Markets Explore

Diffusion of Innovations

The diffusion of innovations describes how an idea spreads, from innovators and early adopters through the majority and on to laggards.

The questionWho adopts this next, and what about the idea itself is speeding up or slowing down adoption?

Geoffrey Moore's version of the adopter curve is called the technology adoption life cycle.

Model · Everett M. Rogers, 1962

Schematic of the five adopter categories on an idealized normal curve, with the standard shares of adopters.

When to use it

A client is launching a product or program and the plan treats "the market" as one group that will respond at one speed.
Adoption was strong among the first enthusiasts and has stalled, and you need to know whether the next group needs something different.
An internal rollout of a new system or process assumes everyone moves on the same schedule.
You are deciding whether to spend on changing the innovation itself, for example making it easier to try, or on pushing it harder.

When to avoid it

You need a number for next year's adoption. The category percentages are not a forecast. Fit a diffusion model to your own data, or collect it first.
The decision is mandated top-down or made by a committee. Individual timing then says little about what happens.
The question is which stage a product's sales are in and what to spend on it. The product life cycle fits that better.

Overview

Diffusion of innovations explains how a new idea, product, or practice spreads through a group over time. Everett Rogers sorted adopters into innovators, early adopters, early majority, late majority, and laggards by when they adopt, and named five attributes of the innovation that speed or slow adoption. A consultant uses it to decide which group to reach next and what to change in the offer.

The percentages are idealized slices of a normal curve, not measured constants, and a category describes one person’s timing on one innovation, not their character. Moore’s Crossing the Chasm extended the model to technology marketing and argued that the step from early adopters to the early majority is where many launches stall. Use the model to find the group not yet adopting and the attribute holding it back.

The adopter categories, and what speeds adoption

Innovators

About 2.5 percent in the standard curve. Adventurous, tolerant of failed bets, and linked to sources outside the local group. Peers often do not treat them as guides, so their adoption does not persuade anyone.

Early adopters

About 13.5 percent. Respected locally, selective, and the group others watch before deciding. Their endorsement carries an idea to the majority, which makes them the group worth studying.

Early majority

About 34 percent. Deliberate. They adopt just ahead of the average member, rarely lead, and want evidence from people like themselves.

Late majority

About 34 percent. Skeptical, and they adopt when the cost of holding out rises or when most of their peers have already moved. They need low risk and little effort.

Laggards

About 16 percent. The last to adopt, often with tighter resources and stronger ties to existing practice. Resisting can be rational for them, because a failed switch costs them more.

Five attributes of the innovation

Relative advantage over what it replaces, compatibility with existing values and practice, complexity, trialability, and observability. All five are judged by the adopter, not by the maker, and they explain much of why two innovations with the same merit spread at different speeds.

What it is not

A forecast. The 2.5, 13.5, 34, 34, and 16 percent shares are idealized slices of an assumed normal curve, cut at one and two standard deviations from the average adoption time. They are not measured constants, and real adoption curves vary in shape.
A personality typing. A person who is an early adopter of a phone can be a laggard on a new accounting method. Category describes timing for one innovation.
A value judgment. Diffusion research has been criticized for a pro-innovation bias, and Rogers acknowledged the problem. Late adoption can be the right decision.

The steps

Define the innovation and the adopter

Decide the unit of adoption (a person, a team, a firm), the behavior that counts as adoption, and the population. A purchased license is not adoption if nobody uses it.
Place the people who have adopted

Pull usage data and sort adopters by when they started. Note who they sit near and who watched them. Use your own timing, not the textbook percentages, as the reference.
Rate the five attributes through the next group's eyes

Interview people who have not adopted and ask what stops them. Score relative advantage, compatibility, complexity, trialability, and observability from their answers, not from the team that built the thing. The weakest attribute is your first target.
Choose the move for the next group

Low compatibility means changing the fit with existing work. High complexity means simplifying. Weak trialability means a pilot or trial with an exit. Weak observability means making use visible, through respected people. If relative advantage is the weak one, the offer may be wrong for this group. In technology markets, test whether early adopters' references actually persuade the majority, since Moore argued that they often do not.
Track by group and revise

Measure adoption monthly by segment. If the pattern does not look like a bell, stop using the category labels and describe the groups by what they said and did.

In practice

Six months after launch, a new case-management tool at an insurer had stalled at roughly a fifth of claims handlers. The rollout plan assumed the rest would follow once the first users spoke up. Interviews showed why they had not. The early users were enthusiasts who had built their own workarounds, and nobody else could see how the tool fit into the handoff to the payments team, which it did not. Nobody the others respected was using it where they could watch.

Mapped to the five attributes, compatibility and observability were the weak ones, with trialability close behind. The changes were specific: rework the handoff template, have two respected senior handlers run their own claims on the tool in the team stand-up, and offer a two-week trial with a way back to the old process. The enthusiasts needed none of this. The next group needed all of it.

What you leave with

A definition of adoption specific enough to count.
A map of who has adopted, in what order, and who influenced whom.
A score on the five attributes as the next group sees them, and the weakest one named.
A plan for the next group: who to reach, through whom, and what to change in the offer.

Failure points

Assuming the first group predicts the second

Early adopters tolerate rough edges and will work around gaps. A pilot full of them produces optimism that the majority does not share, because the majority wants proof, support, and low effort.

Using the percentages as a plan

"We have 16 percent and need 34" treats the cut points as targets. They describe an idealized curve. A real rollout can flatten, stop, or never reach the majority.

Blaming late adopters

When people do not adopt, teams call them resistant. Often the innovation does not fit their work, costs them more, or fails in a way the early users never met. The refusal is information about the product.

Rating the attributes from the builders' side

The team that built it scores it easy to learn and clearly better. Those scores describe the team. The attributes only count as the adopter perceives them.

Counting acquisition as adoption

Licenses sold, accounts opened, and training attended can all rise while use stays flat. A dashboard built on those numbers shows a curve that is not happening.

The failures above are usually Framing: the wrong problem was chosen before this framework was applied.

Origin

Everett Rogers, then a rural sociologist at Ohio State University, published Diffusion of Innovations in 1962. The book pulled together hundreds of earlier studies from rural sociology, medical sociology, education, and other fields. The field had grown out of midwestern farm research in the 1920s and 1930s, and the study most often cited as setting its pattern is Bryce Ryan and Neal Gross's work on how Iowa farmers took up hybrid seed corn, published in 1943. Later editions of Rogers's book renamed the stages an individual passes through, from awareness, interest, evaluation, trial, and adoption to knowledge, persuasion, decision, implementation, and confirmation.

The five adopter categories are defined by timing against the average adoption time on a normal curve: innovators, early adopters, early majority, late majority, and laggards. Geoffrey Moore's Crossing the Chasm (1991) applied the curve to high-technology marketing and argued that the step between early adopters and the early majority is where many launches stall.