Dialogic Organization Development in Coaching: A Field Guide

Dialogic organization development in coaching examines how people’s conversations shape what they consider true, possible and worth doing. It asks how a different inquiry might open a different action, while keeping technical facts, resources and decision authority visible.

The approach is associated with Gervase Bushe and Robert Marshak. This guide explains its main distinctions, develops a complete fictional inquiry and trial, and provides practical contracting, conversation and review tools. The case includes disagreement about the new language, a failed handoff and an update that arrives late.

Start with a mindset, not a meeting format

Bushe’s chapter prepared for the 2026 edition of Practicing Organization Development emphasizes that diagnostic and dialogic are mindsets rather than lists of methods. The same method can be used with different assumptions, and practitioners may combine the two orientations.

A diagnostic inquiry might compare an actual process with an explicit requirement. A dialogic inquiry attends to how participants make meaning and how their interactions create possibilities for action. Asking everyone to discuss a preselected answer does not automatically make the work dialogic.

For example, one function says, “The workshop blocks every request.” Another says, “Sales treats questions as approvals.” A coach can examine how those accounts organize the next conversation: people defend, withhold uncertainty or seek a senior ruling. The coach can also ask for actual records. A date either was promised or was not; making room for multiple interpretations does not make that factual distinction disappear.

Do not turn the contrast into “data versus people.” Data can become material for inquiry, and an open conversation can reveal a missing technical check. The practical question is what this situation requires, including what a preferred method might overlook.

Inquiry can change what happens next

Questions do more than collect an account. “Who failed to follow the process?” directs attention toward individual compliance. “When does a request become a commitment?” makes another issue discussable. Neither question is neutral in its effects.

A coach should therefore offer their own framing for examination. Ask participants what the question assumes, whose work it includes and what it leaves out. If a colleague says the framing ignores a real workload problem, investigate that concern rather than naming it resistance to dialogue.

Our active-listening guide can support checking meanings. Listening is not a promise to adopt every suggestion; it should make the suggestion and its implications understandable before a decision.

Understand generative images, emergence and purpose

A generative image opens an attractive new possibility

In his 2013 theory-of-practice paper, Bushe describes generative images through two features: they make previously unseen actions possible to consider, and people are attracted to using those possibilities. An image may be a phrase, idea, story or object, rather than a literal picture.

The useful test goes beyond whether the group likes a slogan. “Work together better” may attract agreement while changing nothing. “Check the promise while it is still a question” might lead sales and workshop staff to discuss uncertainty before anyone offers a date. That is a candidate generative image if it changes their options and they want to act on it.

It remains a candidate. One colleague may hear an extra approval gate; another may hear a way to avoid rework. Explore those meanings rather than defining the phrase so tightly that participants must accept the coach’s interpretation. Do not require a metaphor when ordinary language works better.

Emergence does not remove the need for resources

An emergent approach leaves room for possibilities that were not fully specified at the start. Participants can discover a useful connection or organize an action the sponsor did not design. It does not follow that they have permission to spend money, alter contracts or displace other people’s work.

Bushe’s 2021 account of generative change distinguishes an ongoing purpose from a fixed future solution and emphasizes learning through experiments and leaders’ work after ideas emerge. This is a conceptual account, not a controlled trial showing that every emergent process outperforms planning.

A purpose such as “help colleagues make commitments they can explain” can allow several approaches. “Install this intake form by Friday” specifies a solution. Either may be appropriate, but the second should not be presented as an open invitation to decide whether the form is needed.

Distinguish an open answer from an open implementation route

The 2026 chapter distinguishes high-engagement work, in which stakeholders contribute to accomplishing an existing change vision, from generative work guided by shared purpose and self-organized initiatives. Explain which kind of influence is available.

If a required security control must be implemented, people might influence the training, sequence or support. They cannot decide to remove the requirement merely because the meeting invites creativity. An honest boundary allows a more useful conversation than an invitation whose apparent freedom is withdrawn later.

