Where ChatGPT Ads Actually Rank in Digital Advertising History
Why the infrastructure you ignore is always more important than the interface you obsess over
The same conversation keeps playing out. Someone asks if ChatGPT ads are the most important development in digital advertising, and immediately the room divides into two camps: the true believers who think AI will eat everything, and the grizzled veterans who’ve seen this movie before and aren’t impressed.
The answer is somewhere in between, which is probably where most reasonable people should land.
But here’s what’s missing from nearly every discussion: context. Real context. The kind you only get from understanding how we actually got here, not just the highlight reel of innovations everyone already knows about.
So let’s walk through where agentic advertising—which is what ChatGPT ads really represent—fits into the actual story of digital advertising. Because the ranking matters less than understanding why certain innovations compound while others just iterate.
1995: Two Guys in a Basement Built the Plumbing
Summer of 1995. Kevin O’Connor and Dwight Merriman are sitting in O’Connor’s basement in Alpharetta, Georgia, building what will eventually become DoubleClick.
The ad server at that time was an ISDN line and a 486 PC on Dwight’s desk with the cover off, because it kept overheating. O’Connor was learning about advertising from textbooks while Dwight coded. They were going to the library to read Ad Age and Adweek to see if they had any competitors.
This isn’t a polished origin story. This is two engineers solving a basic infrastructure problem: how do you deliver ads to multiple websites without doing it manually, one at a time?
They created a system to display banner ads across a network of websites and track their performance to better target internet users. The product caught on. Google would eventually acquire DoubleClick in 2007 for $3.1 billion, and those systems would quietly power digital advertising for the next two decades.
Here’s the critical insight most people miss: without ad servers, none of the other innovations happen. Not search ads. Not programmatic. Not real-time bidding. Not any of it. This was foundational infrastructure.
You can’t build a house without plumbing. DoubleClick was the plumbing. Everything else was just deciding what rooms to build and how to decorate them.
2002: When Google Made the Economics Work
The prevailing myth is that Google invented pay-per-click advertising. The reality is messier and more interesting.
Overture, founded in 1998 as GoTo.com, was the pioneer of the pay-per-click (PPC) advertising model. Jeffrey Brewer presented the concept at TED in 1998, and it was genuinely innovative—advertisers bid for placement in search results based on keywords, paying only when someone clicked.
But innovation without execution is just an interesting idea. Google did something smarter. In February 2002, Google introduced a new version of AdWords that adopted Overture’s pay-per-click auction model, two years after launching their first ad program which had sold ads on a CPM basis.
The brilliance wasn’t just adopting CPC. It was what Google layered on top: Quality Score. Google introduced the Quality Score, a metric that determined ad placement not just on the bid amount but also on the relevance of the ad to the search query.
This fundamentally rewired the incentive structure. You couldn’t just buy your way to the top anymore. Your ads had to be relevant. Your landing pages had to convert. Your click-through rates mattered. Small businesses could suddenly compete with enterprise budgets if they were smarter about targeting and creative.
Performance marketing as we know it started here. Everything became measurable. Everything became optimizable. And billions of dollars started flowing into a channel that rewarded people who were good at math and testing over people who were good at relationships and schmoozing.
This was the moment digital advertising became a real industry, not just a curiosity for early adopters.
2005: The Exchange That Created Liquidity
The Right Media Exchange launched officially April 1, 2005, and scaled significantly in the following 18 months. Most people don’t remember this moment. They should.
Before Right Media, buying and selling display advertising was manual, inefficient, and relationship-driven. Negotiating deals, managing insertion orders, everything took forever. The friction in the system was enormous.
Right Media created a true marketplace. The idea being that an ad exchange would be a marketplace for buyers to seamlessly connect with sellers to buy their inventory on an auction basis. But the real technology innovation wasn’t just matching buyers and sellers—it was the ability to set buying rules which pick only the ad impressions that matched the target buying criteria whilst rejecting those that didn’t.
Think about what this enabled. You could suddenly buy individual impressions based on data, in real-time, at scale. By combining three separate pools of supply and demand, every campaign could bid on more inventory, giving it more chances to find the right user. Every publisher was getting three times the demand. And the networks tripled the effective size of their networks.
The launch of the ad exchange did for display what keyword bidding did for search—it created liquidity. It’s a great equalizer, allowing start-ups and small companies to get as fair a chance in bidding for media as the biggest advertisers.
