Organisational Development: a map of the field — and its neighbours

What OD is

Organisational development is the practice of helping organisations, teams and groups work better and change well, by working with the people in them rather than doing change to them. It is a field with its own history, schools and arguments, and with neighbours that are often mistaken for it.

The classic definition comes from Richard Beckhard (1969): OD is an effort that is planned, organisation-wide and managed from the top, to increase an organisation’s effectiveness and health through planned interventions in its processes, using behavioural-science knowledge. Today’s definitions put more weight on collaboration and values. The OD Network describes the field as building system-wide capacity, effectiveness and vitality, guided by humanistic, developmental, inquiry-based and evidence-informed values.

In practice, “OD” is used in three ways:

• A field of knowledge about how groups and organisations function and change. It draws on social psychology, systems theory, organisational behaviour and, in one strand, psychoanalysis.
• A professional role. Many organisations have OD practitioners or OD teams, often sitting in or near HR, alongside external OD consultants.
• A set of values: participation, respect for people’s own knowledge of their work, learning from experience, and attention to process (how things are done) as well as content (what is done).

OD overlaps with neighbouring fields but is not the same as them. Change management tends to focus on implementing a defined change, such as a new system, structure or policy. Learning and development focuses on individual skills. HR manages the employment relationship. OD is more concerned with the health and capability of the whole system, and with how the people affected take part in the change.

Who defines OD: the OD Network and the CIPD

Definitions of OD are partly a question of professional territory, and they differ by country. The US-based OD Network keeps the classic values-based definition quoted above. In the UK, the CIPD, the HR professional body, publishes a factsheet on organisation development that it abbreviates “ODV”, a label it uses to distinguish organisation development from organisation design. It describes OD as “a planned and systematic approach” to enabling sustained organisational performance through the involvement of an organisation’s people. It is the same territory, framed from HR instead of from behavioural science. Where OD ends and HR or learning and development begin is therefore a professional boundary as much as an academic one.

Clinical work and OD side by side

If you trained in therapy or mental health, much of OD will feel familiar, because many terms have a clinical cousin. The table shows where the two line up and where they part. Readers from other backgrounds can skip ahead to Where OD came from.

Where OD came from

OD grew from two post-war roots, one American and one British. They met early. Knowing both explains why OD can feel both familiar and foreign to clinicians.

The American root: Lewin, action research and T-groups

Kurt Lewin, a German-American social psychologist, argued that you understand a system best by trying to change it. His action research cycle (plan, act, observe, reflect, repeat with the people involved) is still the backbone of OD. In the mid-1940s Lewin and colleagues found that groups learned powerfully from feedback on their own here-and-now behaviour. That became the T-group (training group). In 1947, the year Lewin died, the National Training Laboratories (NTL) began at Bethel, Maine, and T-groups spread through American business and public services. Survey feedback (collecting staff views and feeding them back to the group to act on) grew from the same tradition.

The British root: the Tavistock

In Britain, psychiatrists and psychoanalysts who had worked on wartime problems (officer selection, rehabilitating returning soldiers) founded the Tavistock Institute of Human Relations after the war. Wilfred Bion’s work with groups, Eric Trist’s studies of coal mining, and later Isabel Menzies Lyth’s hospital study brought psychoanalytic and open-systems thinking into the study of work. This strand became known as systems psychodynamics (section 8).

Where they met

In 1947 the Tavistock Institute and Lewin’s Research Center for Group Dynamics at MIT jointly launched the journal Human Relations. So the two roots shared a journal from the start. Over the next decades OD became a recognised profession in the US, mostly drawing on the Lewinian side. The Tavistock tradition stayed closer to psychoanalysis and more influential in the UK, Europe, Australia and South Africa.

1. 1940sBion’s wartime work with groups; Tavistock Institute founded
2. Mid-1940sLewin and colleagues develop action research and the T-group
3. 1947Human Relations launched by the Tavistock and MIT; NTL begins at Bethel, Maine; Lewin dies
4. 1951Trist & Bamforth: changing a mining method breaks up working groups (the socio-technical idea)
5. 1955–1961Jaques (1955) and Menzies Lyth (1960) on social defences against anxiety; Bion’s Experiences in Groups (1961); first Leicester group relations conference (1957)
6. 1960sBeckhard’s definition of OD (1969); Schein’s process consultation; Argyris on organisational learning
7. 1967Miller & Rice, Systems of Organization: primary task and boundaries
8. 1968–1973Campbell and Dunnette review T-group outcomes (1968); Lieberman, Yalom and Miles report on encounter groups (1973)
9. 1970s–1990sDiagnostic models (Weisbord 1976; Burke–Litwin 1992); appreciative inquiry (1987); ISPSO founded (1983); The Unconscious at Work (1994)
10. 2009 onBushe & Marshak name “dialogic OD”; psychological safety becomes a central idea in team research

Blue markers: the British (Tavistock) strand. Green: the American (Lewinian) strand and later developments.