For wider stakeholder relationships, see our systemic team-coaching guide. A leader’s account alone does not establish what the whole system needs.

Read change research without turning it into a guarantee

Hastings and Schwarz’s study, published in the 2022 journal issue, examined 79 change cases through 47 narrative interviews. The researchers constructed and coded event timelines, checked reported success against evidence of changed organizational states, and associated movement between diagnostic and dialogic processes with successful outcomes.

The retrospective case design supplies useful evidence about patterns and leader inquiry. It is not random assignment of organizations to coaching methods, and its sample percentages should not become a client’s predicted chance of success. Multiple perspectives were available for some cases, not every case.

For practice, the study supports taking both operational evidence and participants’ accounts seriously. It does not tell a coach to abandon an investigation whenever someone dislikes it. The next inquiry should address the actual unresolved question.

Contract for real influence before inviting a group

Write a short working agreement with the sponsor and explain it to participants. The ICF Code of Ethics calls for clarity about roles, responsibilities, confidentiality and information exchange. An OD or facilitation role should be explained rather than silently added to an individual coaching engagement.

1. Purpose: What is worth investigating, and why now? Avoid promising an answer the inquiry has not produced.
2. Influence: What may participants propose, try or decide? Which requirements and decisions belong elsewhere?
3. Participation: Whose experience and labor matter? Explain how people can contribute, decline a question or raise a concern outside the group.
4. Information: Who receives raw accounts, de-identified themes and the final decision? Do not promise anonymity that the group size or stories cannot support.
5. Resources: Who can approve a trial, and how will the people doing the work accept its load?
6. Closure: When will participants learn what happened to proposals? Who maintains the work after the coaching or facilitation ends?

The HSE Management Standard on change emphasizes timely information, consultation, opportunities to influence proposals, timetables and support. A workshop does not by itself establish that these needs have been met.

Where the sponsor wants endorsement of a settled decision, name that honestly. Where a complaint requires a formal route, preserve it. A coaching conversation should not become a substitute investigation or a condition for someone’s concern to be taken seriously.

Complete fictional example: when “cleared” means different things

This invented case concerns an equipment distributor’s routine customer enquiries. Facilitator Nora contracts for a 75-minute inquiry on 6 November 2026 and a 45-minute review on 19 November. The six internal participants are Sela, the operations director; Amir in sales; Kat in the workshop; Jo in dispatch; Rory in service; and Mei in stock control.

Sela initially asks Nora to “get everyone behind faster workshop approvals.” In contracting, Nora asks whether the answer is already fixed. Sela acknowledges that neither the location of the delay nor the right change is established. She agrees the purpose is to make commitment status clear before a customer receives a date. Nora is facilitating this group inquiry, not privately assessing the six colleagues.

1. Make the invitation and its boundaries concrete

The invitation explains: “We will examine three de-identified routine enquiries and explore how information becomes a commitment. You may question the framing, pass on an account or send Nora a concern by work email before the session. Sela will receive the agreed themes and decisions, not a confidential individual assessment.” All six confirm they will attend.

Sela can authorize a small trial within existing duties but cannot waive technical checks. Urgent safety-related reports stay on the existing incident route and are outside this trial. The organization’s approved enquiry system will hold customer details and operational records. Nora’s coaching materials will use de-identified examples.

Participants receive the questions in writing as well as hearing them. They can write an account instead of delivering a polished story. Passing is not recorded as opposition to change. A person who needs a confidential formal route can use it without first telling the workshop.

2. Explore actual episodes before proposing the new account

At the inquiry, the three historical records show different states: one request was sent without a recipient acknowledgment; another had a reference number but lacked a serial number; the third had been received, but nobody had specified the next update. The group had called all three “cleared” in internal conversation.

Amir: When I say cleared, I mean I have sent everything I have.

Kat: I hear that as workshop approval. A reference number only means a record exists.

