For months, discussion around artificial intelligence (AI) visibility highlighted Reddit as an important citation source for results.
The community platform was said to be “outranking” corporate websites in citations, as Semrush put it last September. And on a hotel discovery webinar in July, Javier Hernandez, senior SEO manager at Propellic, said “Reddit is a real visibility lever” in AI results.
But it appears the tide has turned: AI visibility platform Promptwatch reported that AI nearly stopped citing Reddit in August.
Travel companies are playing on an AI visibility field that is constantly shifting, making it difficult to maintain a strong optimization strategy.
“Reddit lost almost its entire citation footprint in ChatGPT within a single day,” Klaas Foppen, founder and CTO of Promptwatch, wrote in a blog post.
Foppen said that from July 18 through Aug. 7, Reddit held 3.83% average share of citations on ChatGPT, among the largest of any domain. Then, on Aug. 14, that share ducked under 1%. From Aug. 14 to 17, it fell to an average of 0.52%, indicating an 86.4% relative decline.
Noah Greenberg, CEO at Stacker, called out a similar decrease cited by Petra Labs, which said Reddit citations fell by 80% on Aug. 16.
“Three months ago, every headline, every expert was standing on a soapbox talking about how Reddit was the future, taking up the majority of citations,” Greenberg wrote on LinkedIn.
What has dropped are the citations (what you and I can see in the answer) but not the retrieval part (the source).
Nicolas Sitter, Lighthouse
But the fall in attribution to Reddit doesn’t necessarily mean sourcing has fully stopped, according to Nicolas Sitter, head of AI labs at Lighthouse and co-founder of Hotelrank.ai, which was acquired by Lighthouse.
“What has dropped are the citations (what you and I can see in the answer) but not the retrieval part (the source). ChatGPT still heavily 'consults' Reddit. It just doesn't cite it in the answer,” Sitter told PhocusWire. “This has already happened before, and the behavior quickly reverted to the original.”
He still sees Reddit as important, particularly in the long term. “On a more global level, there's a lot of product decisions that will drive noise, and you cannot optimize against that week to week.”
No one is able to promise what a large language model (LLM) will deliver in a response, a Reddit spokesperson told PhocusWire. LLMs are updating often, and that means ranking signals are changing and what is cited now might not be cited in the future.
“We are at a time where models are constantly shifting the weights of how they show results and this is going to be an incredibly volatile system for a long time to come,” Greenberg said.
A changing landscape
AI visibility is hard to read in terms of execution and channels, according to Tom Rudnai, founder and CEO of Demand-Genius.
Greenberg cautioned against creating a rigid roadmap, given the current conditions.
“We are at a point in time where you should not set a long-term AI visibility strategy based on anything you read, or you will end up like digital media companies in the late 2010s who went all in on video for Facebook and then got left holding the bag,” he wrote. “Things are just too volatile right now to set a long-term strategy.”
Sitter doesn’t think what’s happening is a perennial problem.
“I think people say it's unpredictable, but that's very short term,” Sitter weighed in.
The nature of the technology also lends itself to unpredictable tendencies.
“Ask 10 times the same question, you won't get 10 times the same answer,” Sitter said. “So why would AEO be fully predictable?”
Things are just too volatile right now to set a long-term strategy.
Noah Greenberg, Stacker
Meanwhile, Brennen Bliss, CEO of Propellic, framed what’s happening as a symptom of how early things are with AEO or GEO.
“In travel, a big share of the queries we care about get routed through live web search inside the chatbot, the ‘searching the web…’ moment you see in ChatGPT,” Bliss said.
Optimizing for those queries can look pretty similar to traditional optimization for ranking, he said.
“The unpredictable part is the model's stored, trained knowledge,” Bliss said. “The feedback loop there is long: Your content has to make it into the training data, and then you wait for the next model release to find out whether it worked. It's very hard to test and iterate when one loop takes months.”
According to Rudnai, travel companies need to build a foundation with historical marketing practices to establish an understanding of who they are, who they exist for and how they win.
“The channels you communicate that across will change, but the core is the same,” Rudnai said. “Brands that have defined and documented these fundamentals—which all brands should have anyway—will be able to pivot with the unpredictability comfortably enough.”
Tips for navigating
Bliss compared what’s happening now to the beginnings of Google Search.
When Google Search was new, the algorithm swung constantly. Any slight shift could significantly alter the level of visibility for any site. Now, the same thing is happening with AI what surfaces, he said.
“Outside of Google, the frontier labs (OpenAI, Anthropic and the rest) are still early in how they index and rank sources, so their changes are bigger and blunter,” Bliss said. “Content deals with publishers and forums like Reddit have also come and gone, which explains a lot of the swings. Generative engine optimization will be a moving target for a while.”
He anticipates this movement will follow the same arc as Google, slowly becoming more stable with bigger updates quarterly to suppress unwanted content.
“If you've got the resources to optimize and pivot as it moves, it's worth the effort,” Bliss said.
For his part, Rudnai cautioned against chasing hype.
“You’ll find no shortage of ‘experts’ selling quick wins and hacks,” he said.
“Fundamentally though, the key is building high quality, differentiated content and distributing it widely to maximize human and AI influence,” Rudnai continued. “The distribution strategy needs to evolve continuously, but the foundation doesn’t: produce differentiated, opinionated content and back it up with original research.”
The number one piece of advice Rudnai would give to travel companies is to treat AEO as a discipline instead of a project.
“I always suggest brands reframe the question everyone is asking; not ‘how do I get AI to cite us?’ but instead, ‘why should AI cite us?’” he said.
Phocuswright's Travel Marketing AI Summit London
Join Phocuswright in London on Sept. 8 to hear more from Tom Rudnai, founder and CEO of Demand-Genius.
Take Phocuswright's annual industry survey
Are you keeping pace with AI? Phocuswright is conducting it’s annual global survey to understand how travel companies are actually putting AI to work. Complete the survey to benchmark your progress via an exclusive summary of the findings as well as a chance to win 1 of 2 tickets to The Phocuswright Conference 2026 or 1 of 2 tickets to WiT Singapore 2026.