The rise of generative AI and large language models (LLMs) has opened the possibility of disrupting Google’s long-standing dominance in the search engine market. One such challenger is Perplexity, an AI-powered search platform that aims to provide users with summarized answers and cited sources instead of the traditional list of web page links.
While still dwarfed by Google’s massive search volume, Perplexity’s approach offers a glimpse into how AI could reshape the search experience and the multibillion-dollar search advertising industry.
What is Perplexity AI?
Perplexity describes itself as “your AI-powered Swiss Army Knife for information discovery and curiosity.” It sits on top of ChatGPT 3.5 and its own model. The premium version also includes access to GPT-4, Claude 3, Mistral Large and an Experimental Perplexity Model.
While it’s already seeing significant traction, responding to over 2 million queries per day, it has a long way to go to make a dent in Google, which gets that many searches in 20 seconds. However, consumers are starting to find value in the summary of data and information. Getting to a summarized “answer” more quickly can be valuable.
How to approach AI-powered search platforms like Perplexity
So, how should brands and search engine marketers think about a platform like Perplexity? There are three things to consider:
1. Do your research
You must be using these tools. I’m not suggesting that you have to use them exclusively, but you simply can’t take for granted that you will get to it later. These tools are here and are evolving rapidly.
Do some research, run some queries, then refine them with different criteria. See what the responses are.
Do the responses favor your business?
Do the responses cite your business?
Is there new information you had not considered before that should be something you work into your marketing and content strategy?
Below are two simple examples of searches: “best car insurance” and “car wash near me.”
Comparing the two results from Google and Perplexity, I first notice the lack of ads in the car insurance example.
Allstate takes up pretty much the whole page with their search ad. Perplexity gives the user more of an answer with citations. The information the user may find helpful and dive deeper, but getting to a specific website is harder.
In some cases, this may improve the user’s experience by removing the click and summarizing the most “helpful” (as deemed by the model) information.
The map details in the car wash search are similar, but Google Maps provides the ability to link to or call the business directly from search results. This requires a click away from the Perplexity results to get there.
2. Sources = rankings
In Perplexity, one of the key differences is the citations or sources. This is critical for two big reasons:
It provides valuable insight into the reasoning behind the answer. While many language models now offer some insight into their sources, prominently featuring them allows users to understand the response better and explore further if desired.
From a search marketer’s perspective, these sources can be likened to the new organic search listings. If Perplexity provides the answer directly, users may not need to visit your site to take action. However, being cited as the source of the answer could become the next best option for brands as consumers become more familiar with these models and brands aim to retain their search traffic.
Dig deeper: LLM optimization: Can you influence generative AI outputs?
3. Conversations vs. searches
Perplexity follows a similar approach to other LLMs, allowing you to carry on a “conversation.” That means you can ask one question, and the next one carries the prior question’s context.
For example, when I asked, “I’m a 45-year-old male who runs about 20 miles per week. What are the best running shoes for me?” Then, I followed that response by simply saying, “What if I up my mileage should my options change?”
The model kept the knowledge that I am 45 and am looking for shoe recommendations. I didn’t need to resubmit that context. It also suggested some related queries relevant to the conversation and took me deeper into my research with simple answers.
What’s next for Perplexity?
Perplexity has developed a robust model and user interface, making it easy for users to learn and utilize. But what’s next? They need to determine their revenue model.
In the example about running shoes, you might have noticed what didn’t occur. Perplexity didn’t offer a direct link to purchase the shoes, even when asked about buying a specific type. Despite the request to buy that brand, it couldn’t provide a direct link to Asics. Even the sources didn’t link directly to the Asics site.
Commerce and conversions aren’t at the forefront of these models right now. The revenue model for Perplexity is the same as that of OpenAI and others. They offer a freemium model, with the ability to upgrade for additional features for $20 per month.
As they earn some revenue with this model, I expect things to come that drive more commerce transactions. They may move to an affiliate or PPC model for these queries where Asics can buy access to this “answer” with a direct link to Asics.com.
Don’t think for a second that the $110 billion dollars search market is going to go to zero. If anything, the speed and rate at which these tools can provide answers should generate more search volume and transactions and not less.
The bigger question isn’t what it does to search volume but rather what it does to the revenue model. For now, we have to wait and see what happens.
