AI Makes Chat the New Battleground for Ad Platforms
The meteoric rise of ChatGPT to over 100m users in mere months has ignited fierce competition among ad platforms, struggling to adapt to this rapidly changing landscape. The focus is primarily on whether Google can reclaim its early advantage in AI, before OpenAI and Microsoft disrupt their core search, but this isn’t the only field of contention.
Microsoft Advertising’s introduction of a Chat API for publishers encroaches on Google’s profitable DoubleClick revenue stream. By offering monetization options for publishers building chatbots, Microsoft not only fuels demand for publishers using the GPT-4 API from Microsoft-affiliated OpenAI, but also broadens the reach of its ad network. While Bing's AI chat experience might not displace Google as the default search engine, it could potentially outstrip Google as the default ad network for monetizing AI chat experiences.
Simultaneously, Meta (formerly Facebook) revealed that click-to-message ads have reached a $10 billion run rate, and their paid business messaging service on WhatsApp is expanding at 40% per quarter. With over 2 billion active users on WhatsApp, a platform that has traditionally been under-monetized, Zuckerberg identifies an opportunity to leverage AI for increased profitability: “I anticipate a surge in interest in AI agents for business messaging and customer support once we perfect the experience,” he said.
Chat as a medium differs fundamentally from traditional ads, hence, we should anticipate a substantial variation in the economics of advertising via this new channel.
The Engagement Factor in Advertising
Advertisers' interest in AI chat extends beyond its vast user reach; it also lies in its potential to deliver a more targeted and personalized experience. AI chat allows businesses to interact with customers one-on-one, gaining insights into their preferences and habits. The capacity to correct or offer inline feedback to AI responses or to pose follow-up questions can enhance the user experience, as demonstrated by ChatGPT.
My experience with click-to-message ads suggests that a well-optimized messaging funnel can cut engagement costs in half compared to traditional ads. For instance, we found that personalizing the initial message a user receives to align with the ad they viewed, doubled our click-to-message conversion rate when developing a FinTech chatbot product. This was achieved with a less sophisticated chatbot (pre-ChatGPT), suggesting that even greater results could be achieved with today's technology.
Potential for Quicker Ad Fatigue
The messaging inbox is a highly personal space where users communicate with their friends, family, and partners. Brands that interrupt this space with commercial messages can face backlash if not done tastefully. Users can quickly tire of seeing the same ad repeatedly in a confined space, and may be more irritable when interrupted while reading messages compared to casually browsing the timeline.
In my experience running ads on Facebook Messenger, this format incurs a higher number of users marking the message as spam, and ad fatigue sets in more quickly. I've observed a greater overall backlash with click-to-message ads, with more users marking the ad as spam or leaving an irate message upon realizing they've been directed to a chat experience.
strong>User Experience Challenges
For many products, the chat experience can be slower than using a website, which is why chat is not the dominant web interface. We are still in the pre-Apple phase of AI, where computers started as terminals until Apple popularized the graphical user interface. Asia is likely to lead the way in this area, as the use of WeChat messaging is prevalent for a wider range of daily tasks. Brands on WeChat often incorporate dynamic menus and 'mini programs' within the app, enhancing the chat experience by making it more interactive and user-friendly.
The Potential of Closed-loop Measurement
The integration of AI chat with other services presents exciting possibilities for cross-platform advertising. For instance, a customer might initiate a conversation with a business on WhatsApp, continue it via email, and finalize it on the company's website. Advertisers can track these interactions across platforms, given a user is logged in, and utilize this information to create a comprehensive picture of a customer's journey. This level of tracking, previously unattainable, could revolutionize the advertising industry.
However, the way chat ads are measured will largely depend on their format. If the primary method of monetizing AI chat is to serve sponsored recommendations when the user asks a commercially relevant question, the economics might resemble Google Ads today: limited volume but highly valuable real estate. If that’s the case then all the same measurement issues we face with Google Ads today will be relevant. Last-click attribution models will be biased towards crediting AI chat, and proper incrementality testing will be needed to determine how many of those credits sales were actually deserved.
User Privacy Concerns
Not all consumers may be comfortable with the idea of their conversations being used for ad targeting. They’re likely to regard conversations with AI as highly confidential, similar to their internet browsing history. Companies will need to balance their aim for more effective advertising with the necessity to respect user privacy and preferences.
For instance, Snap faced backlash after a failed attempt to integrate AI into the chat experience, which left users worried that the AI knew their location even with location sharing turned off. Italy also temporarily banned ChatGPT over concerns regarding the use of personal user information in the training data.
Despite these challenges, the potential benefits of AI chat advertising are too compelling to disregard. As AI technology continues to evolve exponentially, we can expect more businesses to adopt this form of advertising. Whether Google, Microsoft, or a new contender will emerge as the dominant force in this space is still uncertain. However, one thing is clear: the landscape of digital advertising is undergoing a transformation, but measurement issues are perennial..