of digital marketing, but measuring its impact is not always straightforward. Recent reports of missing ChatGPT tracking parameters highlight why businesses need a broader approach to analytics, attribution and SEO performance.
Why AI traffic measurement deserves more attention
The way people discover businesses online is changing. Traditional search engines remain important, but customers can also use AI assistants to research products, compare services, find recommendations and explore potential suppliers.
For businesses, this creates new opportunities to reach prospective customers. For marketers, it introduces a measurement challenge: how do we accurately assess the contribution of AI platforms to website traffic and conversions?
Recent reports concerning ChatGPT have brought this question into focus.
In October 2026, marketers began reporting a sharp decline in website visits attributed to ChatGPT. One explanation was that some links were no longer carrying the tracking information previously used to identify visits from the platform.
The issue was highlighted in reporting by Search Engine Land, drawing on analysis shared by Cedric Huesler of Adobe.
The findings raised an important question for anyone responsible for digital marketing: could some businesses be underestimating their AI-driven traffic because of how their analytics systems classify website visits?
The situation is a useful reminder that analytics data must be interpreted carefully. A change in reported traffic does not always mean customer behaviour has changed.
Understanding the role of tracking parameters
To understand the issue, it helps to look at how website analytics identifies traffic sources.
When someone visits a website, analytics software attempts to determine how that visitor arrived. Depending on the information available, a session might be attributed to organic search, paid advertising, social media, referrals, email or direct traffic.
One way to identify a campaign or traffic source is through UTM parameters.
These are small pieces of information added to a URL that help analytics platforms identify where a visit originated.
For example:
https://example.com/services?utm_source=chatgpt.com
The utm_source parameter identifies the intended traffic source as ChatGPT. When a website’s analytics implementation captures the parameter correctly, it can help distinguish those visits from traffic arriving through other channels.
However, campaign parameters are only one way of identifying a visitor’s origin.
Analytics platforms may also use referral information supplied by the browser. If someone follows a link from another website, this information can help identify the referring domain.
Problems can arise when neither signal is available.
If a visitor arrives without a recognisable campaign parameter or referral source, the analytics platform may classify the session as direct traffic or otherwise leave its origin unidentified.
Direct traffic, therefore, does not necessarily mean someone typed a URL into their browser or used a bookmark. It can also include visits where the original source could not be established.
This distinction is particularly relevant as more people use AI assistants to find information and navigate to websites.
What the October reports revealed
The issue gained attention after analysts observed a decline in recorded ChatGPT referrals around the rollout of GPT-6.
According to Search Engine Land’s report, Huesler examined approximately 500 pages where ChatGPT had previously accounted for at least 5% of visits.
His comparison covered 9–13 October 2026 against the preceding week.
The reported findings included approximately 31,000 fewer daily visits attributed to ChatGPT and 26,000 additional daily direct visits across the same pages. Search traffic also declined by around 7%, while approximately two-thirds of the pages losing ChatGPT referrals experienced an increase in direct traffic.
These observations suggested that attribution changes might explain at least some of the apparent decline.
However, the figures should be understood in context. The analysis covered a selected group of pages over a short period, and the increase in direct traffic did not account for every missing ChatGPT visit.
The findings were evidence worth investigating, rather than proof that all the missing visits had simply been relabelled.
There is also an important subsequent development. Follow-up reporting from Search Engine Journal on 16 October described Huesler’s account that OpenAI had identified a bug and begun rolling out a fix. The status and effectiveness of that fix should be checked against the latest information before drawing conclusions about current tracking behaviour.
For marketers, this reinforces the need to distinguish between a temporary measurement problem and a lasting change in traffic or customer behaviour.
Source: Search Engine Land’s original report on ChatGPT referral tracking.
Why businesses should care about attribution
For business owners, the practical concern is not simply whether a particular traffic source appears correctly in a report. It is whether marketing decisions are being made using reliable information.
Imagine a company that publishes helpful content to attract potential customers through search engines and AI assistants.
Its analytics previously showed a steady stream of ChatGPT referrals. Suddenly, those referrals decline, and direct traffic increases.
Without investigating the change, the business might conclude that its content is no longer being recommended by AI platforms. It could then reduce its content investment or redirect its marketing budget elsewhere.
If the apparent decline is caused partly by an attribution issue, that response could be premature.
There are three broader lessons here.