The T-group arc: from mainstream to cautionary tale

T-groups were not a side-show. Highhouse’s history records how the method grew inside the National Training Laboratories until it was part of OD’s mainstream, and then declined as doubts about its outcomes accumulated. An early review by Campbell and Dunnette (1968) had found that T-group experiences moved self-reported attitudes and interpersonal awareness, with thin evidence that any of it carried over into job performance. A harder look at encounter groups, a related method, found real harm for some participants. Groups: from the therapy room to the team tells that part of the story, with its numbers and their definitions.

The consulting stance

Before any model or tool, OD is a way of being with a client system. Edgar Schein called it process consultation.

Three ways to help

Schein distinguished three models of helping:

• The expert. The client knows the problem and buys an answer: a survey, a restructure, a training course. This works only if the client has diagnosed the problem correctly.
• The doctor. The consultant investigates, diagnoses and prescribes. This assumes the client gives accurate information, accepts the diagnosis and follows the prescription. In organisations, often none of these hold.
• The process consultant. Consultant and client inquire together. The client keeps ownership of the problem and learns to solve similar problems later. Schein’s advice: start here, and move into expert or doctor mode only once you understand the situation.

The consulting cycle

Most OD work follows a version of the action-research cycle:

1. Entry. Who called, why now, and what did they ask for? The first request is data. It is rarely the whole problem.
2. Contracting. Agree purpose, sponsor, client system, boundaries, confidentiality, timescale and how you will both know if it helped. Re-contract when things shift.
3. Data gathering. Interviews, focus groups, observation, surveys, documents, and your own experience of the system.
4. Feedback and sense-making. Share themes with the people who gave the data, and make sense of them together. How the system responds to its own data is often the most useful information you get.
5. Action. Plan and carry out changes with the people involved: in structure, roles, processes, relationships or ways of meeting.
6. Review and exit. Did it help, for whom, and what else changed? Leave the system more able to do this itself.

The habit of asking for the source

The most useful habit in this field is also the simplest: when a number or a model is handed to you, ask where it came from. This page uses a recurring exhibit, “Ask for the source”, for four cases. The change-failure statistic shows a caveat that vanished as the number spread. The positivity ratio shows a claim, its debunking and a formal correction. The generative-AI pilot figure shows the same pattern a generation later. And a short correction note shows that this page’s own research made a mistake too.

Making sense of a system

OD has its own versions of a formulation framework: models that tell you where to look and how the parts connect.

Open systems

The basic idea under almost every OD model: an organisation is an open system. It takes in inputs (people, money, referrals, information), transforms them through its work, and sends out outputs (services, products) into an environment that keeps changing. Trouble in one part shows up somewhere else. A problem that looks like poor morale on one team may begin in how work is handed over from another.

Three widely used models

Culture: Schein’s levels, and how far leaders can change it

Edgar Schein’s three levels of culture appeared in print in a 1984 article and became canonical in his book Organizational Culture and Leadership. They are artifacts (what you can see and hear), espoused beliefs and values (what people say matters) and basic underlying assumptions (what is taken for granted). The wording of the middle level varies by edition. The fifth edition says “espoused beliefs and values”; many summaries shorten it to “espoused values”. If you quote the levels, cite an edition.

Can leaders change culture on purpose? A review in the Journal of Management covering 74 studies (Kim and Toh, 2022) found that leaders have a disproportionate influence on culture and that “when exercising such influence, they are often unsuccessful at creating functional cultures”. Cultures tend to track a leader’s traits and past experience more than present function. Schein’s own account of how assumptions change is that they shift through new, repeated experiences that survive contact with reality, not through posters or decrees.

Change, loss and “resistance”

This is where a clinical background helps most, and where OD’s everyday language can sound oddly thin to clinicians.

Organisational life often talks about “change resistance” as a problem to overcome. A clinician will recognise something else: people grieving. Two writers make this link explicit.

• Peter Marris (Loss and Change, 1974) argued from studies of bereavement and of communities being rehoused that people need continuity of meaning. Any change, even a welcome one, breaks the link between what we did and why it mattered. The loss has to be worked through before people can commit to the new.
• William Bridges distinguished change (the external event: a new structure starts on Monday) from transition (the inner process of letting go). Transition has three phases: an ending, a confusing neutral zone, and a new beginning. Organisations usually plan for the change and ignore the transition.

Survey research supports a link between how people defend against anxiety and how they respond to change. Bovey and Hede surveyed 615 staff across nine organisations and found that some defence mechanisms were associated with intention to resist change. This is an association, not a demonstration of cause.

Change management and OD: a family quarrel

Change management (the Kotter, Prosci and ADKAR lineage) tends to be programmatic: a defined end state, a sponsor and a sequence. OD is more values-based, built on participation and on the idea that people who help design a change will own it. Both cite Lewin, and over the years those citations have turned into folklore. Three pieces of that folklore are worth checking before you repeat them.

