Google handles about five times the search-like queries attributed to ChatGPT, according to OpenAI consumer product lead Adam Fry. That comparison describes activity inside the chatbot, though, rather than the volume of traffic reaching client websites.
For web hosting resellers, referral data offers the more useful signal. It shows which visits arrive, what resources they consume and whether AI channels deserve space in client reporting.
Quick take:
- Fry estimated that Google’s search-like query volume is approximately five times ChatGPT’s.
- OpenAI counts a search whenever a ChatGPT answer requires a web request.
- Big Technology reported that ChatGPT’s weekly active users increased from 1 billion to 1.2 billion over two months.
- Ahrefs found that Google generated 24.74% of tracked referrals across 118,671 sites, while ChatGPT generated 0.36%.
- The comparison lacked details about the Google baseline, measurement period and geographic scope.
Three metrics that shouldn’t be confused
OpenAI’s internal search measure records questions that cause ChatGPT to retrieve information from the web. Those requests may not generate a visit to the cited page because the answer can remain inside the chatbot.
Ahrefs’ referral tracker works at a different level. It observes visits sent to participating websites and identifies the referring service. Its November update showed Google contributing a much larger share of traffic than ChatGPT.
Website referrals matter more directly to hosting clients. They help explain changes in bandwidth, page requests and server load. Internal search activity may demonstrate changing user behavior, but it does not automatically create a sales opportunity for a web host.
Marketing teams should likewise avoid combining prompts, search requests and referred visits under one “AI traffic” label. Each metric answers a separate question and should be reported independently.
Where the estimates diverge
Fry discussed the comparison during a public dinner hosted by OpenAI in New York. Big Technology later shared the transcript.
He explained that early ChatGPT adoption centered on writing and coding requests. The following year brought more questions involving current events, recognizable people and places, financial information and health. Those subjects often need fresh information, prompting the chatbot to search online.
Ahrefs published a separate February estimate that placed ChatGPT’s search volume at about 12% of Google’s. Its model started with a July 2025 study from Harvard University and OpenAI that counted 2.5 billion prompts per day. Ahrefs then classified roughly 65% as likely traditional search-type requests before comparing that portion with Google’s volume.
The different definitions produced different results. Ahrefs categorized prompts by their purpose, whereas OpenAI counted web requests initiated by users. Google’s own third-quarter disclosure offered only a query-growth rate, not an absolute count that could support a direct comparison.
Why this matters for resellers
For agencies reselling hosting, the key concern is client liability. Claims about traffic must be precise, especially when they influence capacity, pricing and performance promises.
Referral analytics create a legitimate upsell path. Resellers can configure clearer reporting, watch traffic quality and connect visit patterns to resource usage. They can also help clients distinguish an informational query from a visitor who reaches a landing page and takes action.
White-label dashboards need separate fields for AI referrals and any separately supplied search metrics. Combining them may make ChatGPT appear responsible for more demand than it currently sends to websites.
There is also no basis in the reported evidence for treating ChatGPT as a replacement for Google. Fry’s ratio measures search-like activity, while Ahrefs measures outbound visits. Neither demonstrates whether a user stopped searching on Google and moved the same request into ChatGPT.
A practical cPanel workflow
Say you operate WordPress hosting for five clients and discover ChatGPT referrals in their reports. Before changing their plans, compare three figures: referred visits, server requests and bandwidth consumption.
If referrals are rising but resource use remains modest, retain the current package and improve attribution. If visits are increasing alongside sustained demand, consider a plan with more capacity. Document the evidence in the client report.
A useful monthly checklist includes:
- Referrals by source, including ChatGPT and Google
- Page views or meaningful actions reached through each source
- Bandwidth and resource use during peak periods
- Whether AI referrals change the sizing assumptions for the next term
This approach gives buyers a defensible recommendation rather than an AI-themed sales pitch.
What to watch next
Monthly referral shares will provide the clearest evidence of whether ChatGPT is becoming a meaningful acquisition source. Query volume offers broader behavioral context, but it cannot substitute for measured website visits.
For resellers, the practical stance is straightforward: track referrals accurately, connect them to infrastructure demand and keep reporting labels distinct. Treat AI search as a trend to monitor, not yet as a reason for broad hosting upgrades.
Key takeaways
- Separate prompts, web searches and website referrals instead of grouping them as AI traffic.
- Compare referred visits with bandwidth and resource use before changing client plans.
- Give clients precise attribution in white-label reports and monthly dashboards.
Frequently asked questions
How much search-like query volume does ChatGPT have compared with Google?
Fry placed ChatGPT at approximately one-fifth of Google’s volume. He did not disclose the Google baseline, measurement period or geographic coverage.
Why is ChatGPT’s share of website referrals much smaller?
A search request may be answered within ChatGPT without sending the user to a source website. Internal activity and outbound visits are therefore different measurements.
Should a reseller upgrade hosting plans because of ChatGPT growth?
Not from query estimates alone. Use referral visits, bandwidth and server demand to establish whether additional capacity is needed.
How should AI channels appear in client reports?
Give each source its own attribution label and state the measurement period. Keep prompts, searches and referrals separate wherever the available data permits.