Dec 3, 2024 5 min read

ChatGPT Ads, Perplexity Agents, SMB Expectations

Dialog: Exchange Number 10

See our special "cyberweek" membership offer.

As ChatGPT Turns Two, Ads Await

Roughly a week ago, ChatGPT turned two. OpenAI was founded in 2015 but its breakthrough product launched in November, 2022. In two short years, ChatGPT has become a credible challenger to Google.

One of the things OpenAI CEO Sam Altman said in a podcast before the launch of SearchGPT is that he wanted to "reinvent" the search experience. "I don’t think the world needs another copy of Google," Altman remarked. He also said that he "hates ads" and prefers that people pay to access services because that avoids conflicts or editorial bias. But he did acknowledge it might be possible to create a better ad model for an LLM-search hybrid.

OpenAI CFO Sarah Friar, formerly the CEO of Nextdoor, told the Financial Times this week that OpenAI was considering introducing ads, but said there were "no active plans to pursue advertising." Yet the company has been hiring advertising executives from Google and Meta. So it's pretty clear ads are coming; it's only a question of timing. The company is following in the footsteps of its smaller rival Perplexity, which has begun piloting advertising – out of necessity.

OpenAI is burning through cash, and while the company is predicting a dramatic increase in users and revenue in 2025 ($11.6 billion), ads will probably be necessary to supplement paid subscriptions and enterprise revenues.

Source: Dialog Local Consumer Tracking Survey (2024)

According to our research, one of several features of AI-based search that consumers like is the absence of advertising. And while it's not the main factor they appreciate, it's a significant differentiator vs. Google. So there is some risk in introducing ads – depending on the implementation and user experience.

News & Noteworthy

Optimization for AI: being a trusted reference, tracking brand mentions.
Why Gen-AI is a bridge to the post-smartphone future.
Hume + Claude: Is the "voice-first" AI future starting to emerge?
Enterprises rhetorically bullish on AI but not ready to implement.
Microsoft denies it's training LLMs on Office 365 user data.
Google explains how Gemini is being deployed by retailers.
ChatGPT is showing inaccurate citations of news publisher content.
World Labs can now turn a still image into an interactive "3D world."
ElevenLabs' new GenFM podcast creator, mimics NotebookLM.
Former Googler Hugo Barra building an OS for AI agents.
AI content labeling draws skepticism, negative user reactions.
The AI boom is just getting started.

Perplexity Shopping: Agents in Action

Speaking of Perplexity, the company recently introduced a shopping experience for Pro subscribers. That's noteworthy in itself but it's really the first exposure, on the consumer side, to what is likely to become much more common in 2025: agents.

Next year may well be "the year of agents," for both B2B workflows and online consumers. We don't know all the ways they'll show up and be "productized." But the basic idea is that AI-based agents will automate and execute steps or processes that previously were manual. In a consumer-shopping context that means collapsing the funnel and the multiple steps that consumers might otherwise perform on their own – and radically reducing the time involved.

"Best Noise Cancelling Headphones"

The Perplexity shopping experience is a kind of hybrid between current product search and agents; the agents part isn't as obvious. But Perplexity Shopping locates relevant products, compares them in the background and then offers immediate buying options through the site. This can significantly accelerate research and purchase decision-making. And there's no reason this can't extend to (local) services.

More agent powered experiences will become available to consumers in 2025, which may include transactions: book a ticket, make a reservation, buy a product at a specific price (not unlike automated stock market trades). The implications for merchants, retailers, publishers and advertisers are potentially significant if not obvious. It's all about data, not your website. While we don't know exactly how all this will all play out, Google and Amazon will be impacted. And they of course will evolve in response.

Digital Agencies, SMB Expectations

One of the things that our recent agency survey revealed is that many digital agencies are worried about AI-savvy competitors and AI DIY tools displacing them. They cited a number of risks and concerns:

AI-savvy agencies might outcompete older agencies
AI could devalue services and impact pricing power
Could enable clients to bring more services in house
Challenge of keeping pace with client expectations in a dynamic market

The high-level response to all this is: offer better and more service to clients. Small businesses (SMBs) represent the largest client segments for smaller agencies but surprisingly also for the larger agencies (200+ headcount) in our sample. Yet most of these agencies aren't delivering against the full range of SMB expectations.

Most SMBs don't use agencies or third party vendors to help with marketing. But for those that do, there are a broader set of needs and expectations that agencies are either not aware of or failing to address.

SMBs working with agencies want to see marketing ROI of course. But they also want to better understand customers, improve customer service, get technology and software advice, better understand their own businesses and competitors' and want help with overall strategy. These are things outside the scope of most current agency-client relationships.

Moving into these more strategic areas and offering a broader range of services will be important for both customer acquisition and retention as a new generation of simplified AI-based marketing tools puts increasing pressure on business as usual.

Dialog: Year in Review, Outlook 2025

Join us for the Dialog's year in review and 2025 outlook webinar on Wednesday, December 18. You'll receive a separate email invitation with more details this week. Neal Polachek and I will look back at our data and "what we've learned" in 2024 and then ahead to next year, which is certain to be even more eventful. Mark your calendars.