How innovations spread: diffusion of innovations explained

Diffusion of innovations is the study of how a new idea, practice or product spreads through a group of people or organisations. Everett Rogers, whose textbook drew the field together, described diffusion as the process by which an innovation is communicated through certain channels over time among the members of a social system. Each part raises a practical question: what is new, who hears from whom, how long deciding takes, and which relationships shape it.

The theory’s most useful move is a change of vantage point. People do not adopt an innovation as its promoter describes it. They adopt, adapt or reject it as they experience it. Is it better than what they do now? Does it fit their routines? Can they try it safely? Can they see it working for someone like them? Sound ideas can spread slowly for exactly these reasons, and weak ideas sometimes spread quickly.

This guide explains the core of Rogers’ framework, what research on organisations adds and where the familiar labels mislead. Throughout, we follow a fictional reusable-cup deposit scheme being introduced across independent cafés in a mid-sized city. Customers pay a small deposit for a cup, return it to any participating café and get the deposit back.

Four elements: the idea, the channels, time and the social system

An innovation does not have to be new to the world. Sahin’s detailed review of Rogers’ theory quotes his definition: an idea or practice perceived as new by the individual or other unit adopting it. For a café that has only ever served takeaway drinks in disposable cups, a deposit scheme is an innovation, even though reusable cups are hardly an invention. Our guide to innovation versus invention draws the same line between creating something and bringing it into use.

Channels range from mass media to conversation. Time appears in each person’s decision, in how early they adopt compared with others, and in the rate of adoption overall. The social system here comprises café owners, baristas, customers, the scheme organiser and the city waste department.

The study that set the pattern for diffusion research makes this concrete. Rogers, Singhal and Quinlan’s history of diffusion research traces that pattern to Bryce Ryan and Neal Gross’s 1943 study of hybrid seed corn among Iowa farmers. Farmers named salespeople as their main first source of information, but neighbours and friends as what convinced them. Despite higher yields, widespread use took about twelve years. Adopting also meant giving up an old practice: choosing next year’s seed from the best-looking ears.

Plotted cumulatively over time, innovations that do spread often trace an S-shaped curve: a slow start, acceleration as adopters influence peers, then a levelling off. The curve describes what happened; it is not a forecast. Diffusion is also where innovation work meets the change work of putting something into use. An idea has not spread until people have changed what they do.

Five attributes, judged from the adopter’s side

A 2005 paper Rogers co-wrote with Medina, Rivera and Wiley states the core finding plainly: innovations perceived as relatively advantageous, compatible, easy to understand, observable and divisible for trial are adopted more rapidly. The five perceived attributes are:

• Relative advantage: how far it seems better than what it replaces, in cost, convenience, status or whatever matters to the adopter.
• Compatibility: how well it fits existing values, experience and needs, and in practice, equipment and routines.
• Complexity: how difficult it seems to understand and use.
• Trialability: whether it can be tried on a limited basis first.
• Observability: whether its use and results are visible to others.

The key word is “perceived”. Greenhalgh and colleagues’ systematic review stresses that these are not fixed properties. It is the interaction among the innovation, the intended adopters and a particular context that determines the rate of adoption.

In the cup scheme, the organiser sees the advantage as fewer disposable cups in public bins. A café owner asks about margins, queues and counter space. A customer asks whether the deposit is worth the bother. A barista asks what happens when a dozen dirty cups come back mid-rush. The scheme also combines a product, a process and a business-model change: a cup, a washing and refund routine, and a deposit arrangement. Each part can be perceived differently.

Timing matters as well. Rogers argued that preventive innovations diffuse rather slowly, partly because the rewards of adopting are delayed, and the 2005 paper adds that poorly observable innovations, including many prevention interventions, diffuse more slowly. The cup scheme has a similar shape, in our reading: a café pays for extra washing this morning, while the benefit of fewer cups in public bins comes later and is shared by everyone.

The attributes are not equally well supported. Tornatzky and Klein’s meta-analysis of 75 articles found that compatibility, relative advantage and complexity had the most consistent relationship with adoption. Dearing, citing Rogers, calls trialability and observability less consistently important, except perhaps for risky or complex innovations.

Turn the attributes into questions for adopters

In our invented example, the scheme is run by a small social enterprise whose coordinator, Lena, planned a straightforward launch: an information pack for every independent café, a press release from the waste department and a sign-up form. Before committing, she ran a two-week pilot with six cafés. Afterwards she used the attributes as an interview checklist, asking owners and baristas to walk through a recent busy morning rather than whether they liked the scheme. Our guide to customer interviews that go beyond polite agreement explains why.

The table records what the fictional cafés reported and what Lena changed. It is an illustration, not research data.

