Conversational Intelligence Three Levels Matrix: Practice

The Conversational Intelligence three levels matrix describes Judith E. Glaser's distinction between transactional information exchange, positional advocacy and transformational exploration. A leader can use the distinctions to choose what a conversation needs next. The useful question is what people understand and can responsibly do afterward.

This guide gives a brief explanation, original planning questions and a complete fictional launch conversation. It uses observable behavior and reported experience. It does not turn colleagues into ranked personality types or claim to read their hormones from a meeting.

Use the three levels as a repertoire

Glaser's published work and 2013 dashboard are the sources for the names below. This compact explanation is followed by an original planning worksheet, not a reproduction of the proprietary dashboard or a formal assessment.

All three purposes can matter. A clear instruction may be exactly what a time-sensitive task needs. Advocacy can reveal expertise and a genuine conflict. Joint inquiry is useful when something is genuinely open. A meeting can move between these purposes without classifying every sentence.

Glaser uses neuroscience language in presenting her approach. A category assigned by a coach is not a measurement of cortisol, oxytocin, dopamine or brain activity. Stay with what was said, what changed and what participants report. Neither a quiet room nor apparent agreement proves trust.

Plan the task before choosing a conversational style

Start with the work: a factual update, a reasoned recommendation, an unresolved design question or a decision explanation. A ninety-minute workshop can consist entirely of repeated arguments. A five-minute factual check can uncover the distinction that makes a later decision possible.

Ask privately: What are we doing now? What does this task require? What change in our interaction would help? These are reflection prompts. If you use colored notes, explain their purpose. Red, yellow and green must not become physiological readings or an employee trust score.

Describe behavior rather than labeling a person. “We have repeated both recommendations; can we check the assumptions behind them?” gives people something to answer. “You are stuck at Level II” can make disagreement sound like a personal deficiency. People may prefer ordinary meeting language or decline the framework.

State what is fixed before asking for ideas. A deadline imposed by a contract may be fixed for the participants, while scope or an exception request remains open. If nothing can change, explain the decision and invite questions about understanding, implementation and unresolved concerns. Do not advertise an open design session to secure compliance with a completed decision.

Prepare a seven-field conversation worksheet

The following original worksheet is a preparation aid. It does not reproduce Glaser's proprietary matrix or score participants. Use one row per important agenda item and adjust the fields to the work. These filled answers come from Theo, a fictional operations leader considering a proposed 29 January 2029 launch.

Keep an evidence column in a working version if necessary. An account leader's recollection, a dated customer requirement and an approved contractual change are different kinds of information. Do not promote one into another because the meeting would otherwise end without a neat answer.

Use a complete six-step meeting sequence

1. State the task and boundaries. Explain who decides, what is open, what is fixed and how contributions will be used. Agree an alternative for someone who cannot discuss an issue in that group.
2. Check the facts. Establish sources, dates and definitions. Mark conflicting accounts and missing information rather than creating a supposed shared fact by majority vote.
3. Hear positions and reasons. Ask what each recommendation protects. Summarize it for correction before looking for a compromise.
4. Find a genuine question. Choose something the group does not yet know. The answer must be capable of changing the proposal.
5. Develop and test options. Generate possibilities, then evaluate them against the criteria. An attractive option is still conditional until its dependencies have been checked.
6. Close responsibly. Record the decision or learning action, its owner, receiving work, unresolved issue and review. A discussion can end with a hold.

Allow pauses and suitable processing time. Offer a written first round where it helps access or reflection. If a participant needs a private route or specialist process, do not make public disclosure the price of being heard. Serious concerns may require an appropriate workplace process beyond the coach's role.

For Theo's forty-five-minute meeting, the agreed agenda is five minutes for purpose, eight for facts, ten for positions, twelve for options and ten for closing. Those blocks total forty-five. They are one original allocation, not a requirement to spend a fixed proportion of every meeting in each level.

Work through a conditional launch option

The bounded fictional case runs from 8–24 January 2029. Lara coaches Theo; Amina leads the account work and Seth the technical work. The 12 January meeting, bounded written checks on 19 January, five-minute returns on 22 January and private closing review had been arranged before this account. That earlier administration and the existing business records are unmeasured here.

They aim to close three planning questions: what customer need is described, what technical dependency must be satisfied, and what authorization route permits release. The agreed audiences may receive selected business planning material, not private coaching content. Theo can prepare an operational recommendation; he cannot change a customer commitment by himself. The proposed launch date, Monday 29 January, is beyond the measured case. No actual launch or changed customer date is claimed.