Learning, defences and safety

Mainstream OD has developed its own ideas about defences, without using psychoanalytic language.

Single-loop and double-loop learning

Chris Argyris and Donald Schön described two kinds of organisational learning. Single-loop learning corrects errors within the existing rules, like a thermostat switching the heating on when it gets cold. Double-loop learning questions the rules themselves: why is the thermostat set to that temperature? Argyris found that professionals and managers are often good at single-loop learning and bad at double-loop learning. Under threat, they protect themselves with defensive routines: smoothing things over, making issues undiscussable, then making the undiscussability undiscussable.

Psychological safety

Amy Edmondson (1999) studied hospital and manufacturing teams. She found that teams learn better when members believe it is safe to take interpersonal risks: to admit mistakes, ask questions or disagree. She called this psychological safety. It is now one of the most used ideas in team development, including in healthcare safety work.

What Argyris and Schön added

Argyris and Schön’s Organizational Learning (1978; second edition 1996) supplies three further terms. Espoused theory is what people say guides their actions; theory-in-use is what their actions show. The gap between the two is where defensive routines live. These are conceptual tools: a book that names a mechanism is not evidence that the mechanism changes performance. The field’s own debates (for example, between programmed and emergent views of learning) are mapped by Easterby-Smith, Thorpe and Jackson (1999).

Does the learning organisation improve performance?

The psychodynamic strand: systems psychodynamics

This is the part of OD built from psychoanalytic ideas, the ideas many clinicians trained in. It joins psychoanalytic thinking about anxiety and defence with open-systems thinking about task, boundary and role.

The core claim: work stirs up feelings, and some work does so by its nature (caring for people in distress, holding risk, deciding who gets scarce help). Groups and organisations develop shared ways of managing those feelings. Some help the work. Some quietly replace it. The consultant studies the person-in-role-in-system: not people’s inner lives as such, but how task, structure and emotional life meet.

Key ideas

How the ideas are used

Systems-psychodynamic consultants work through role consultation (one-to-one work on how someone takes up their role), team and organisational consultation, group relations conferences (temporary learning organisations where members study authority and group life as it happens; the first Leicester Conference was in 1957), organisational observation (adapted from infant observation), and social dreaming.

Groups: from the therapy room to the team

Group therapy gave organisational thinkers a vocabulary before OD had one of its own. Some of it travels well to work teams. The history of how OD tried to import the rest is the field’s most useful cautionary tale.

The vocabulary that crossed over

Wilfred Bion’s Experiences in Groups (1961) described group life as a constant swing between the work group, which gets on with its task, and basic-assumption behaviour (dependency, fight/flight, pairing), in which the group manages a shared anxiety instead of the task. The Tavistock tradition carried this into organisational consultation, through studies such as Trist and Bamforth’s coal-mining study and Menzies Lyth’s hospital study. What section 8 calls a social defence is a close relative of Bion’s idea that a group can protect itself from anxiety at the expense of its work.

S. H. Foulkes’s group-analytic tradition (1948, 1964) added a relational layer: the matrix, the web of communication in which a group’s members think and feel together. It is the ancestor of the intuition that a team can think in patterns no single member authored. Irvin Yalom’s list of therapeutic factors (The Theory and Practice of Group Psychotherapy; the standard list has about eleven, including universality, interpersonal learning, cohesiveness, altruism and the instillation of hope) is a clinician’s checklist of what a group can do that a one-to-one room cannot.

Cohesion and safety: related ideas, different currencies

Amy Edmondson’s psychological safety is a shared belief, held by members of a team, that the team is safe for interpersonal risk-taking. Unlike Bion’s concepts, it can be measured. On the therapy side, a meta-analysis by Burlingame, McClendon and Yang (2018) pooled 55 studies and found a moderate association between group cohesion and outcome (r = .26).

The T-group and the encounter group: therapy technology at work

The training group is the historical case where the analogy was not just drawn but enacted, at scale. Highhouse’s history records how the method grew inside the National Training Laboratories until it was part of OD’s mainstream, and then declined as doubts about its outcomes accumulated. An early review by Campbell and Dunnette (1968) found that T-group experiences moved self-reported attitudes and interpersonal awareness, with thin evidence that any of it transferred to job performance.

The hardest look came from Lieberman, Yalom and Miles’s Encounter Groups: First Facts (1973): 210 Stanford volunteers in 18 groups, followed up afterwards. About a third gained meaningfully. But 16 participants (8%) were “casualties” by the study’s strict definition: an enduring, significant negative change that could be attributed to the group, still present at least eight months later. Counting any negative change at all, the figure was about 16%. The casualties clustered around group leaders with an aggressive, charismatic style.

What OD took from the episode

Edgar Schein’s Process Consultation Revisited (1999) is the guardrail vocabulary OD developed: an explicit contract, clarity about roles, the client system rather than the individual as the unit, and dependency watched on purpose as an occupational hazard of the helper. His framework is about helping a client system see its own process without becoming its therapist (see What the work looks like).

