What the research says

Should change be top-down or bottom-up?

The short answer

The research does not settle it: about half the studies support starting change close to the work, and nearly as many find it depends. In one survey, employees backed change most when middle managers started it and top managers carried it out, yet that pattern was the rarest, about 3% of cases (Heyden et al., 2017). In an empowerment experiment, managers' sense of control rose only where they felt their own bosses backed them, though their units' measured performance also improved. More freedom for managers does not help everywhere. This is the most studied part of the Loop, and the most contested.

Based on 13 studies from the Atlas of the Middle: 0 rated high strength, 10 medium, 3 low. Designs: qualitative case (6), survey (4), experiment (2), qualitative interviews (1).

Who starts it, and who leads which part

One survey compared where a change began with how much support it drew. In another, of managers at Chinese acquirers, more middle-manager involvement in slower integration of the people side went with less employee resistance.

Empowerment needs backing from above

In one randomized study, empowerment raised managers' sense of control only under a condition, and a utility's change program found top-down direction and shared leadership working together throughout. In a cluster-randomized trial, training middle managers raised their units' psychosocial safety climate compared with matched controls. After COVID-19 hit, climate stayed higher in the trained units, while a strong top-management response lifted communication and commitment in the control units too. At a French bank, one study reports that middle managers backed by structure and top management adapted the strategy to local conditions.

More freedom isn't always better

Some studies find limits: high autonomy weakened the payoff from initiative for risk-takers, gains depend on experience and unit size, and room to act can be misused.

When the middle shapes it

Several cases show change shaped or carried by the middle, sometimes after the leader who started it has left.

Opening it up can sideline the middle

When a large financial company opened strategy making to front-line employees, its middle managers' traditional role as strategists was undermined. They took different routes to reclaim it over time.

Broader evidence from the Evidence Library

These come from syntheses and trials on organizational change generally, not only middle managers, graded with the same rubric as the Atlas studies.

  • Brookman-Frazee et al. 2025 · two linked cluster randomized trials, 65 mental health programs and school districts (387 providers) · rated High strength

    A strategy aimed at program leaders raised how faithfully a practice was delivered, as rated by observers blind to condition, but did not change provider certification. A separate strategy aimed at frontline providers had no significant effect on any outcome.

    This compares working through leaders with working on frontline providers, not top-down with bottom-up design, so it informs the Atlas question without settling it; it comes from health and education services.

  • Patterson et al. 2010 · series of systematic reviews of longitudinal and intervention studies (NHS evidence report; searches in 2006) · rated Medium strength

    A small number of studies supported involving employees in the design and implementation of changes that affect their work, and in health care, employee involvement through quality improvement teams was linked to better patient outcomes.

    It leans toward involving people close to the work, as about half the Atlas studies do, but the evidence base is small and dated.

Still untested in the Evidence Library (Loop Part 3): no study compares team-designed with manager-designed implementation.

What I take from it

I don't read this as a vote for either direction. The pattern I take from it is that a change needs real input from the people closest to the work and visible backing from the top, at the same time.

This is my reading, Gabriel Lopez's, not a finding of the studies above.

Common questions

Is bottom-up change more effective than top-down?

The studies we mapped do not settle it. Of 67 studies that bear on whether change should start close to the work, 34 support it, 30 find it depends, and 3 push back.

Does empowering middle managers improve results?

In a randomized program at a large trucking company, empowerment raised unit managers' sense of control only where they felt their own bosses backed them, and it also lifted their units' measured performance.

Every summary on this page was written by Unfrozen from the study's abstract or full text and checked twice against the source. Loop coding is exploratory and provisional until an independent human coder checks it. Spot an error? Email info@unfreeze.us and it gets fixed.

More questions