While big AI companies are delving into legal gray areas to find new data sources for large-scale language models, the picture is black and white for website operators and publishers. Traffic to websites is drying up, and as search engines transform into generative AI chatbots, a real drought could be in the offing.
To learn more about how search engines are changing in the AI era and what website operators can do to adapt, check out the founding of BrightEdge, a 17-year-old enterprise SEO platform. We spoke with Jim Yu, CEO and Executive Chairman.
Are Perplexity citations currently driving traffic?
BrightEdge recently published some interesting research on Perplexity, an AI chatbot platform looking to compete with Google in the search space (when I spoke to Perplexity CEO Aravind Srinivas in January, he said he was PageRank's search engine).
What immediately caught my attention in this study was Perplexity's claim that more people are clicking on quote links. “According to BrightEdge analysis, the number of referrals from Perplexity to branded sites has increased by nearly 40% month-over-month since January,” the media release states. “This is the first evidence that Perplexity is not being used solely for generated content. It also acts as a search engine that directs users to your site through citations” (emphasis mine)
As a regular user of Perplexity, this resonated with me. (Disclosure: Perplexity provided me with a year of Pro service while researching the January article). When I first started using Perplexity Pro, quote links appeared in the chatbot's replies to questions I posed, but there were very few prompts to click on the links. But recently I started noticing that Perplexity was encouraging users to click on these quote links. Many of my replies now include phrases like “according to search results” or “key evidence”, which tends to tempt people to click on the accompanying link.
I asked Yu if this could be the reason why he's seeing an increase in referral traffic from Perplexity to his website. Although he wasn't interested in that specific question, he said that Perplexity has a clear position in the search space, diametrically opposed to his ChatGPT.
“I think the core of the Perplexity AI experience is that it provides well-formulated, well-researched answers to questions,” he said, adding that it is much more transparent than ChatGPT in that regard. I pointed out that it was expensive. He added that Perplexity not only provides sources, but also “great AI summaries.”
It's clear that the main value users get from Perplexity is the summaries of the source material. But it could also be in the company's interest to have people click on quote links as a way to gain user trust in Perplexity. Either way, his BrightEdge results in terms of referral traffic are encouraging from a publisher's perspective.
Awakening the Giants: Google's search generation experience
While Perplexity has captured the attention of many early adopters of the technology, the 800-pound gorilla of search is quietly preparing its own AI-based search. Google's “Search Generative Experiences” (SGE) have been an optional feature in Google Labs for nearly a year, but for most of that time they were geographically limited to users in the United States. It has recently been expanded to other countries, but I have yet to find it in the UK (where I live). So, I asked Yu about his impressions of Google's SGE.
“Google is really somewhere in between the two,” he said, referring to ChatGPT and Perplexity. He said SGE is what he considers “much more transparent” than ChatGPT. He added that while it's “closer to Perplexity than the GPT model,” it's “less obvious” in terms of how citation links are assembled.
Yu has watched Google go through a series of iterations to improve SGE over the past 11 months or so.
“In the early days, that experience wasn’t cached,” he said. “So it was pretty slow to load. Now that it's cached, it's pretty fast. You can also see that we're doing industry-specific tuning.”
In particular, industry adjustments include implementing “guardrails” in industries such as healthcare and finance as a form of risk management in how AI search is implemented in these industries.
“The number one place they use AI is your health,” Yu said of Google's testing of SGE. “The reason is that they are very careful about that. So when you search for a disease in the AI version, [of Google Search], you will see a disclaimer at the top. You should consult your doctor. The second thing is that it takes up a large portion of the page and clearly emphasizes certain sites such as the CDC (Centers for Disease Control) and his NIH (National Institutes of Health).They are [also] Compare Mayo Clinic and major medical institutions. ”
These same sources also rank highly in Google's current non-AI searches, but Yu points out that they are even more prominent in AI search results. He says this extra weighting, along with the disclaimer, is because “AI generates certain opinions,” so Google naturally wants to protect itself.
This seems to suggest that Google SGE will focus more on authoritative websites than ever before. That means it will be harder for new entrants (for example, new health information websites) to rank well in Google's AI searches. So I wondered what a web operator could do to optimize her AI search, considering that AI search will obviously be harder to get cited (especially in an industry like healthcare). asked Yu.
“If you produce great content and are a trusted brand, you have a good foundation to build on,” he said. “The future is that you need to better understand what conversations you have a right to win as a brand. Please try to create it.”
Will publishers benefit from AI citations?
Getting cited in search results is great, but of course websites want traffic coming back from those citations. BrightEdge's research shows that referral traffic from Perplexity increases, but that doesn't necessarily happen with Google SGE. Especially since Mr. Yu has stated that his SGE doesn't display quotes as prominently as his Perplexity. So I asked him what website operators, especially the media companies that rely on their traffic, can do to adapt to his AI search era.
He responded that media sites “need to be very keenly aware of which areas they have the right to win and in which they have the most expertise.” As such, brand authority will become even more important as “AI engines will need to somehow attribute trust to different sources.”
He said AI engines will increasingly rely on “authority graphs” that build metadata about publishers to organize sources. This is already done in traditional search engines, but now he is entered in LLM. “All the elements you want to surround your content with, you want the AI to know how to attribute it,” he says.
All of this means that being cited in Google SGE or Perplexity (or Bing's OpenAI-based solutions) will increasingly drive your search optimization strategy. In fact, Perplexity's Aravind Srinivas said something similar in his January interview. According to Srinivas, the more Perplexity's products cite a particular web page of his, the more important it becomes.
“Next generation rankings will be cited,” Yu said. “As a publisher, we want to be the first cited source, because all other information after that will be generated from the AI based on the original source.” Therefore, the graph will be an original source of information on various topics. So I think this is very important for publishers. ”
Yu's argument is that being an original source of information, or in other words, a “trustworthy source,” will be important for publishers in the future. Unfortunately, even if you are the original source for a particular topic, there is no guarantee that traffic will return to them. But there's at least some hope that Perplexity is starting to drive traffic back to publishers, and we can only hope (and pray) that Google follows suit.
YouTube.COM/THENEWSTACK
Technology moves fast, so don't miss an episode. Subscribe to our YouTube channel to stream all our podcasts, interviews, demos, and more.
subscribe