The behavioural tradition at work

Where the group tradition’s organisational evidence is thin, the behavioural tradition is more heavily quantified. Its promises are narrower, though, and the numbers need their outcome labels.

Organisational behaviour modification

Luthans and Kreitner’s Organizational Behavior Modification (1975) systematised an antecedent–behaviour–consequence (A-B-C) approach to measurable work behaviour: define the behaviour, measure it, change what comes before and after it, and measure again. Its main effect claim rests on a meta-analysis by Stajkovic and Luthans (1997) of studies from 1975 to 1995. They found d = .51 on task performance, which they describe as a 17 per cent increase. The effect was stronger in manufacturing than in service settings. You may see higher productivity figures attributed to OB Mod elsewhere; the published abstract says 17 per cent.

Albert Bandura’s social learning theory (1977) is the bridge position: behaviourism widened to take in cognition, which is why neither the behavioural nor the cognitive camp can claim or dismiss him cleanly.

Worksite stress programmes: CBT and ACT

Richardson and Rothstein (2008) pooled 36 studies (N = 2,847) of occupational stress management programmes and found an overall effect of d = 0.526 on psychological outcomes. Cognitive-behavioural programmes consistently produced larger effects, and adding further components tended to reduce the effect, not raise it.

The cleanest mechanism evidence comes from a randomised trial by Bond and Bunce (2000). In a media organisation (N = 90), they compared an acceptance and commitment therapy (ACT) programme, a problem-focused “Innovation Promotion Program” and a waiting list. Both programmes did better than the waiting list on mental health and work-related measures. ACT’s benefit was mediated only by acceptance of undesirable thoughts and feelings, which is an emotion-focused mechanism.

The pairing of psychological flexibility with work outcomes belongs to a different paper, Bond, Flaxman and Bunce (2008), a study of work redesign. A systematic review of ACT for professional staff burnout (Towey-Swift and colleagues, 2022) reads the effects on burnout as less consistent than the effects on mental health.

Behaviour-based safety

Behaviour-based safety (BBS) is the best-known operational arm of the tradition: observing safe and unsafe behaviours at work, giving feedback and reinforcing the safe ones. Carra and colleagues (2024) reviewed 230 papers from the 1970s to 2023 and describe an evolution towards positive reinforcement and observation, blurred boundaries with neighbouring approaches, and a discussion of the approach’s limits.

The critical treatment is Sutherland, Makin and Cox (2000), which applies the behavioural approach to safety while documenting its implications for management. Critics worry about the same few things: that “unsafe act” slides into “unsafe person”, and that measurement becomes monitoring. The critique is gathered on the critique shelf.

Four ethical hazards when clinical technology meets an employer

• Covert shaping. Kazdin (1975) is the standard early statement of the control, consent and coercion questions raised when the person being changed is an employee, not a client.
• Surveillance and blame. The observation and feedback machinery of BBS is the apparatus its critics aim at (Sutherland et al.; Carra et al.).
• Therapy-like intimacy in work groups. The T-group record (Lieberman et al.; Roback) is the field’s own case study; Schein’s process consultation supplies the guardrails.
• Voluntariness when the employer is the client. A stress-management or flexibility programme offered “for you” is usually bought by management for outcomes such as performance, absence or safety. If the stated promise is performance and the measured effect is symptom relief (as in the 2024 trial above), the basis for consent is off by a category.

Strengths-based and dialogic OD

A large part of current OD practice works almost the opposite way to depth psychology. It is worth knowing well, because you will meet it everywhere.

Appreciative inquiry

Developed by David Cooperrider and Suresh Srivastva in the 1980s, appreciative inquiry (AI) starts from what works rather than what is wrong. A typical cycle has four stages: Discover (what gives life here at its best?), Dream (what could be?), Design (what should be?) and Destiny or Deliver (how do we make it happen?). The theory is that organisations move in the direction of what they repeatedly ask about.

Diagnostic and dialogic OD

Gervase Bushe and Robert Marshak (2009) described two families of OD. Diagnostic OD treats the organisation as something that can be assessed against a model, then changed by planned intervention. Dialogic OD treats an organisation as made of conversations and stories. Change comes from changing the conversations: who takes part, which questions get asked, and what new language emerges. Dialogic methods include appreciative inquiry, Open Space, World Café and other large-group events.

Complexity thinking and Stacey: a neighbour, not a twin

Ralph Stacey’s complex responsive processes (Stacey, 2001) is a way of seeing organisations. Order arises from local interaction without a planner, the formal and informal organisation cannot be separated, and power and ideology run through the turn-taking of conversation itself. It describes a way of seeing and is not a tested method.