Yahoo acquired Right Media in 2007 for $850 million, and programmatic advertising was born. The impact is hard to overstate. This changed who could participate in digital advertising and how efficiently capital could flow through the ecosystem.
Here’s the deeper pattern: ad servers made digital advertising possible, CPC made it economically viable, and ad exchanges made it liquid. These three innovations created the operating system that everything else runs on.
2007: Facebook Proves Social Data Is Different
November 2007. Mark Zuckerberg stood before an audience of corporate bigwigs and representatives of the Madison Avenue agency establishment to announce what was arguably the start of social media advertising.
The unveiling of Facebook Ads wasn’t just about creating another ad platform. Facebook launched its Facebook Ads platform, part of which includes Beacon and Facebook Marketplace, along with Pages for brands and businesses. They were giving businesses the ability to create branded pages and run targeted ads based on an entirely new kind of data—social data.
What made Facebook different wasn’t the ad format. It was the targeting. For the first time, advertisers could target based on interests, behaviors, and connections that people voluntarily shared. Not “women 25-34” but “women 25-34 who like yoga and recently got engaged and live in Austin.”
The precision was unprecedented. The scale was massive. And the results were undeniable. In February 2015, Facebook announced that it had reached the milestone of attracting more than two million active advertisers to its site.
Facebook proved that social networks could be advertising platforms. Instagram, Twitter, LinkedIn, TikTok—they all followed the playbook Facebook wrote. But more importantly, Facebook demonstrated that different types of data create different types of value. Social data wasn’t just another targeting variable—it was a fundamentally different substrate for building an ads business.
2012: Amazon Captures the Moment of Intent
While everyone was focused on Facebook’s social graph, Amazon was quietly building something different. In 2012, Amazon launched Amazon Marketing Services (AMS) exclusively for first-party vendors.
This was different from Google and Facebook because of one critical advantage: purchase intent. Amazon knows what you searched for last week and what items you’ve been viewing and ultimately what you purchased. Google knows what you’re interested in. Facebook knows who you are. Amazon knows what you’re about to buy.
In 2015, Amazon launched Amazon Marketing Services (AMS), a suite of cost-per-click advertising solutions. AMS offered three primary ad formats: Sponsored Products, Headline Search Ads (later renamed as Sponsored Brands), and Product Display Ads.
The impact took a while to materialize, but when it did, it was enormous. Amazon went from a standing start in 2012 to becoming the third-largest digital advertising platform behind Google and Facebook. They did it by sitting closer to the point of purchase than anyone else.
The lesson here: proximity to transaction matters more than scale of attention. Amazon had fewer users than Facebook, less search volume than Google, but they owned the moment when people were ready to buy. That turned out to be worth more than anyone expected.
2021: When Apple Forced the Industry to Grow Up
The ATT privacy framework was introduced for all Apple devices after the release of iOS 14 (and enforced after iOS 14.5) to limit the amount of user data app developers can share with other companies.
April 26, 2021. The sheer panic in the industry was palpable. Facebook was losing their minds. Every mobile advertiser was scrambling. The Financial Times estimated ATT cost the major tech platforms including Facebook, YouTube, Snapchat and Twitter a combined $9.85 billion in advertising revenue in 2021.
With ATT, app users had to opt-in to data tracking via a popup. Because most users opted out, this created a massive challenge for advertisers who had built their entire measurement stack on device-level tracking. The apocalypse everyone predicted seemed imminent.
But here’s what actually happened: the industry adapted. Despite these setbacks, the mobile advertising industry has not only recovered but thrived in the post-ATT landscape. eMarketer predicts in-app video ads will capture over 30% of total US mobile advertising spend for the first time this year.
New signals were found. Better models were built. First-party data strategies accelerated. The predicted apocalypse never came.
What ATT really did was accelerate a shift that was already happening—moving away from surveillance-based advertising toward more privacy-conscious approaches. It also established a precedent: platforms will unilaterally change the rules when they feel pressure to do so, and the industry will adapt faster than anyone expects.
2025: Agentic AI and the Shift from Influence to Instruction
Which brings us to now. OpenAI appears to forecast a billion dollars in new revenue from “free user monetization” in 2026. That figure is forecast to grow to nearly $25 billion by 2029.
