ChatGPT Referral Traffic Increased 8x in 5 Months Driven by Third-Party Mentions and Page Optimization

Zeeshan · 2026-06-08T09:08:46.812Z

Pulled our ChatGPT referral numbers this week. Last 5 months vs Aug to Dec 2025, sessions went from 114 to 879, close to 8x, at near 99% engagement. It is now one of our TOP referral sources. This was not luck. Two things did the work 👇 1. We earned placements on third-party listicles and review sites, especially the “best X” roundups that models often reference when giving recommendations. 2. We cleaned up our own pages by adding clear answers, definitions, and positioning right at the top, making it easier for models to understand, lift, and credit the page. What stood out: the off-site signal seems to carry more weight than the page itself. Anyone else seeing real LLM referral traffic yet? And is it coming from your own pages or third party mentions?

14 comments

· Sorted by Oldest

Alexandre C.
·
·
Zeeshan what are you using to earn those placements and to monitor LLMs?
Zeeshan
·
(edited)·
Alexandre C. Basically we look at our target queries in the LLMs, see which citations are already ranking for them, then reach out to those specific sources and pay a placement fee to get our link in there. The thinking is that whatever is already getting cited by the models is the best place to start, since they've already proven the models trust them. Working well for us so far 🙌
👍1
🙌1
Alexandre C.
·
·
That's awesome Zeeshan congrats 🙂
Lauren R.
·
·
Super interesting, thanks for sharing your process and learnings!
Nicolas L.
·
·
Super insightful! Zeeshan. Curious - do you run any validation on the specific queries that you are considering for the analysis? How is your selection process?
Ilya S.
·
·
can you share what your strategy is around third-party listicles? How does this work - do you just reach out to post authors asking to add or offer a guest post?
Zeeshan
·
·
Lauren R. Really appreciate that! 🙌
Zeeshan
·
·
Thanks!Nicolas L. Yeah, we start from commercially relevant queries, the "best X" and "top tools for Y" type prompts plus the actual questions our buyers ask. Then we run each one a few times across models since outputs shift run to run, so single checks don't cut it. We prioritize the ones where competitors get cited and we don't. Those gaps convert best.
Zeeshan
·
·
Ilya S. Yeah, pretty much. There's no single playbook, it depends on the site and the author. Sometimes we do a 3-way exchange to keep things clean, or swap a placement on a partner site if they want a link back. That said most of the time it's a paid placement, so it really comes down to solid outreach and negotiating the best possible deal.
Ilya S.
·
·
Zeeshan fair enough, thanks for the details!
Nicolas L.
·
·
Appreciate all the detail Zeeshan!
Cristina R.
·
·
Just in case, we use Profound to monitor visibility and sentiment across LLMs and it’s true what you said. You need to make sure you are fetchable by LLMs (both tech SEO, and on page SEO, especially meta data, meta description, url, title, etc) and then make sure that there are “easily quotable” answers in your content. The LLMs won’t get to the good part if there is a lot of fluff, it’s hidden behind UI, etc. And just ensure you are analyzing the prompts that actually matter, which is as you said, the commercial ones. The ones that actually get the searcher a recommendation for your business/product/service.
Cristina R.
·
(edited)·
Profound (and I promise I’m not in any way connected to them I’m just a happy user), allows you to see where the citations come from for the recommendations, and this is huge, because you can ser where your competitors are being cited. And then you can create content that addresses those gaps or reach out to the same sites they are being cited in and see if you can get cited too.
Zeeshan
·
·
Cristina R., great point! 🙌🏻 Especially around making the content “easily quotable.” That’s exactly where I think a lot of traditional SEO pages miss the mark: the answer exists, but it’s buried too deep for models to confidently lift or cite. Also agree on citation-gap analysis. Seeing where competitors are being referenced gives you a much clearer outreach/content roadmap. Curious, which Profound plan are you using, and has it been worth it mainly for citation tracking?
❤️1

Alexandre C.
·
·
Zeeshan what are you using to earn those placements and to monitor LLMs?
Zeeshan
·
(edited)·
Alexandre C. Basically we look at our target queries in the LLMs, see which citations are already ranking for them, then reach out to those specific sources and pay a placement fee to get our link in there. The thinking is that whatever is already getting cited by the models is the best place to start, since they've already proven the models trust them. Working well for us so far 🙌
👍1
🙌1
Alexandre C.
·
·
That's awesome Zeeshan congrats 🙂
Lauren R.
·
·
Super interesting, thanks for sharing your process and learnings!
Nicolas L.
·
·
Super insightful! Zeeshan. Curious - do you run any validation on the specific queries that you are considering for the analysis? How is your selection process?
Ilya S.
·
·
can you share what your strategy is around third-party listicles? How does this work - do you just reach out to post authors asking to add or offer a guest post?
Zeeshan
·
·
Lauren R. Really appreciate that! 🙌
Zeeshan
·
·
Thanks!Nicolas L. Yeah, we start from commercially relevant queries, the "best X" and "top tools for Y" type prompts plus the actual questions our buyers ask. Then we run each one a few times across models since outputs shift run to run, so single checks don't cut it. We prioritize the ones where competitors get cited and we don't. Those gaps convert best.
Zeeshan
·
·
Ilya S. Yeah, pretty much. There's no single playbook, it depends on the site and the author. Sometimes we do a 3-way exchange to keep things clean, or swap a placement on a partner site if they want a link back. That said most of the time it's a paid placement, so it really comes down to solid outreach and negotiating the best possible deal.
Ilya S.
·
·
Zeeshan fair enough, thanks for the details!
Nicolas L.
·
·
Appreciate all the detail Zeeshan!
Cristina R.
·
·
Just in case, we use Profound to monitor visibility and sentiment across LLMs and it’s true what you said. You need to make sure you are fetchable by LLMs (both tech SEO, and on page SEO, especially meta data, meta description, url, title, etc) and then make sure that there are “easily quotable” answers in your content. The LLMs won’t get to the good part if there is a lot of fluff, it’s hidden behind UI, etc. And just ensure you are analyzing the prompts that actually matter, which is as you said, the commercial ones. The ones that actually get the searcher a recommendation for your business/product/service.
Cristina R.
·
(edited)·
Profound (and I promise I’m not in any way connected to them I’m just a happy user), allows you to see where the citations come from for the recommendations, and this is huge, because you can ser where your competitors are being cited. And then you can create content that addresses those gaps or reach out to the same sites they are being cited in and see if you can get cited too.
Zeeshan
·
·
Cristina R., great point! 🙌🏻 Especially around making the content “easily quotable.” That’s exactly where I think a lot of traditional SEO pages miss the mark: the answer exists, but it’s buried too deep for models to confidently lift or cite. Also agree on citation-gap analysis. Seeing where competitors are being referenced gives you a much clearer outreach/content roadmap. Curious, which Profound plan are you using, and has it been worth it mainly for citation tracking?
❤️1