Rory: Customers ask when they will hear again. Neither meaning answers that.

Nora: What changes if sending, accepting the next task and approving a date stay distinct?

Amir says the current pattern encourages him to seek a date quickly; Kat says it encourages her to avoid messages that sound like commitments she cannot make. Mei adds that missing stock information can make both positions reasonable. Their accounts are not merged into a verdict that one function caused every delay.

Nora’s first summary, “The workshop needs to respond earlier,” repeats Sela’s opening assumption. Kat challenges it: “We sometimes cannot identify the item.” Nora withdraws that summary and asks the group to record the missing-information problem alongside response ownership.

3. Let participants challenge the attractive phrase

Kat proposes “a promise under construction.” Amir likes the idea of discussing feasibility before a date is offered. Mei objects that it could imply the group is already promising something. Rory says customers need plain status language.

They keep the practical question “What is the next accepted step?” rather than adopting Kat’s phrase as mandatory vocabulary. That question makes a new action attractive to Jo and Kat: a short exchange that names the next person, what they have accepted and when the sender will hear again. Amir still dislikes the possibility of another gate.

The group considers two alternatives. A mandatory long form for every enquiry would standardize information but add work before the need is known. A short early exchange on selected uncertain requests could reveal what information is needed, but it depends on an actual recipient responding. Neither is declared the proven solution.

4. Authorize a bounded trial with willing owners

Sela authorizes three routine enquiries on 10, 12 and 16 November. Jo agrees to coordinate each exchange. Kat accepts a ten-minute workshop discussion slot on each date; Amir agrees to supply the available item information and keep unaccepted dates out of customer communications. All six accept the 19 November review before the trial starts.

The original trial card reads:

The three ten-minute slots describe Kat’s reserved discussion time only. They are not the whole labor cost of the trial, which includes preparation, coordination, updates and review. No claim of overall time savings follows from that reservation.

5. Keep the actual failure and the late update

On 10 November, Kat accepts the next task after Jo supplies the serial number. She promises an update on 11 November and provides it that day. Her update identifies a further technical question; it does not approve a delivery date.

On 12 November, Kat asks for the missing serial number. Amir writes “workshop approved” in the internal record anyway. Kat notices and explicitly rejects that status. Jo pauses the request; Amir corrects the record and acknowledges that receiving a question was not approval. No customer date has yet been sent for this enquiry.

The serial number is supplied on 13 November. Kat then accepts the next task and promises an update on 16 November. She supplies it that day, identifying a parts dependency. The later repair is preserved separately from the failed first exchange.

On 16 November, the third selected enquiry starts with clear acceptance and an 18 November update date. An urgent report arrives on 17 November and follows the existing incident route. Kat misses the routine update on 18 November. Jo asks on 19 November; Kat provides the update that morning, apologizes for the missed date and explains that the next technical check is still pending. Jo and Amir acknowledge receiving that status before the afternoon review.

The urgent report is not counted as a fourth routine trial request or claimed as a successful trial outcome. Its effect on capacity matters, but it does not erase the missed commitment.

6. Close with a decision that matches the evidence

At review, Sela initially says the group has “eliminated unclear handoffs.” Amir points to the second request. Sela corrects her account: two first exchanges had clear acceptance; one needed a repair. Two update commitments were met on time, and one was late. There is no comparable prior cohort or completed customer-delivery result.

The new question helped participants notice distinctions they had previously collapsed. That is a useful account of this small trial, not proof of organization-wide transformation. Amir still worries about delay, and Kat says current staffing cannot support a ten-minute discussion for every enquiry.

Sela does not authorize a wider rollout. She asks Mei to examine the existing serial-number field and Rory to draft clearer status wording. Both accept those internal tasks and a 24 November discussion. The discussion’s result and any new resource approval remain future. Nora’s engagement ends after the 19 November review; the organization owns subsequent decisions.