1. Traffic-source reports need context
A change in a channel’s reported performance should prompt questions before it prompts action.
Has the number of visitors actually fallen? Have conversions changed? Is another channel showing an unusual increase? Has the analytics implementation or the referring platform changed?
These questions help separate changes in customer behaviour from changes in measurement.
2. AI’s contribution may be difficult to quantify
Someone might discover a business through ChatGPT, visit its website and later return directly. Another person might use an AI assistant for research before searching for the company by name.
In both cases, the customer journey may involve AI even when the final visit is attributed to another source.
This does not mean every direct visit should be credited to AI. It means that last-click attribution and referral reports cannot always capture the complete path to a purchase or enquiry.
3. Marketing reports should communicate uncertainty
SEO professionals are often expected to explain why traffic has risen or fallen.
When attribution is uncertain, the most useful report is not necessarily the one with the most confident explanation. It is the one that clearly distinguishes observed changes from possible causes.
For agencies, consultants and in-house teams, that distinction helps maintain trust and prevents unexplained changes in reporting from becoming unnecessary strategic concerns.
Five practical steps for marketers
Rather than reacting immediately to changes in AI referral traffic, I recommend taking a structured approach.
Step 1: Establish what has actually changed
Start by reviewing your analytics over the relevant period.
Compare ChatGPT referrals, direct traffic, total sessions and conversions. Examine the pages that historically attracted AI referrals rather than relying solely on site-wide totals.
Look for unusual patterns, including:
- A sudden decline in identifiable ChatGPT referrals.
- An increase in direct traffic to the same landing pages.
- Stable overall sessions despite changing source attribution.
- Changes in enquiries, purchases or other important conversions.
Compare equivalent periods where possible, accounting for seasonality, campaigns and other known changes.
These patterns can help identify a possible tracking issue, but they cannot establish the source of unattributed visits on their own.
Step 2: Audit your analytics configuration
Review how your analytics platform processes campaign parameters and referral information.
For businesses using Google Analytics 4, examine session source and medium, campaign data, landing pages and relevant conversion reports.
Check that your tracking implementation is working correctly and that redirects or other website changes are not removing campaign parameters before analytics captures them.
It is also worth documenting any known tracking changes so that future comparisons are made with the appropriate context.
If a referring platform changes its behaviour, your own analytics configuration may not be the cause, but understanding how your setup processes incoming visits is still essential.
Step 3: Use server-side evidence where available
Server logs can provide another perspective on website activity.
Depending on your setup, they may show requests, requested URLs, timestamps and available request headers. This can help you investigate whether a page is continuing to receive visits even when a particular referral source appears to decline.
However, server logs do not automatically identify the original source of every visitor. If referral information is missing, the logs may not reveal whether a request originated from ChatGPT or another source.
Treat this as complementary evidence rather than a replacement for analytics.
Step 4: Measure conversions, not just sessions
Traffic is a useful indicator, but commercial outcomes are usually more important.
Track the metrics that matter to your business, such as qualified enquiries, sales, bookings, conversion rates, customer acquisition costs and revenue.
If recorded ChatGPT referrals decline while total conversions remain stable, investigate attribution before concluding that the channel has lost value.
If both traffic and conversions decline, a genuine performance issue may exist, but you should still consider tracking problems alongside other possible causes.
The objective is to make decisions based on the strongest available evidence rather than one isolated metric.
Step 5: Review your reporting after platform changes
When a significant tracking issue is reported, annotate the affected period in your analytics dashboards and marketing reports.
This makes it easier to understand historical fluctuations and avoids treating a measurement change as a confirmed change in performance.
It is equally important to check for subsequent fixes. If a platform restores its tracking behaviour, the resulting increase in reported referrals may reflect improved attribution rather than a sudden increase in the number of visitors.
In other words, both a decline and a recovery in recorded traffic need to be interpreted in context.
Should businesses change their SEO strategy?
A potential tracking issue is not, by itself, a reason to overhaul an SEO strategy.
It does, however, highlight why businesses should think more broadly about online discovery.
Search engines remain a major source of information and website visits. AI assistants offer another way for people to discover brands, explore services and evaluate options.
Businesses should aim to build a digital presence that helps customers find accurate, useful and credible information, regardless of the platform they use.
For most organisations, that means continuing to invest in the fundamentals of SEO while paying attention to how AI-powered discovery fits into the wider customer journey.