Attribute	Question to ask adopters	What the cafés found	Design change
Relative advantage	On a busy morning, what does this cost or save you?	Cash refunds slowed the queue	Refund as credit on the next drink
Compatibility	Where does this fit your equipment and routines?	Small glasswashers could not keep up at peak times	Twice-weekly collection of dirty cups
Complexity	What would a new barista need to know?	Staff were unsure about cups from other cafés	Any cup accepted anywhere; one-page counter card
Trialability	What would let you try this without committing?	Owners wanted a way out	Two-week trial, no fee, no penalty for leaving
Observability	Where would people see this working?	Cups vanished into bags; customers did not know where to return them	Bright cup colour, window stickers, a visible collection bike

Dearing’s review of intervention design separates an experimental demonstration, a field test for collecting data, from an exemplary demonstration, a persuasive event meant to influence adoption. A flaw is useful evidence in a test but damaging noise in a showcase. Lena kept the pilot experimental, so a café’s complaint counted as evidence rather than bad publicity. As in any concept test, she agreed its boundaries in advance; our guide to what a prototype, MVP or pilot can show explains the limits of such evidence.

Some of the best changes came from the cafés. Rogers calls this re-invention: users changing an innovation as they adopt it. Sam, a barista at one pilot café, started handing out a wooden token for the next drink instead of cash, and Lena made a version of it standard. Dearing recommends being clear about which parts of an innovation are central and which can change. Here the central rule was that any participating café accepts any cup and honours the deposit. How the refund is given could vary.

A decision unfolds in stages

Rogers describes an innovation-decision process with five stages: knowledge, persuasion, decision, implementation and confirmation. Sahin’s summary notes that mass media and outside sources matter most for first knowledge, while interpersonal and local channels matter more for persuasion. Rejection is possible at every stage. In organisations, Greenhalgh’s review found the process messier, often moving back and forth.

A café owner might first read about the scheme in the waste department’s newsletter, then be persuaded, or not, by another owner. Implementation is where uncertainty becomes practical: where the crate goes, who counts tokens, what to tell a confused customer. At confirmation, the owner looks for signs that the decision was right, and conflicting messages can reverse it.

Rogers calls that reversal discontinuance. It may be replacement, where something better takes over, or disenchantment, where the adopter is dissatisfied. In the pilot’s second week, one owner nearly left because unreturned cups had piled up behind the counter. The collection round addressed the cause.

Match support to the stage. An information pack serves knowledge, not persuasion. Counter help serves implementation; a check-in after the first month serves confirmation. A communication plan built around the questions people actually have does more than one launch campaign aimed at everyone.

Influence travels through people others watch

Rogers calls diffusion “a very social process that involves interpersonal communication relationships”. Close peers’ evaluations usually carry more weight than outside expertise, so people look to similar others who have already adopted. Rogers calls similarity between people homophily and difference heterophily, and Greenhalgh’s review notes that adoption is more likely when people resemble current users. Yet Rogers and colleagues’ 2005 paper explains that while very high heterophily is likely to slow diffusion, some is necessary, because a source must know more about the innovation than the receiver. Sahin quotes Rogers naming as one of diffusion’s “most distinctive problems” that participants usually differ from one another.

That tension shapes two roles. A change agent introduces an innovation on behalf of an organisation; Lena is one, and she has never run a counter during the morning rush. An opinion leader is someone whose advice peers seek informally. Dearing describes opinion leaders as nearby, credible and accessible near-peers, and warns that formal authority often does not bring influence. The waste department has authority and budget, but little credibility on running a café.

In our fictional city, Tomasz has run a café near the market for fifteen years. Other owners ask his advice about suppliers, and he started their informal messaging group. Dearing cautions that asking opinion leaders to promote something in ways they normally would not puts their standing at risk. So Lena did not make Tomasz an ambassador with a script. She asked whether other owners could visit his café on an ordinary morning and watch cups coming back, tokens changing hands and a crate filling up.

Rogers, Singhal and Quinlan also describe a network effect: for interactive innovations such as email, each new adopter makes the innovation more valuable to earlier and later adopters, and at a critical mass diffusion becomes self-sustaining. A cup scheme is not a communication technology, but its return network behaves similarly: a deposit is easier to accept when there is a return point near your office. That is our interpretation, not Rogers’ claim, and it argues for recruiting cafés in clusters.

A stakeholder map helps if it records who asks whom for advice, not only who holds formal power. Greenhalgh’s review adds a caution: when opinion leaders have been trained to influence peers, the effect has generally been positive but small, and opinion leaders can also argue against an innovation.

Adopter categories describe timing, not character

Rogers grouped adopters by how early each adopted compared with others in the same system. Newcastle University’s TheoryHub summary sets out the five categories and their idealised shares: innovators (2.5 per cent), early adopters (13.5), early majority (34), late majority (34) and laggards (16).

The categories exist only after the fact, for one innovation in one system; Sahin notes that only adopters of an innovation that did spread form the distribution. The neat percentages come from assuming a normal curve cut at standard deviations from the mean adoption time. The same café could be early to offer oat milk and late to go card-only. Rogers did generalise about each category’s typical traits, but nothing in the definition makes them personality types.