The link to OD runs through the dialogic school. Bushe and Marshak (2009) described a “bifurcation in OD not fully acknowledged or discussed in OD textbooks”. Dialogic OD sets aside diagnostic OD’s core premise (that reality can be objectively measured and then fixed) in favour of a social-constructionist one: realities are conversations, so changing the conversations changes the system. By their own later account, dialogic practitioners hold, explicitly or implicitly, complex responsive systems perspectives of the Stacey kind (Bushe and Marshak, 2015).

Appreciative inquiry, honestly

Cooperrider and Srivastva named appreciative inquiry (AI) in 1987, and Cameron, Dutton and Quinn’s positive organisational scholarship (2003) became its wider academic programme. Both are conceptual. The best published check comes from AI’s own advocates: Bushe and Kassam (2005) examined 20 published AI cases. All began with positive storytelling, all followed the four-stage model and all kept to the founding principles, and “only seven (35%) showed transformational outcomes”.

The 35% needs its definition: it is the share of published cases (a selected set) with transformational outcomes. It is not a success rate for AI. Bushe’s later account of the mechanism is that AI leads to transformational change when it creates enough disruption to set off self-organising processes aimed at what people widely want (Bushe, 2011). For a related cautionary tale from positive psychology, see the positivity-ratio exhibit in the evidence section.

The critique shelf: power, justice and critical management studies

Every field that promises to help organisations also takes a side about whose interests count. This shelf gathers the critics in one place and grades them by the kind of evidence they offer, so that you can weigh each one fairly.

Critics from inside OD, already met on this page

Agility, design thinking and the digital claims

Agile, design thinking and “digital transformation” arrive with their own vocabularies and are often presented as new. The honest question for each is whether it is a continuation of OD’s thinking, a rebranding, or something else. The answer differs between what these approaches say they value and what is observed when they are practised.

Agile: continuity in values, contested in practice

The first value in the Agile Manifesto (2001) is “individuals and interactions over processes and tools”. That sounds like an OD sentence written by software people, and the continuity of values needs no interpretation.

The peer-reviewed record then shows a gap between stated values and observed practice. Annosi and colleagues (2016) studied what they call the dark side of agile and found large-scale agile implementations associated with reduced individual and team learning. Khanagha and colleagues (2022) describe how short-deadline peer pressure and control by leaders can work against the innovation agile promises, and which leadership behaviours mitigate it. A fair summary is that agile is a cousin of OD in what it declares, and a less certain relative in what it does under scale.

Design thinking: the same shape

A conceptual review by Micheli and colleagues (2019) finds design thinking to be a coherent but under-specified family of practices. Its real points of contact with OD are at the level of values: putting users at the centre, iterating, and tolerating ambiguity.

Dunne (2018) studied how firms implement it. The barriers were unclear goals, the work of establishing legitimacy, cultural resistance, leadership turnover, and a bind between keeping design thinking isolated (safe but marginal) and assimilating it (influential but diluted). Those are the classic problems of organisational change, which OD already theorises.

Digital transformation and the GenAI headline

An influential practitioner essay by Davenport and Westerman (2018) asks why so many high-profile digital transformations fail. It is a useful discussion of cases, and it is an essay, not a measured failure rate. More recently, a statistic about generative AI has gone through the same life cycle as a much older one about change in general.

Wellbeing by design

Wellbeing at work is often treated as something done to individuals: resilience training, a mindfulness app, a helpline. A long-running strand of research and, more recently, a formal standard treat it as a property of how work is designed.

The standard

ISO 45003:2021 is described by ISO as the first international standard on managing psychosocial risks at work. It places hazards in how work is organised, puts responsibility on the organisation’s management system, and treats worker participation as critical to controlling risk (ISO scope statement). A peer-reviewed commentary by Jain and colleagues (2022) confirms its status as the first such standard.

The mechanism: sixty years of work-design research

• Demand and control. Karasek (1979) showed that strain follows from high job demands combined with low decision latitude, and drew the implication that jobs should be redesigned.
• Job demands and resources. Bakker and Demerouti (2007) set out the job demands–resources (JD-R) model, a shared grammar for why work design produces burnout or motivation.
• Primary prevention. The European framework PRIMA-EF codifies a prevention hierarchy in which primary prevention means fixing the work, not hardening the worker.

What the intervention evidence says, in two currencies

Two systematic reviews of organisation-level work-design interventions answer different questions, and are kept apart on purpose.

The bridge from the behavioural evidence

The previous sections showed that individual programmes can improve how people feel without necessarily improving how work goes. The 2024 workplace trial of mindfulness and ACT (Christodoulou et al.) reduced stress but did not reduce work limitations. One reading is that if teaching people to cope does not restore their capacity to work, attention turns to how the work is organised. That is an argument, not a proven result, and it can be read alongside the overview of organisational-level interventions by Aust and colleagues (2023).

What the work looks like

OD interventions are usually grouped by the level of the system they work with.

Two practices worth knowing by name

Process consultation. Schein’s Process Consultation Revisited (1999) describes a way of helping a client system see its own process without the consultant taking over as its expert or its therapist. Dependency, where the client leans on the helper, is the occupational hazard it names. It is the working form of the consulting stance described earlier.

