With various AI platforms continuing to gain momentum, it would be super beneficial to have the ability to parse out leads from ChatGPT, Claude, Perplexity, etc. from into their own, new Original Source category: AI Search.
This way, HubSpot CRM users can more easily see data in their dashboards at a high level and can pinpoint which popular AI platforms are producing leads. This is especially important since, currently, the Original Source they get lumped into just covers any website that isn't their own (outside of search engines).
This new option allows marketers to make more informed decisions on their strategies and what channels to pay attention to more.
Fabulous idea! We have to use additional properties like 'Original Traffic Source Drill-down 1' to identify those leads, which is annoying. AI platorms definitely deserve a dedicated Original Source option at this stage.
Really love this idea– thanks for posting it to this forum!
I've been thinking about this from a few different angles, and I would love to hear more customer perspectives about what would be most valuable. Here are some of the data points I'm currently thinking through:
Leads referred directly from AI (e.g., a person is in an AI tool, they ask a question, your sources are cited, and they click that link over to your site)
Content crawled by flagship AIs (i.e., what if we can return server logs to quantify which bots have crawled which pieces of content)
Anything else related to AI-influenced leads that you would find helpful to track?
Those both sound great! That covers everything I think my team could utilize in terms of tracking. In turn too, being able to see the "first page seen" and other contact attributes will be great to have since we know a lot of tools don't have (or freely offer) the ability to see those kinds of insights.
I think we need a solution for this challenge, but this solution as proposedi think will cause confusion.
ChatGPT isn't search so "AI Search" doesn't make sense as a label.
An organic search on Google that generates a click from a Gemini summary's cited page—where does that get filed? "AI Search" or "Organic Search"?
Perplexity is testing Ads in their results. Surely other LLMs will follow—Is that "Paid Search" or "Paid AI Search". They're going to be all the same Search Ads.
As basically all search tools go to AI-first, what's the differece between "Organic Search" and "AI Search"?
HubSpot is already recognizing AI traffic in drilldowns, the question is how should it rollup.
Totally agree with this. As more people discover content through tools like ChatGPT and Perplexity, it’s getting harder to track that traffic in a meaningful way. Right now, it all just ends up under generic buckets like Referrals or Other, which isn’t helpful. Having AI Search as its Original Source would give marketers much clearer insights into what’s actually working.
But if someone is searching for something on AI, and they come to your website, I would conclude that say that label makes sense. But AI isn't a traditional search engine. It does get hairy with Google implementing AIOs and AI Mode, but we already know we won't get proper tracking for that in any Google tool. So there is nothing HubSpot would be able to do about that no matter what, no marketer can.
As for ads, sure paid AI could be a thing, if perplexity adds the proper UTM parameters to the click data. There is both paid social and paid search already in there, so yes, that would be different in the grand scheme of things.
Overall, just think there needs to be some different way to classify leads that come in from ChatGPT, Claude, or Perplexity. It can help internal or external marketing teams/agencies see what (if any) ROI is coming from those tools separate from traditional search engines and other referrals.
Happy to see more HubSpot users suggesting this—hopefully we’ll soon get better reporting.
I think a good label for this type of traffic (LLM chatbot referrals) would be "Organic LLM". When/if ads are introduced in LLMs at some point, a similar channel and logic could be introduced as "Paid LLM". This would allow us to both bucket all LLM traffic and segment it into paid and organic, following existing conventions in HubSpot (for eg. paid and organic search). It would also align with major analytics platforms like Google Analytics and the source/medium split—where, for example, chatgpt.com would be the source and organic the medium.
I suggest "LLM" as it’s short, simple, and already adopted by a large part of the community—business execs, analysts, marketers, SEOs, etc.—and covers all AI chatbots, as they’re built on LLMs. Gen AI is another option, but in that case, terms like chatbots or conversational AI would seem more accurate. That’s likely also why “LLM platforms” is gaining traction and adoption as a standard term.
Ideally, Organic LLM should also be available in reports covering web traffic—not just leads.
My existing HubSpot idea related to this can be seen through below link. It also covers the specifics on distinctions between sources like google.com/search and gemini.google.com.
Hello HubSpot Community - Good suggestion! We took it and ran with it.
Announcing a new web traffic source, Contact Original Source, and Contact Latest source: "AI Referrals".
This source will track referrals from ChatGPT, Claude, Gemini, Perplexity, Grok, and a few other sources. Full Knowledge Base article to come in September.
@shygum you could use a workflow that adds contacts where first page seen/first referring site contains markers that it came from an AI source (such as ChatGPT) and then changes the data value Original Source to AI.
@jeffvincent do you have any details on what criteria this attribution uses? For example, we had an MQL come in where "utm_source= chatgpt.com" (the MQLs occured after the AI value for Original Source was added into our system) but weren't picked up by it (was attributed to "Other Campaigns" instead).
Q: Can this work against data retroactively? A: For contact data in CRM, source category will be automatically re-calculated at their next update (eg when they submit a form, or visit your webpage again). For web traffic, we do not plan to retroactively apply the categorization, because that task is very large and risky.
Q: What's the cut-off date where this new beta filter started populating? A: Once your portal was enrolled, the new filter started populating when the first LLM sourced lead came in. The category was created in your portal at that moment.
Q: Will the logic be updated by HubSpot in the future, and is portal customization possible? A: Excellent question, we're discussing this actively. Our plan IS to keep the logic updated (this AEO space is changing fast!) but also hoping to support some level of customization for industry-specific LLM etc