Create content that answers genuine customer questions
Useful content remains central to an effective search strategy.
Businesses should explain what they offer, who they help, how their products or services work and what customers should consider before making a decision.
For a local service provider, that might mean detailed service pages, clear pricing information where appropriate, answers to common questions and examples of completed work.
For an ecommerce business, it could involve comprehensive product information, accurate specifications, useful comparisons and transparent delivery and returns policies.
For a B2B company, it might include industry analysis, case studies, original research and practical resources that help potential customers evaluate suppliers.
The goal is not to produce content solely to attract AI citations. It is to provide information that genuinely helps people make informed decisions.
Strengthen credibility and expertise
Customers need reasons to trust the information they find.
Clear business information, demonstrable expertise, original insights, credible references and relevant customer evidence can all contribute to a stronger online presence.
For SEO professionals, this means looking beyond the number of pages published or keywords targeted. The quality, accuracy and usefulness of the information matter too.
No particular content format guarantees that an AI assistant will cite or recommend a business. A strong, trustworthy website nevertheless provides a sound foundation for both traditional search visibility and broader online discovery.
Keep technical SEO in good shape
Important website pages should be accessible, indexable where appropriate and easy for search engines to understand.
Review crawlability, internal linking, page performance, canonicalisation, structured data where relevant and the accuracy of important business information.
These activities support the overall quality and discoverability of a website, although they cannot guarantee inclusion in AI-generated answers.
Most businesses will benefit more from maintaining strong technical foundations and useful content than from chasing every new AI optimisation trend.
A better framework for measuring AI visibility
If referral data alone is incomplete, businesses need a broader measurement framework.
I would combine four areas of analysis.
1. Website analytics
Track identifiable AI referrals, landing-page engagement and conversions. Continue to use these metrics, but recognise that source attribution may be incomplete.
2. Search performance
Monitor organic impressions, clicks, relevant search queries and landing-page performance. These measures help explain broader search trends, even though they do not capture every form of AI discovery.
3. AI visibility monitoring
Test a consistent set of relevant customer questions across selected AI assistants. Record whether your business is mentioned, cited or recommended, and repeat the tests over time.
Remember that AI responses can vary. A small collection of prompts cannot provide a definitive measure of total AI visibility or prove that those mentions generate website visits.
4. Customer and commercial data
Ask customers how they discovered your business where appropriate, and connect marketing reporting to qualified leads, sales and revenue.
This can help identify influence that conventional referral reports miss, particularly when customers use multiple channels before converting.
Together, these methods provide a more rounded view of performance than relying on a single traffic-source metric.
The bigger lesson: Build a measurement strategy that can adapt
The wider digital marketing landscape will continue to change.
Platforms can modify how they present links, pass referral information and support campaign tracking. Browsers and privacy settings can also affect the information available to analytics systems.
Businesses cannot control every part of that process. They can, however, make their measurement practices more resilient.
That means documenting changes, checking unusual results, combining multiple sources of evidence and avoiding major strategic decisions based on one unexplained fluctuation.
It also means being honest about the limitations of available data.
There will be situations where the original source of a visit simply cannot be established with confidence. Acknowledging that limitation is more useful than assigning traffic to a channel without sufficient evidence.
Final thoughts
The recent reports concerning ChatGPT referral tracking highlight an important issue for anyone working in SEO and digital marketing: the numbers in an analytics dashboard do not always tell the complete story.
A decline in recorded AI referrals might reflect a genuine change in visitor behaviour, an attribution problem or a combination of factors. The only sensible response is to investigate the evidence before drawing conclusions.
For business owners, that means resisting the temptation to make immediate budget decisions based on unexplained changes in traffic.
For marketers and SEO professionals, it means combining reliable measurement with a broader understanding of how customers discover, evaluate and choose businesses.
AI-driven discovery deserves attention, but it should be measured alongside the rest of your marketing activity rather than treated as an isolated channel.
My recommendation is to focus on three things: maintain strong SEO fundamentals, measure the commercial outcomes that matter, and investigate unexpected analytics changes before acting on them.
As AI continues to influence how people find information online, businesses that combine useful content with sound measurement will be better equipped to understand what is working and where to invest next.
Further reading
This article draws on reporting by Search Engine Land and subsequent industry coverage of the October 2026 tracking issue. For the original findings, see Search Engine Land’s report on ChatGPT referral tracking and Search Engine Journal’s follow-up coverage.