Behaviour-based safety. The observation-and-feedback approach to workplace safety is described in the behavioural section, together with its critics: the worry that “unsafe act” becomes “unsafe person”, and that measurement becomes monitoring (Sutherland et al., 2000). Whenever a practice changes someone’s behaviour at work, the question on the critique shelf applies: who benefits?

A worked example

A mental health service asks for help because “morale is low on the duty team”. Here is how an OD engagement might go:

1. Entry: the service manager wants a wellbeing day. The OD practitioner asks why now. A senior clinician left recently, two serious incidents happened, and there is a new waiting-time target.
2. Contract: agree to spend six weeks understanding the duty system, with anonymised themes going back to the whole team and the manager together. Clinical concerns about individual staff will go through normal line management, not the consultant.
3. Data: interviews with duty staff, managers and partner teams; sitting in on handovers; looking at referral and incident data; noting the consultant’s own experience. (Everyone seems to expect to be blamed.)
4. Sense-making: themes fed back in a facilitated session. Working hypothesis: since the incidents, decisions are pushed up to the senior on shift, who is now overloaded. The waiting-time target rewards closing calls quickly, which cuts across careful risk assessment.
5. Action: the team and manager redesign the duty rota and decision rules, set up a fortnightly case-discussion group, and take the target conflict to senior management.
6. Review: three months on, look at escalations, staff experience and service-user feedback. Then the consultant steps back.

The wellbeing day may still happen. But the work addressed structure, task and anxiety together, and the team did most of the thinking.

What carries over, and what to put down

A summary of the crossings made throughout this page. Cilliers and Henning’s study of clinical psychologists moving into OD describes this transition from the inside.

Ethical questions that become sharper

• Dual roles. If you are a clinician in the same organisation, can you be its OD consultant? Who knows which hat you are wearing?
• Consent. The sponsor’s permission is not the same as each person’s consent. What happens to people who would rather not take part?
• Power. You are often paid by those with most power. Whose interests does the work serve? Do not let psychological language hide structural problems such as unsafe staffing, discrimination or unfair pay.
• When clinical need appears. Someone discloses distress during an interview. You are not their therapist. Know in advance how you will signpost or refer.
• Culture. Most OD theory comes from North American and European institutions. Ideas about authority, voice, hierarchy and emotional expression need translating, not just applying, in other settings, including much of Asia. Peer-reviewed non-Western OD literature was not found in the sources behind this page. The one Asian practitioner source used here is a grey-literature survey (ICF Singapore, 2020).

What the evidence says

Compared with clinical research, OD’s evidence base can feel thin. That is partly because organisation-level change is hard to study. It is also because parts of the field have not tried very hard.

OD in general

Robertson, Roberts and Porras (1993) meta-analysed 52 evaluations of planned organisational change. They found that interventions did change work settings, though modestly, and that changes in individual behaviour were associated with changes in organisational outcomes. But not all OD helps. An earlier large-system study by Porras and Wilkins (1979) found mostly negative effects on attitudes and behaviour. More broadly, an overview of 52 systematic reviews (Aust et al., 2023) found moderate or strong evidence for some kinds of organisation-level intervention on the psychosocial work environment, health and retention, but not for others.

Appreciative inquiry

Mostly case studies. Bushe and Kassam’s meta-case analysis found that 7 of 20 published AI cases (35%) showed transformational outcomes (see section 11).

Systems psychodynamics

Strong as theory, history and rich case description: Menzies Lyth’s study, institutional consultations and recent ethnographies such as Fitzsimons et al. (2024) on “defensive organising”. Weak on independent, comparative evidence that it improves organisation-level outcomes. Geldenhuys (2022) notes that much of the evaluation comes from inside the same practitioner community.

Thin in 1977, thin in 2013

The diagnosis has hardly changed in almost forty years. Margulies, Wright and Scholl (1977) noted that there was an abundance of OD literature but very little research on its effects that could withstand rigorous testing; research existed on “only six interventions”. The first genuine meta-analysis, by Neuman, Edwards and Raju (1989), found effects on satisfaction and other attitudes. It is not evidence about performance. Barends, ten Have and Houtman (2013) reviewed the field again and concluded that “few well-controlled studies exist to evaluate effective organizational change”.

That is not a scandal. Different kinds of evidence answer different questions. Conceptual work and process evidence are OD’s strongest currencies, and controlled evaluations of effect its weakest. The honest posture is to say which kind of evidence you are holding, not to inflate the third kind or to let folk statistics stand in for it. Barends and colleagues’ companion paper (Difficult but doable) sets out what evidence-based change management would involve.

Effect sizes, and what each one measured

Numbers appear throughout this page. This table gathers them with the outcome each one is about. A number without its outcome label is easy to misuse.

