Planned change has a longer academic pedigree than most practitioners assume, and the oldest model is still the simplest. Lewin (1947) described change in three stages: unfreeze, in which the forces holding the current state in place are deliberately loosened so people accept the status quo cannot continue; change (or "move"), in which new behaviour is actually adopted; and refreeze, in which the new behaviour is reinforced until it becomes normal rather than a temporary initiative that reverts once attention moves elsewhere. Behaviour, in Lewin's terms, sits in a quasi-equilibrium held in place by competing forces, and skipping the unfreeze stage is the most common reason change programmes meet quiet, durable resistance rather than open conflict.

Kotter's eight steps (1995, expanded in Leading Change, 1996) sequence that same logic into an operational checklist for large transformations: establishing a sense of urgency, forming a guiding coalition, creating a vision, communicating the vision, empowering others to act on it, generating short-term wins, consolidating gains to produce more change, and anchoring new approaches in the culture. Kotter's own study of over a hundred transforming companies found failure concentrated at specific steps — skipping urgency, or declaring victory after the first win instead of using it as momentum. The model is not uncontested: emergent-change scholars argue it assumes change is plannable, top-down and linear in a way real organisations rarely are, and a leader who treats the steps as a checklist can complete every one and still watch the change fail to stick.

Hiatt's ADKAR model, developed by Prosci from research across hundreds of organisations and set out fully in Hiatt (2006), sequences what has to be true for one individual to change rather than what the organisation does: Awareness of why change is needed, Desire to support it, Knowledge of how to change, Ability to apply new skills, and Reinforcement to sustain it. Its use is diagnostic: when change stalls, ADKAR asks which letter is missing for the people affected, rather than assuming the whole programme has failed — a workforce aware of why a system is changing but untrained is failing at Knowledge, not Desire, and the two need different fixes.

No one model is complete. What all three can accommodate is that a named, visible sponsor is one of the strongest single predictors of success, because sponsorship supplies Lewin's unfreeze, anchors Kotter's coalition, and gives ADKAR's Desire stage a credible source.

Change models describe how an organisation moves people through a planned transition; diffusion of innovation describes something related but distinct — how a new idea or technology spreads through a population once it is available, whether or not anyone is actively managing the process. Rogers (1962) found that adoption over time traces a consistent S-shaped curve rather than a straight line: slow at first, accelerating sharply once a critical mass takes hold, then levelling off as the population approaches saturation. Plotted instead as adopters per unit of time rather than cumulatively, the same pattern forms a bell curve, and it is this bell curve that Rogers divides into five categories.

Innovators, roughly the first 2.5% of a population, are venturesome risk-takers often more connected to other innovators outside the organisation than to the mainstream within it; they adopt before an innovation has much social proof, and they tolerate failure well, which means their enthusiasm is a weak signal of whether the broader population will follow. Early adopters, the next 13.5%, are usually the most locally respected and socially integrated group in the population, which is why they function as opinion leaders — others watch what the early adopters do, not what the innovators do, before deciding whether an innovation is credible. The early majority, at 34%, adopt only once the innovation is established as sound and are deliberate rather than quick, while the late majority, also 34%, are sceptical and adopt mainly under social or economic pressure once most others already have. Laggards, the final 16%, are the most locally oriented and the most reluctant, often adopting only when continuing not to is no longer a viable option.

The practical weight of the model falls on the early majority, not the innovators, and this is the point most often missed by anyone judging early momentum too confidently. A pilot that delights its innovators and early adopters has won only the easiest 16% of the population; the 'chasm' Moore (1991) identified between early adopters and the early majority is where many promising pilots stall, because the early majority is not persuaded by the same arguments that moved the innovators — it wants proof the innovation works at scale, for people like them, before it will commit. Rogers's proportions are a well-evidenced empirical regularity across many studies of innovation diffusion, not a law that reproduces identically in every context; a population's actual split depends on how the innovation is introduced, who sponsors it and how visible its early results are, so the categories are a diagnostic lens for thinking about a specific rollout, not a fixed prediction to apply mechanically.

Rogers (1962) also identified five characteristics of an innovation itself, independent of who is being asked to adopt it, that predict how quickly it spreads. Relative advantage is the degree to which the innovation is perceived as better than what it replaces — and the operative word is perceived: an innovation that is objectively superior but not experienced as better will spread slowly regardless of its real merits. Compatibility is how well it fits existing values and the way work already gets done; a technology that requires abandoning a familiar routine carries an adoption cost before its benefits arrive. Complexity, unusually among the five, correlates negatively with adoption speed — the easier an innovation is to understand, the faster it spreads, so a technically elegant but hard-to-explain solution can diffuse more slowly than a cruder, intelligible one.

Trialability is whether the innovation can be tried on a limited, reversible basis before full commitment — a far smaller psychological ask than an irreversible one, especially for the early and late majority, who wait for evidence the innovators could do without. Observability is whether results are visible to others; a technology whose benefits are private spreads only by word of mouth, while one whose results are visible — a shorter queue, a route finished early — recruits its own adopters. The factors interact: strong trialability can rescue a complex innovation by lowering the cost of finding out whether the complexity is worth it, while a simple innovation with neither trialability nor observability can stall even when genuinely advantageous.

These factors are also where change leadership has the most leverage, since several can be deliberately engineered: a phased pilot manufactures trialability, a dashboard of pilot results manufactures observability, and training that maps onto existing routines improves compatibility. None of this guarantees adoption — a genuine, unresolved disadvantage for one group, such as unreliable connectivity, still slows adoption for that group however well the other factors are engineered, and the honest response is to name that limit rather than assume communication can close a gap that is actually technical.

A change-and-adoption plan is where the models of the first three lessons stop being separate theories and become one sequenced document. Sponsorship comes first, not as a formality but as the single move that does the most work across every model covered so far: a named, visible sponsor supplies Lewin's unfreeze (someone with the authority to say the status quo is ending), anchors Kotter's guiding coalition, and gives ADKAR's Desire stage a credible source. A rollout with no sponsor, or a sponsor named on a slide but never visibly spending authority on the change, tends to fail quietly rather than dramatically — not through open resistance, but through a slow accumulation of reasons not to prioritise it.

Communication follows sponsorship, and its job is to do deliberately what the adopter-category and adoption-factor lessons describe happening naturally: build awareness and desire (ADKAR) ahead of training, and manufacture observability (Rogers) so a pilot's results recruit adopters rather than requiring every one of them to be individually persuaded. The content of that communication should differ by adopter category rather than repeat the same message to everyone — early adopters respond to being first and to a credible path to real advantage, while the early and late majority need visible, replicated proof that the innovation already works for people like them, which is exactly why a premature organisation-wide launch forfeits the chance to generate that proof.

The rollout itself should be phased deliberately around diffusion theory rather than around convenience: target a pilot group likely to include genuine early adopters, resource that pilot properly enough to produce an honest result either way, and only then widen to the early majority using the pilot's own results as evidence — which is Rogers's trialability and observability, engineered on purpose, and Kotter's short-term wins, generated deliberately rather than hoped for. A rollout that skips the pilot and goes straight to the whole organisation denies itself both the evidence the early and late majority need and the chance to fix what the pilot would otherwise have caught early, at the cost of fixing it later at a far larger scale.