A reusable inquiry and decision record

Use this original sequence to prepare your own work. It is a practical aid rather than a validated OD instrument or a substitute for specialist facilitation competence.

1. Collect concrete accounts: “Describe an occasion when this became difficult. What happened before and after?” Separate the record from the explanation.
2. Explore organizing language: “What does this word mean to you? What action follows when you use it? Who might hear something different?”
3. Invite another possibility: “What could become possible if this distinction changed? What might the new account conceal?” Let participants challenge your question too.
4. Compare proposals: Record the purpose served, people affected, resource demand and objections. Attraction to an idea is evidence of interest, not proof of feasibility.
5. Secure actual acceptance: Record the authorized action, willing owner, receiving party, boundary and review. An assigned name without acceptance leaves a dependency open.
6. Review and communicate: Keep the original criterion, actual evidence, unsuccessful attempt and unresolved effects. State which decision was made and what remains pending.

Ask separately whether the conversation changed understanding, whether people acted differently and whether an operational result followed. These can diverge. Participants might understand each other better while the selected action proves too costly or the service remains constrained.

Support inquiry and follow-through in Coachful

Coachful is our #1 recommendation for organizing the coaching resources and agreed discussion that support this work. This is our editorial recommendation as Coachful’s publisher. Software can preserve questions and actions; it cannot guarantee generativity, equal power or confidential group behavior.

Create an original preparation or review worksheet and deliberately assign it through Clients > client > Worksheets & resources. A copied private link does not grant access, and staff preview does not submit a participant’s answer. Explain who will receive any response before requesting it.

If you choose a Community channel for agreed, de-identified themes, check the actual audience. Selected members can define the intended group, but eligible Program or resource audiences, an active Offer grant and moderator access also matter. Removing a directly selected person does not prove they lack another access route. Staff access remains a separate consideration.

Check who can start posts and who can reply. A discussion invitation is misleading if members can only read. Agree another route for sensitive concerns, such as the facilitator’s stated work email; a shared channel is not an anonymous complaint mechanism. The fictional trial above keeps operational customer records in the organization’s approved system.

Use the Coachful features overview to assess the wider workflow. Tell participants where decisions and follow-up will appear, who owns them and when facilitation ends. Do not mistake a posted action for an accepted or completed commitment.

Troubleshoot common design failures

The sponsor has already chosen the answer. Offer an honest implementation consultation or revisit the contract. Do not invite people to discover a solution they are only allowed to endorse.

The new language hides a material problem. Bring the factual requirement back. In a contrasting hypothetical case, a machine fails a required inspection. Renaming its status cannot authorize operation. The technical control must be addressed through the appropriate process.

Participants talk only to the coach. Invite them to ask each other for examples and check interpretations directly. Keep enough structure for access and turn-taking without making yourself the final interpreter of every account.

Nothing follows the event. Check whether anyone accepted the action, whether resources exist and whether the sponsor communicated a decision. More enthusiastic wording will not supply missing time or authority.

Frequently asked questions

Is dialogic OD the same as having a good discussion?

No. It concerns assumptions about how meaning and change develop. A discussion that gathers views for a predetermined answer may be useful, but its purpose and influence should be described accurately.

Can a coach use this perspective individually?

Yes, to explore how a client participates in recurring conversations and which new actions they can responsibly consider. One client’s account does not establish the whole organization’s meaning, agreement or readiness.

Must every inquiry produce a generative image?

Do not force a phrase or declare transformation because one appears. Explore whether participants see a new option they want to use, and test what that option actually requires. A useful factual distinction can be more valuable than an imposed metaphor.

Does emergence mean abandoning plans?

No. Keep participation, requirements and resource decisions explicit while allowing the content of possible action to develop. A trial still needs willing owners, appropriate authority and an honest review.

How should success be reported?

Separate changed understanding, actual behavior and operational outcomes. In the fictional case, two first exchanges were clear, one required repair and one update was late. Those observations support a bounded next decision, not a claim that the organization has transformed.