In 2013 Google began selling Glass, a tiny computer with a small screen above the right eye, to early buyers it called Explorers. The price was $1,500. A year earlier, thousands of people at Google’s I/O conference had watched skydivers wearing it drop toward the roof of San Francisco’s Moscone Center.
First impressions were glowing. Time named Glass one of the best products of the year. The tech blogger Robert Scoble said he could not imagine a day without it, and was photographed wearing it in the shower.
Then the excitement wore off. Glass was buggy, it felt awkward, and it had no clear function. People near a wearer worried about being filmed, and some bars and restaurants banned it. In January 2015 Google said it would stop taking orders and close the Explorer program.
But in a tractor factory in Jackson, Minnesota, a single Explorer unit bought in 2013 had already started something else.
Two markets that don’t talk
Geoffrey Moore, who advises technology companies, gave this pattern a name in his 1991 book Crossing the Chasm. Early fans and the cautious majority buy for different reasons, so a gap can open between them.
Moore built on Everett Rogers’s adopter groups (the curve itself is on the diffusion page) and split them into two markets. The early market holds the technology enthusiasts, who love the technology itself, and the visionaries, who see a chance to remake how things work. Neither group, Moore said in a 2017 podcast interview, looks to peers for guidance.
The mainstream starts with the pragmatists. “The pragmatists in particular use peer referencing a lot,” Moore said. They want a pilot, data and references, and in his telling they want to buy the same thing their competitor bought, while early buyers want something different. That, in Moore’s account, is the gap: mainstream buyers tend to trust people who think like them, so the early fans can’t vouch for the product to the people who come next.
A new restaurant packed with food bloggers in its first week tells a family very little. The family asks the neighbors.
Moore’s prescription is to stop chasing everyone and win one niche first, a beachhead, made of “pragmatists in pain,” people whose problem is bad enough to take a risk. And you have to bring what he calls the whole product, everything the buyer needs to make it work, because pragmatists won’t meet you halfway the way early buyers do.
Glass’s second act, as Wired told it, has that shape, though nobody at Google is quoted invoking Moore. The factory had a painful problem: workers walked about 50 feet to laptops to check each custom-built tractor’s specifications. Google’s workplace version came with clip-on safety frames, stronger security and partners who sold complete hardware and software packages.
A consultant’s map, not a law
The chasm came from Moore’s experience in high tech. In 2006 the marketing researchers Jacob Goldenberg, Barak Libai, Eitan Muller and Renana Peres wrote that his theory “is not backed up by organized empirical data, but rather provides only a few anecdotal examples.” It also cuts against classic diffusion theory, which pictures communication flowing steadily from early adopters to everyone else.
When researchers went looking, they found something real but not universal. In a 2002 study of new consumer electronics, Goldenberg, Libai and Muller found that a third to a half of sales histories rose, sagged for a sustained stretch, then climbed past the first peak. They called it a saddle. In their simulations, the less the early and mainstream buyers talked to each other, the more often it appeared.
So the gap shows up often enough to plan for, but not always: of the seven product categories in the 2006 study, cell phones showed no clear saddle. Nor is a beachhead a guarantee. Google stopped selling Glass Enterprise in 2023.
The pilot that never spreads
The chasm can open inside a company too. Imagine an operations manager whose two most curious planners have rebuilt the weekly scheduling in a new automation tool. They demo it at every team meeting. Everyone else nods and keeps the spreadsheet.
More demos from the same two people won’t help much, if Moore is right, because pragmatists don’t take their cues from enthusiasts. In his account, what moves them is a colleague with the same workload who has used it for real, and a whole product: templates, IT sign-off, someone to call when it breaks.
The tractor factory took a blunter route: following the on-screen instructions became required, “like a torque tool,” one manager said.
Leaders first, frontline later
AI tools at work show a similar split. In Boston Consulting Group’s 2025 survey of more than 10,600 white-collar employees in 11 countries and regions, more than three-quarters of leaders and managers said they used generative AI several times a week. Regular use among frontline employees had stalled at 51%. With strong leadership support, BCG found, the share of employees who felt positive about it rose from 15% to 55%. Only about a quarter of frontline staff said they had that support.
A year later BCG reported a break: 74% of frontline employees were regular users, driven mainly by older workers, people in operational roles and lagging markets.
These are self-reported surveys by a consulting firm that advises companies on AI. Leaders versus frontline staff is not the same split as enthusiasts versus pragmatists, the leadership figure is a correlation, and none of it tests Moore’s theory. It does show the shape he described: frontline use stuck around half for a few years while most of their bosses had moved ahead.