The research behind decisions that stick

A decision gets made in the room. Everyone nods. A week later, the same agenda item is back, or the decision quietly stopped happening the way it was agreed. This page is not our opinion about why that happens. It is a plain-language summary of six published sources — five peer-reviewed studies and one policy report — that each explain a piece of the pattern. Every claim below is cited; follow the DOI to read the original.

Why doesn't agreement in the room survive the week?

People appear to agree, then work against the decision afterward

Oreg, Vakola and Armenakis (2011) reviewed decades of research on how people actually react to organisational change. One consistent pattern across that body of work: visible agreement during a change process and genuine commitment are not the same thing. When people go along with a decision in the meeting without a real chance to raise what they actually think, the gap does not disappear — it resurfaces afterward as quiet non-compliance rather than an open objection. That is why the same item keeps returning to the agenda: the disagreement was never resolved, only postponed.

Choice and rationale drive acceptance — not being told, and not just having access

Gagné, Koestner and Zuckerman (2000) studied what actually makes people accept an organisational change, grounded in self-determination theory. Their finding: acceptance depends on two things — understanding the genuine reason for the change, and having some real say in how it gets implemented. A change that is simply announced, however clearly explained, produces shallower acceptance than one where the people affected had a hand in shaping it. This is why a decision made without the people who hold the trade-off does not hold once they leave the room — they were told the what, not given a stake in the how.

A one-off intervention reverts once attention moves on

Weick and Quinn (1999), in a foundational review of organisational change research, draw a distinction between "episodic" change — a deliberate, time-boxed intervention with a clear start and end — and "continuous" change, the ongoing, cumulative way an organisation actually adjusts day to day. Their argument: treating change as an episode (a rollout with a launch date and a project close-out) interrupts the old pattern temporarily, but does not alter the underlying, continuously-adjusting behaviour that produced it. Once the project team moves on, the old rhythm reasserts itself — because it was never actually replaced, only paused.

The plan on paper and the work on the ground are different things

Orlikowski (1996) followed a real organisation adopting new technology over time and found that transformation does not happen the way it is designed on paper — it happens through the ongoing, improvised adjustments that the people doing the work make as they go. When a plan is built without consulting the people who know what the process actually looks like day to day, the plan and the practice diverge from day one, and the gap between them is exactly where a rollout stalls.

Why doesn't access to a new tool become a changed way of working?

A successful pilot does not automatically scale

Haefner, Wincent, Parida and Gassmann (2023) examined what happens when organisations try to move from a working AI pilot to organisation-wide use. Their finding: the same structural barriers that a pilot team quietly worked around — often with informal support, extra attention, or a highly motivated group — reappear at full scale, where none of those workarounds are available. Access does not become use, and use does not become a changed way of working; it is the same pattern that stalls scaling after a pilot, just at a larger size.

Fast adoption, slow uptake — and the barrier is not access to the model

The OECD, BCG and INSEAD (2025) published a large policy-oriented study of AI adoption across firms. It found that organisations are adopting AI tools quickly, but real operational uptake lags far behind adoption. The barrier is not access to the technology. It is skills, data readiness, cost, and organisational readiness — the practical conditions that determine whether a tool that is technically available actually gets used to do the work differently.

What does this mean in practice?

Read together, these six sources describe the same mechanism from different angles: a decision or a change holds when the people who have to live with it had a genuine say in shaping it, and it reverts or gets quietly worked around when it was designed and announced without them. That is not a motivation problem, and it is not something more training or a better launch event fixes — the studies above are explicit that the objections are rational, not a failure of skill or willingness.

It is also the mechanism our structured sessions are built to interrupt: getting the people who know why something isn't working and the people with the authority to change it into the same room, with methods that surface the objection before it becomes a workaround, not after.