Greenhalgh’s review is blunt. It finds little empirical support for “stereotypical and value-laden terms” such as “laggard”, which fail to treat the adopter as someone who engages purposefully and creatively with a complex innovation. Sahin also quotes Rogers’ observation that those who most need an innovation’s benefits are generally the last to adopt. Rogers’ examples are the less educated and less wealthy.

In the example, the waste department’s first progress note called cafés that had not signed up “laggards”. Lena asked for the word to be removed. The non-joiners included a counter-only coffee bar with nowhere to store a crate and a café whose glasswasher was already at capacity: practical constraints, not attitudes.

A popular marketing extension, Geoffrey Moore’s Crossing the Chasm, concerns start-ups selling what he calls disruptive innovations as they move from early adopters, who buy into a vision, to pragmatist mainstream customers, who want proof in use cases like their own. It is practitioner strategy advice, not part of Rogers’ theory or a tested finding. Whether a product is disruptive in Christensen’s sense is a separate question.

When the adopter is an organisation

Classic diffusion studies mostly treated the individual as the unit of adoption, and Rogers and his co-authors called for more attention to organisations. Greenhalgh’s review distinguishes four processes that are easy to blur: diffusion (passive spread), dissemination (active, planned efforts to persuade target groups), implementation (active, planned efforts to mainstream an innovation within an organisation) and sustainability (making it routine until it becomes obsolete).

Dearing and Cox similarly relate diffusion to dissemination, implementation, sustainability and scale-up, and ask why worthy innovations do not spread faster. Lena’s information pack is dissemination. What happens behind each counter is implementation.

Even a small café is an organisation: the owner decides, but the early-shift baristas carry the decision out. Greenhalgh’s review names six elements of system readiness for a particular innovation: tension for change, innovation–system fit, assessment of implications, support and advocacy, dedicated time and resources, and capacity to evaluate. In café terms: do customers already ask for alternatives? Does the cup fit the racks? Who on the early shift supports it? Will anyone notice if returns quietly stop?

Our playbook on assessing readiness for a specific change turns such questions into a decision review. ADKAR can structure support for an individual barista, but it answers a different question. Diffusion asks how a practice spreads across a system; ADKAR asks what one person needs in order to change.

When not adopting is the sensible choice

Rogers himself discussed four major criticisms of diffusion research, which the TheoryHub summary sets out:

• Pro-innovation bias: assuming an innovation should be adopted by everyone, quickly and unchanged.
• Individual-blame bias: holding individuals responsible for problems that belong to the system.
• The recall problem: people misremember when they adopted something.
• Equality: diffusion can widen existing socio-economic gaps.

Dearing adds that interventions of unknown effectiveness, and even known ineffectiveness, often spread while effective ones do not. How far something spreads is not evidence that it works.

For Amira, who runs the counter-only coffee bar, joining meant storing crates she has no room for and washing cups by hand. Declining was rational. Lena treated the refusal as design information, in the spirit of treating an objection as information before treating it as a problem. She offered a return-point-only membership: Amira accepts cups and gives credit, and the collection round takes them away unwashed. Amira agreed to try it. Another café still said no, and Lena recorded its reasons rather than chasing it.

Equality matters at this scale too. If only well-equipped central cafés can join, customers who want reusable cups drift towards them, while smaller cafés further out keep paying for disposables. The recall problem is easier to prevent than to fix: Lena kept a dated log of when each café joined, paused or left, and why.

What the organiser decided after the pilot

Lena and the waste department had to choose: proceed with the city-wide launch or change the approach. They dropped the launch to every café at once. Instead, a second wave would recruit cafés in two neighbourhoods, so return points would be close together. The pilot’s design changes and the return-point-only option became standard, new cafés would get the no-penalty trial, and owners could keep visiting Tomasz’s café.

The press release waited until the stickers and collection bike were visible in those neighbourhoods. Lena would review the log after three months, looking for disenchantment as well as new sign-ups. They were careful about what they did not claim: the pilot showed the scheme could work in six closely supported cafés, not that it reduces waste overall or that every café should join.

Limitations and evidence

Diffusion theory explains how and why something spreads. It cannot tell you whether it deserves to. Whether a reusable-cup scheme reduces environmental impact depends on return rates, washing and transport, and needs its own evidence.

The research base is large but uneven. Much of it comes from agriculture and health services, often studying individuals, so applying it to cafés involves judgement. Tornatzky and Klein’s meta-analysis dates from 1982, and diffusion studies have typically asked people to recall, after the event, when they adopted and what influenced them. Greenhalgh’s review found that few empirical studies addressed the complexity of spreading innovations in service organisations, and that only one of the more than 200 studies it covered examined discontinuance prospectively.

The S-curve and adopter categories are built from innovations that did spread, and they describe the past more reliably than they predict the future. Use the attributes as questions for adopters, the stages as a guide to timing support, and the network ideas as a prompt to find credible peers. The worked example and its table are fictional illustrations, not results.