In 2018, blockchain showed up in 275 of the stories we collected at Advertising Perspectives. By 2023 it was ten.
The metaverse peaked at 77 stories in 2022. NFTs peaked at 46 in 2021. Chatbots had their first moment in 2017. Generative AI ran from 153 stories in 2023 to 194 in 2025, and “AI agents” went from six stories in 2024 to 127 the next year.
Every few years, advertising finds a new technology it can’t stop talking about. We have been in this business long enough to have sat through most of these from the inside, and we wanted to see the cycles side by side. So we counted.
The curves
Stories in our archive that mention each term, by year:
| Term | Peak year | Stories at peak | Most recent full year |
|---|---|---|---|
| Augmented reality | 2017 | 268 | 9 (2025) |
| Blockchain | 2018 | 275 | 1 (2025) |
| NFT | 2021 | 46 | 1 (2025) |
| Metaverse | 2022 | 77 | 19 (2025) |
| Web3 | 2022 | 24 | 0 (2025) |
| Generative AI | 2025 | 194 | still climbing |
| AI agents | 2025 | 127 | 45 already in 2026 |
Our archive includes TechCrunch, which covers consumer technology heavily, so the 2017 and 2018 numbers lean toward tech news. The shape holds across the advertising outlets too, and the shape is the point. Each one rises fast, peaks within a year or two, and falls off almost as fast as it rose.
The pattern, in four headlines
Read the coverage of any one of these in order and the same four stages show up, often in nearly the same words.
1. “Brands should be there.” The first wave of stories is about who’s experimenting. In 2016 Digiday explained why fashion brands were gaga for chatbots, and in January 2017 it showed how five brands had approached their Messenger chatbots. In late 2021, it was mainstream brands experimenting with AR and VR as the pressure to compete in the metaverse loomed.
2. The conference takes over. At Dmexco in 2017, Digiday’s awards led with blockchain, buzzwords and booths. At Advertising Week in October 2023, the preview was how AI was expected to dominate, and the recap was that marketers and agency executives were abuzz about generative AI. When a technology owns the stage at a trade show, it is usually at or near its peak.
3. The reality check. Then the tone turns. As early as April 2017, AdExchanger found that Messenger strategies had gained only modest traction with brands. In January 2022, at the height of the metaverse wave, Digiday was already writing that brands’ FOMO was competing with consumers’ burnout. By March, the advice on NFTs was to put utility over headlines, because “consumers don’t care if brands participate.”
4. The fizzle. Finally the coverage says it out loud. In August 2018: blockchain buzz among advertisers starts to fizzle. In January 2023: as metaverse hype subsides, in-game advertising companies go back to their gaming roots. A month later the question was how the recession could impact brands’ metaverse spending, which is a polite way of saying it was about to stop.
What survived each wave
The interesting part isn’t the hype. It’s what’s left when the hype goes away, because something always is.
Chatbots didn’t become a new ad channel. They became customer service, which is less exciting and far more useful, and they were the rough draft for the conversational AI that arrived six years later.
Augmented reality didn’t become its own medium either. It became an ad format. By late 2018, Digiday reported that Snapchat’s AR ads, once $500,000, were now cheap and pulling in smaller advertisers. When the price of a hype technology drops by an order of magnitude, it’s turning into a commodity, which means it’s working.
Blockchain left the least behind in advertising. In 2017 AdExchanger reported a Nasdaq-based exchange planning to bring Wall Street trading to digital media, and as late as 2018 companies were teaming up to support a blockchain ad model. The underlying problems those pitches named, supply-chain transparency and verifiable delivery, were real. They mostly got solved with better logs, standards and contracts, not a distributed ledger.
The metaverse and Web3 left behind gaming. The in-game advertising companies that went back to their roots in 2023 were selling something that existed before the word “metaverse” was popular and still exists now: ads inside the places people actually spend time.
The pattern across all of them: the technology that survives is the one that slots into an existing workflow and makes something cheaper or faster. The one that asks brands to build a presence in a new place and wait for an audience to show up rarely makes it.
Generative AI and agents: is this one different?
Two things are true at once.
The curve looks the same. Generative AI went from one story in 2022 to 153 in 2023. “AI agents” jumped from six to 127 in a single year. If you only looked at the chart, you’d predict a fizzle around 2027.
But the substance looks different from the last few waves, and it fails the hype test in the right ways. Generative AI didn’t ask brands to go anywhere new. It moved into work they were already doing: writing copy, producing variations, building creative, answering questions. Some of the hype has already burned off. By August 2023, Digiday was explaining why virtual influencers had died off despite the AI boom, which is the same reality check every wave gets. What’s left after that check is a lot of ordinary production work done faster.
Agents are earlier. The 2025 headlines are still mostly stage one and two: Mastercard and Visa building AI shopping agents, Adobe launching a suite of agents, and MediaPost arguing that agents acting as brand influencers will collapse the consumer buying journey.
The detail that makes us take it seriously is a different kind of headline: the plumbing. In October 2025, AdExchanger covered the Ad Context Protocol, an attempt to standardize how agents ask for and buy advertising. Two weeks later, MediaPost reported that agentic ad creation would be tested through the Model Context Protocol. Nobody wrote protocol stories about the metaverse. When an industry starts arguing about standards, it’s usually because there’s real work to connect.
How to tell the difference next time
After a decade of watching these curves, here are the questions we ask about any new thing the industry can’t stop talking about.
Does it fit into work people already do? Generative AI writes the copy you were already writing. The metaverse asked you to build a store in a place your customers weren’t.
Does it get cheaper as it spreads? AR ads fell from $500,000 to self-serve. That’s what working technology does. Technology that stays expensive and bespoke is usually still a demo.
Is anyone building the boring parts? Protocols, measurement, standards, integrations. Hype has keynotes. Technology that lasts has documentation.
Who’s asking for it? If the demand comes from conference agendas and agency decks, be patient. If it comes from the people doing the work, pay attention.
By those tests, generative AI is already past the hype and into the plumbing. Agents are in the middle of the climb, with the first real plumbing showing up. The curve will flatten, as every curve in our archive has. The question is what’s left standing when it does, and this time we think it will be a good deal more than a booth.
Every headline in this piece links to the original story. Browse the archive’s coverage of the metaverse, blockchain, generative AI and AI agents.
Kevin Reynolds is the founder of Advertising Perspectives, MadTech Consulting and data4orce.ai. He is a marketing technology architect and C-level advisor, with a focus on Salesforce and data-driven advertising.