Pooled effect sizes for Neuman et al. (1989) were not read in full text, so none is quoted. The table gives its scope only.

Getting started

A first reading path

1. The Unconscious at Work (Obholzer & Roberts): the gentlest bridge from clinical work, with examples from health and social care.
2. Schein on process consultation: the consulting stance.
3. Block, Flawless Consulting: contracting and feedback, practically.
4. Weisbord’s six boxes and Schein on culture: how to look at a whole system.
5. Bushe & Marshak on dialogic OD: the other half of the field.
6. Menzies Lyth (1960) and Jaques (1995) read together: the classic case and the strongest critique from inside the field.

Learning by experience

• A group relations conference (Tavistock’s Leicester Conference, or conferences run by sister institutions in many countries) is often where the ideas make sense from the inside.
• Practise contracting on small pieces of work: a team away-day, a meeting review. Get supervision from someone experienced in OD as well as clinical supervision.
• Professional communities: the OD Network for mainstream OD, and ISPSO (International Society for the Psychoanalytic Study of Organizations, founded 1983) for the psychodynamic strand.

Habits worth keeping

• Ask for the source. The four exhibits on this page (change failure, the positivity ratio, the generative-AI figure, a correction) are practice runs.
• Keep outcome labels attached to numbers. The effect-size table shows how.
• Try both lenses. One team problem, two lenses is a short exercise in holding two explanations at once.
• Look at the map. The concept map shows how the ideas on this page connect.

Glossary

The first thirteen entries compare OD terms with their clinical cousins. The later entries cover the newer sections and each give a source and a “don’t confuse with” line.

Client system

The people and units the work is meant to help. Compare: the patient, but plural and often not the one who asked for help.

Sponsor

The person who commissions and usually pays for the work. Compare: the referrer, but with far more say over the work.

Contract / contracting

The agreed purpose, boundaries and terms of the work, renegotiated as needed. Compare: the frame.

Diagnosis / discovery

Gathering and making sense of data about the system, usually with it. Compare: assessment and formulation.

Working hypothesis

A provisional explanation offered to the system to test. Compare: a formulation or interpretation, but shared and revisable.

Intervention

Anything the consultant does that affects the system, including asking a question. Compare: broadly the same idea.

Process vs content

How a group works versus what it works on. Compare: process and content in therapy.

Primary task

What a group or organisation must do to fulfil its purpose. Compare: the therapeutic task.

Use of self

The consultant’s reactions and presence as instruments of inquiry. Compare: countertransference.

Social defence

A structure or routine that protects people from the anxieties of the work. Compare: defence mechanisms, held by the system.

Psychological safety

A shared belief that it is safe to take interpersonal risks in a team. Compare: conditions for a working alliance.

Action research

Cycles of planning, acting, observing and reflecting with the people involved. Compare: reflective practice, done collectively.

Stakeholders

Everyone affected by or able to affect the work. Compare: the family and wider network around a patient.

Basic-assumption mentality

Bion’s term for a group acting as if it had met for another purpose than its task: dependency, fight/flight or pairing. Source: Bion (1961). Don’t confuse with: “groupthink”, a different idea from social psychology about conformity in decisions.

Therapeutic cohesion

The sense of belonging and mutual trust in a therapy group, associated with better therapy outcome (r = .26 across 55 studies). Source: Burlingame et al. (2018). Don’t confuse with: team cohesion at work, which serves a task and has not been shown by this evidence to improve performance.

T-group (training group)

An unstructured group in which members learn from feedback on their own here-and-now behaviour, central to early OD. Source: Highhouse (2002); Campbell & Dunnette (1968). Don’t confuse with: a therapy group, which has a clinical contract and clinical responsibility.

Casualty (encounter-group research)

A participant with an enduring, significant negative change attributable to the group, at least eight months on. 16 of 210 (8%) in the study that defined it. Source: Lieberman, Yalom & Miles (1973); Roback. Don’t confuse with: any negative change at all (about 16% in the same study). The two figures answer different questions.

OB Mod

Organisational behaviour modification: an antecedent–behaviour–consequence approach to measurable work behaviour. Source: Luthans & Kreitner (1975); Stajkovic & Luthans (1997). Don’t confuse with: behaviour therapy for a clinical problem, or with any management technique that changes behaviour.

Psychological flexibility

In acceptance and commitment therapy, the capacity to stay in contact with difficult thoughts and feelings and act on one’s values. Source: Bond & Bunce (2000); Bond, Flaxman & Bunce (2008). Don’t confuse with: resilience or toughness. Evidence at work concerns distress, not shown performance gains.

Behaviour-based safety (BBS)

A safety approach based on observing, giving feedback on and reinforcing safe behaviours. Source: Carra et al. (2024); Sutherland, Makin & Cox (2000). Don’t confuse with: safety management as a whole, which also includes design, procedures and culture.

Process consultation

Schein’s approach: helping a client system see and manage its own process without taking over as expert or therapist. Source: Schein (1999). Don’t confuse with: process improvement or process re-engineering of workflows.