The scale is already there. ChatGPT.com recorded around 5.2 billion monthly visits, with 601.5 million unique visitors. OpenAI is hiring aggressively, bringing in people like Fidji Simo, former CEO of Instacart and head of the Facebook app, to build out their ads infrastructure.
But scale isn’t what makes this interesting. What makes this different is the nature of the interaction—and more importantly, what happens after the interaction.
When you search on Google, you’re looking for information. When you scroll Instagram, you’re passing time. When you ask ChatGPT a question, you’re having a conversation with something that understands context, remembers your preferences, and increasingly can take action on your behalf.
McKinsey’s Technology Trends Outlook 2025 report identified agentic AI—artificial intelligence systems capable of autonomous planning and execution—as the most significant emerging trend for marketing organizations.
Think about what that means. An AI that doesn’t just show you an ad, but negotiates on your behalf. That doesn’t just recommend a product, but completes the transaction. That doesn’t just inform, but acts.
This is fundamentally different from anything that came before. Previous innovations changed how ads were delivered, how they were priced, or how they were targeted. Agentic advertising changes who makes the buying decision.
It’s not interruptive. It’s assistive. The AI becomes a proxy, making decisions theoretically in the user’s interest while brands pay to influence those decisions. The shift is from persuading people to instructing machines.
The Actual Hierarchy of What Matters
Here’s the framework that makes sense:
Tier 1: The Infrastructure (Enables Everything)
Ad servers (1995) - Nothing else happens without this
CPC pricing (2002) - Made the economics work at scale
Ad exchanges (2005) - Created true programmatic liquidity
Tier 2: New Substrates (New Types of Value)
Facebook ads (2007) - Proved social data was a different substrate
Amazon ads (2012) - Captured purchase intent at the point of decision
Tier 3: Shifts and Constraints (Force Evolution)
ATT (2021) - Forced privacy-first innovation
ChatGPT ads (2025-2026) - New decision-making paradigm
Notice what’s in Tier 1. Infrastructure. The stuff that’s boring but essential. The plumbing that everything else is built on top of.
ChatGPT ads are important—very important—but they’re an interface innovation, not an infrastructural one. They’re changing how users interact with commercial intent, not creating the underlying systems that make digital advertising possible.
The more interesting question isn’t where ChatGPT ads rank. It’s whether agentic AI will eventually create its own infrastructure layer that future innovations build on top of. That’s the bet OpenAI is really making.
What Actually Matters for the Next Five Years
Here’s where the strategic thinking needs to focus:
AI media will become its own budget line within the next few years, just like retail media and social media before it. Someone will ask what percentage of budget goes to AI media and teams will need to have an answer. The smart money is preparing for that conversation now.
The brands that win in an agentic world won’t be the ones with the biggest ad budgets. They’ll be the ones that figure out how to be algorithmically discoverable and defensible when an AI is comparison shopping on behalf of a user.
It’s more likely that they will pursue a take-rate based model rather than inserting a paid feature into the query. This changes unit economics fundamentally—paying for outcomes, not impressions. That’s actually closer to how affiliate marketing works than how display advertising works.
The product needs a clear value proposition that an AI can articulate without human intervention. Pricing needs to be competitive in a world where comparison happens instantly and comprehensively. Data needs to be structured in a way that AIs can consume and evaluate it.
And experimentation needs to start now. OpenAI posted a job listing on September 24, 2025, seeking an engineer to develop internal marketing technology infrastructure including campaign management and attribution systems. They’re building real infrastructure. This isn’t vaporware.
The Conclusion No One Wants to Hear
Look, infrastructure always matters more in the long run. The ad server, CPC pricing, and ad exchanges were foundational innovations that made modern digital advertising possible. ChatGPT ads are an interface innovation that changes how consumers interact with commercial intent. Both matter, but one enables everything else.
The more uncomfortable truth is that most people overweight recency and underweight foundations. The newest thing always feels more important than the invisible systems that made it possible. But the invisible systems are usually what compound over decades while the shiny new interfaces get replaced by the next shiny thing.
That said, if you’re trying to figure out what to learn, where to test, and how to position for the next five years, agentic advertising is where the action is. The infrastructure is mostly built. Now we’re figuring out what to do with it.
The playbook isn’t written. The winners haven’t won. The rules haven’t solidified.
That’s what makes this moment interesting. Not because it’s the biggest thing that’s ever happened, but because there’s still time to be early. And in this industry, being early to the right thing is worth more than being loud about everything.