Lara spends twenty minutes on 8 January preparing and sending the finite plan, including the promised 17 January 17:00 written return. Theo uses ten minutes on 9 January to read it and prepare and send selected invitations to Amina and Seth. Each uses ten minutes on 10 January reading, preparing and sending acceptance. Theo's ten minutes on 11 January cover reading both acceptances and sending the selected agenda to Lara.

On 12 January, Lara spends ten minutes reading the agenda and preparing; Theo prepares for five minutes. All four then attend for forty-five minutes. Seth declines being assigned a conversation-level label and asks for ordinary language. That choice is accepted without treating it as resistance. The original dialogue below occurs within the meeting:

Amina: “We should protect the announced date. Customers have organized work around it.”

Seth: “The full service cannot be supported yet. The second shift has not completed training.”

Lara: “What does each recommendation protect?”

Amina: “A commitment people relied on, not simply a sales target.”

Seth: “A service we can actually operate when something goes wrong.”

Lara: “Which customer requirements and technical capabilities are distinct?”

Amina: “The largest account needs reporting first. Two other accounts need the complete workflow.”

Theo: “Could reporting be considered separately, while the other needs remain visible?”

Seth: “Possibly, but readiness of the reporting screen does not prove its backend can run independently.”

Amina: “And needing reporting first does not mean the customer agreed to a partial release.”

Lara: “What can we conclude today?”

Theo: “There is an option to investigate. Neither its technical independence nor customer agreement is confirmed.”

The factual distinction changes the question. It does not settle the answer. Staff readiness and customer credibility remain legitimate concerns. The partial-release idea is a conditional proposal, not evidence that everyone reached consensus or that the coach transformed the team.

On 15 January, Lara spends fifteen minutes preparing and sending V1 privately to Theo. It incorrectly says that the group entered Level III and approved a reporting-only release. Theo uses ten minutes on 16 January reading it and preparing and sending a correction. The option was conditional; Seth declined the label.

Lara had promised a written return by 17 January at 17:00 London. She instead uses twenty minutes on 18 January reading Theo's correction, revising and sending V2 at 10:00. Her response is seventeen hours late. The apology and correction do not erase the missed deadline:

I missed the agreed return yesterday at 17:00 and am sorry for the delay. My first account also overstated the meeting. The group identified a reporting-only option to investigate; it did not approve a release. I have removed the claim that everyone entered a particular conversation level. Seth's preference for plain language remains part of the record. V2 now distinguishes the option, the technical question and the customer-agreement question. Please use only the selected business account for the checks already arranged.

Theo spends fifteen minutes on 18 January reading V2 and preparing and sending a limited business question to Amina and Seth. It asks what their existing evidence can establish about customer agreement and technical independence. He offers an additional call as an option while retaining the agreed written route. The request does not include Lara's private coaching notes.

Seth spends ten minutes that day reading the request and replying no to an additional live call. He retains the already agreed twenty-minute written check on 19 January and five-minute read on 22 January. Theo spends five minutes on 19 January reading and sending acceptance; Seth spends five minutes reading that acceptance before the written check.

Seth's twenty-minute block includes reading the technical question, checking the relevant dependency and sending the return below. It is a bounded review of existing information, not completion of further technical validation:

The reporting screen is ready, but it relies on the same backend validation as the complete workflow. I cannot confirm a reporting-only release from the evidence available today. Full validation would need its own scope and time. Please keep the partial option conditional rather than describing it as an approved workaround. I am not accepting another call or an additional testing assignment in this return. I will read the short business close on Monday as agreed; no further action is assumed.

Amina uses her fifteen-minute block on 19 January to read the question, check the existing customer record and send a separate return. It does not establish customer agreement to a partial release. She declines presenting it as agreed. No one contacts the customer in this measured case.

On Monday 22 January, Theo spends twenty minutes reading both returns and preparing and sending a business close to Amina and Seth, plus selected learning to Lara. The business close says:

The two returns do not establish an approved partial release. Reporting still depends on validation that has not been completed, and the customer record does not show agreement to a changed scope. I will keep my internal readiness recommendation on hold. This does not change the announced customer date or authorize further testing. Thank you for the bounded checks. Please read this close in the agreed five-minute slot today; there is no request for another meeting, acknowledgment or new task.

Amina and Seth each spend five minutes reading later that day, without further replies. On 23 January, Lara uses fifteen minutes reading the selected learning and preparing; Theo prepares for five minutes. They hold their twenty-minute private review. Two questions are clearer: what customer need was described and what technical dependency remains. The final authorization route and customer agreement are still unresolved. The aim of closing all three planning questions is not met.

Lara spends ten minutes afterward preparing and sending the finite close below. Theo reads it for five minutes on 24 January without acknowledgment. The case ends before the proposed launch date.