Cyclical psychodynamics

Wachtel’s integration of psychoanalytic and behavioural thinking, in which inner experience and behaviour sustain each other in cycles. Source: Wachtel (1977). Don’t confuse with: cyclical models of change management.

Single- and double-loop learning

Single-loop corrects errors within existing rules; double-loop questions the rules themselves. Source: Argyris & Schön (1978/1996). Don’t confuse with: feedback in general. Both loops use feedback, but only one questions its own assumptions.

Defensive routines

Habitual ways of avoiding embarrassment or threat that also prevent learning, including making issues undiscussable. Source: Argyris & Schön. Don’t confuse with: excuses. They are patterns in how a group talks, often unnoticed.

Espoused theory and theory-in-use

What people say guides their actions, compared with what their actions show. Source: Argyris & Schön (conceptual). Don’t confuse with: hypocrisy. The gap is usually invisible to the person.

Schein’s three levels of culture

Artifacts; espoused beliefs and values (5th edition wording); basic underlying assumptions. Source: Schein, Organizational Culture and Leadership. Don’t confuse with: stated values, which are only the middle level.

ODV (CIPD)

The CIPD’s abbreviation for organisation development, used to distinguish it from organisation design. Source: CIPD factsheet. Don’t confuse with: the OD Network’s “OD”. It is the same territory with a different emphasis.

Reengineering

A 1990s approach to redesigning business processes from scratch. Source: Hammer & Champy (1993). Don’t confuse with: organisational change in general.

“70% of change fails” (a myth)

Myth. A frequently repeated claim with no valid empirical basis, traced back to an unscientific estimate about reengineering. Source: Hughes (2011). Don’t confuse with: a measured failure rate. See the exhibit in the change section.

Evidence-based management

Using the best available evidence, of several kinds, in decisions, and being clear what kind it is. Source: Barends et al. (2013). Don’t confuse with: trusting the consultant’s best anecdote.

Dialogic OD

A family of OD practice built on the premise that realities are conversations, so changing the conversations changes the system. Source: Bushe & Marshak (2009). Don’t confuse with: “just facilitating conversations”. It is a distinct set of premises.

Complex responsive processes

Stacey’s view that organisational order emerges from local interaction and conversation, without a planner. Source: Stacey (2001). Don’t confuse with: complex adaptive systems used loosely, or with dialogic OD itself.

Appreciative inquiry (4-D)

An approach that starts from what works: Discover, Dream, Design, Destiny. 7 of 20 published cases were transformational. Source: Bushe & Kassam (2005). Don’t confuse with: generic positivity, or with a 35% success rate.

Positivity ratio (a corrected claim)

Corrected claim. The proposal that a 2.9 ratio of positive to negative emotion marks flourishing; its mathematics was shown to be unjustified and a correction was published. Source: Fredrickson & Losada (2005); Brown et al. (2013); correction. Don’t confuse with: positive-emotion research as a whole.

Organisational justice

Perceived fairness at work: distributive, procedural and interactional. Source: Colquitt et al. (2001). Don’t confuse with: “fair process” as a vague ideal. It is measured.

Consent-manufacturing

The critical management studies claim that humanistic management techniques can lead people to want what the organisation wanted anyway. Source: Alvesson & Deetz. Don’t confuse with: manipulation as everyday slang. It is a structural argument.

Agile Manifesto values

The four stated values of agile software development, the first being individuals and interactions over processes and tools. Studies describe a “dark side” where practice departs from the values. Source: Beck et al. (2001); Annosi et al. (2016). Don’t confuse with: any specific method (Scrum, Kanban) or with evidence that agile works.

Design thinking

A family of practices that put users at the centre and work iteratively; coherent but under-specified. Source: Micheli et al. (2019). Don’t confuse with: graphic or product design as a profession.

Psychosocial hazard (ISO 45003)

A feature of how work is organised that can harm psychological health, such as demands or low control. Source: ISO 45003:2021. Don’t confuse with: an individual’s stress reaction.

Primary prevention (work-level)

Changing the work to reduce the hazard, instead of helping people cope with it. Source: Leka et al. (2011), PRIMA-EF. Don’t confuse with: a resilience or wellbeing programme for individuals.

Job demands–resources (JD-R)

A model in which job demands and job resources shape strain and motivation. Source: Bakker & Demerouti (2007). Don’t confuse with: Karasek’s demand–control model, an earlier relative (Karasek, 1979).

“95% of GenAI pilots fail” (a withdrawn claim)

Withdrawn claim. A figure from a preliminary 2025 report that was withdrawn and described as unpublished, non-peer-reviewed work. Source: Challapally et al. (2025), MIT Project NANDA; reported by a secondary forensic account. Don’t confuse with: a measured failure rate.

All sources

Every source linked on this page, grouped for independent reading. Book links that go to a library catalogue search are marked.