It’s hard to remember how the internet was before tools like ChatGPT, Claude, Gemini, Perplexity and others came into our lives. The meteoric rise of Large Language Models (LLMs) brought a new era for search. The new platforms offer users with a completely new level of interaction with information, going away from the keyword search to providing full answers, generating content, or even allowing users to engage in a conversation with the LLM. Search ads, we know them, are on their deathbed.
In a strange (un)coordinated step, almost all LLMs offer a $20 a month paid subscription to the service. To the average user, the value unlocked with a paid subscription is questionable while most LLMs continue offering free access because they need to train on enormous sets of data. Restricting access to paid subscribers would limit the continuous improvement of the LLM model.

ChatGPT CFO said that approximately 75% of OpenAI’s revenues comes from the paid subscription, while Sam Altman, OpenAI’s CEO said that OpenAI is losing money on the $200-per-month subscription tier, because people use it more than OpenAI expected. While subscription models are certainly viable, only a very small percentage of users opt in to pay for a service that is also offered for free. It's highly probable that the majority of revenue for these LLM search engines will ultimately come from advertising. This prediction draws parallels from the evolution of other dominant platforms, most notably: Facebook.
Facebook, initially a free service for users, quickly discovered the immense power of targeted advertising. By meticulously analyzing user data, Facebook developed the second BEST advertising ecosystem in history. 
The ads model is incredibly successful, showing the potential to monetize not part of the users, but all, through the personalization of ads, and granting users choice. As of 2023, Meta’s monthly average revenue per user (ARPU) stands at $3.38 globally.
Only one company ahead of Meta with their advertising product is Google, with their dominance in Search, as well as Video. Google made a smart decision to keep Search monetization with Ads, while YouTube offers ads free, or an expensive subscription to YouTube premium. This tiered approach allows users to choose their preferred experience while ensuring a sustainable revenue stream for Google. Google’s ARPU stands at $5.12 globally.
This can’t be an article about Ads Free, or Subscription revenues, without mentioning Netflix. Netflix, was a strong opponent against advertising in their service, but recently introduced an ads-supported tier as well. This shift acknowledges the growing importance of advertising revenues, even for premium content providers. 

What is the LLM Advertising opportunity?

ChatGPT, currently the leading LLM platform, has approximately 300 million weekly users today. In sharp comparison, Meta, has almost 3.3 billion (with a “b”), and Google, has an estimated reach of over 4 billion online users globally.
Based on the unconfirmed figures shared by OpenAI, the company behind ChatGPT, the company generated approximately $1.6b in revenues during 2023, which would mean that their current monthly ARPU is only $0.67. 
By adding an ad supported tier, if compared with Meta, or Google’s ARPU, OpenAI may have an opportunity to increase revenues by between 16.5x and 24.8x!

I would estimate that OpenAI stands to gain at least $26 billion in annual revenues by offering an ads supported tier, and integrating ads into the native results. I would also bet that ChatGPT will feature ads within the coming year or so!
 

(this is hopefully not how ads will look on ChatGPT 😬)

Why will ads dominate LLM search ? 

• Scalability: Advertising offers unparalleled scalability. As the user base grows, so does the potential ad revenue. Subscriptions, while valuable, have a ceiling – the number of users willing to pay a monthly fee, especially given the subscription fatigue users have been experiencing over the past years. 
• User Acquisition: Free access fueled by advertising is a powerful tool for user acquisition. It allows LLM search engines to reach a massive audience quickly, establishing a strong user base before focusing on premium offerings.
• Data-Driven Refinement: Analyzing user interactions with ads provides valuable data that can be used to refine the advertising experience, making it more relevant and less intrusive for users.

Premium LLM subscriptions are not going away

While adtech is the monetization engine of the internet, users are not fans of ads. If Facebook or instagram would launch in 2024, it would likely offer a subscription model before it would introduce ads. 
The future of LLM search is likely to be a hybrid model, combining the scale of digital advertising with the choice of subscription tiers. By strategically leveraging both revenue streams, LLM search engines can achieve sustainable growth, deliver exceptional user experiences, and continue to push the boundaries of artificial intelligence.
I hope this article provides valuable insights into the potential future of LLM search engines!

A word about LLM ads measurement

Perplexity, which is the first LLM search engine experimenting with ads, introduced a non-intrusive ad format with sponsored related questions.
 

(Ads on Perplexity LLM search engine)

In a blog post explaining the move, Perplexity said that monetizing with ads would allow Perplexity to broaden their publisher program (allowing publishers to implement an LLM search engine feature), but also that the advertising program will never share personal information with advertisers.

A harsh statement such as this, means that Advertisers cannot use legacy measurement technologies tracking users impressions or clicks across the Perplexity website or LLM search. 
Advertisers will need to adopt an AI first approach to measuring LLM search ads performance, using tools such as INCRMNTAL, an AI powered marketing insights platform.  

The LLM advertising opportunity is indeed an exciting new era of advertising and AI. For long, Search and Social were the only two fast growing mediums to advertise. It will be interesting to see what place LLM will take within the advertising pie.