Your recommendation now retains both constraints rather than treating an attractive option as approval. You have a clearer account of the reporting need and its technical dependency. Customer agreement and the final authorization route remain open, so the intended three-question closure was not achieved. My earlier report was corrected and my late return remains recorded. This closes our agreed exchange after your scheduled read tomorrow. Any further customer discussion, technical validation or coaching support needs a separate owner, scope and available time.

Make the receiving work visible

The account includes the people who read, check, decline, correct and close the exchange. A sent message alone is not proof of shared understanding; the described reads are actual events in the fictional case, not inferred from a notification.

That is 475 measured person-minutes against 600 allocated, a difference of 125: thirty-five for Lara, forty for Theo, twenty for Amina and thirty for Seth. Meeting participation contributes 220 minutes within that total; 255 are other measured activities. Earlier arrangements, wider business work, further validation and any future customer discussion are outside this account and unknown, not free. The allocation difference is not available practice headroom.

Ask questions that can change your own understanding

A question can sound curious while functioning as an argument. “Do you not think customers deserve better?” seeks agreement with a judgment. “What customer impact concerns you most?” leaves room for information the questioner may not have. Ask what answer could change your recommendation before presenting the question as inquiry.

Acas's practical guidance distinguishes open questions, specific verification and the risks of leading or multiple questions. In coaching, ask one useful question and listen to the answer. A relevant closed question, such as whether a check was completed, can be necessary. Openness is not a rule that forbids precise verification.

Try an original sequence: “Which part is fixed?”, then “What is the source for that?”, then “What remains open?” Follow the answers rather than completing a script mechanically. Someone may identify an obligation that ends one option. That is useful information, even if it disappoints the person who proposed it.

A specialist can hold a firm safety or operating boundary while considering alternatives within it. Willingness to learn does not require abandoning expertise, accepting an unsupported compromise or promising work they cannot perform. Seth's refusal of another call and of unscoped testing both belong in the plan.

Keep advocacy and accountable decisions visible

State a recommendation plainly: the view, reasoning, uncertainty and challenge you want tested. “I recommend waiting for validation because the reporting service shares the dependency; what verified evidence contradicts that?” exposes reasoning. Repeating “what do others think?” without disclosing your own constraint may conceal the information the group needs.

Useful exploration does not transfer authority to the room. A sponsor or accountable leader may decide after hearing alternatives. Explain that process before inviting participation. Retain a dissenting view where appropriate, and say what happens to concerns that the meeting cannot resolve.

Close with practical information: which version, what decision or hold, whose action, what time and which recipient will read or respond. A creative conversation without a usable handoff can create more work. For another practice context, see conversational rituals and trust.

Use the distinctions in writing and review

At the start of an email or document, tell the reader whether it is an update, recommendation or request for contributions to an open question. Include context, scope and a response route that fits the receiving work. A written contribution can be a valid alternative to another meeting.

For a recommendation, identify the reason, alternatives and specific challenge sought. For exploration, give enough background and time for a thoughtful answer. Show how the next version uses the contribution, including a constraint that prevents the preferred option. A long comment thread is not automatically effective collaboration.

Afterward, review concrete observations: Was an assumption corrected? Did an answer reveal a new distinction? Was disagreement heard without assigning a personal label? Did the record preserve a hold or refusal? Ask how participants experienced the conversation and retain different accounts.

A brisk briefing can meet its purpose; an exploratory session may need wider participation or more information. Compare the interaction with its intended task, not an aspiration to maximize a category score. People can question the framework or prefer plain language. Do not make acceptance of the tool a condition of ordinary help.

Keep the practice record separate from shared reports

Coachful is our #1 recommendation for organizing selected reflection prompts, client practice actions and review. A private client workflow can support preparation and follow-up. Keep private coach-team review notes separate from feedback sent to the client and from a selected business account sent to colleagues.

Choose recap sharing, approval and session-type settings deliberately if using recaps. Those controls do not supply recording permission, a training license or agreement to post a team conversation in a community. The demonstration community image shows a product area, not publication of this case. Compare Coachful's features against the actual intended audience and work.

Use a limited action and review date rather than uploading unnecessary organizational records. A marked-complete task does not prove that a customer agreed to a release. For broader practice, see the communication skills guide and choose only the prompts relevant to the client's purpose.

Frequently asked questions

What are Glaser's three levels?

The names are transactional, positional and transformational. This guide uses them briefly to consider the purpose of interaction, followed by original practice questions and examples.

Is Level III always preferable?

No. The task may need a factual instruction, a reasoned challenge, an open design question or a sequence of all three. Useful information exchange is not a lesser kind of person.

Can a coach infer cortisol or oxytocin from a meeting?

No. These conversation labels do not measure hormones or brain activity. Discuss observable behavior and the participants' accounts without presenting a physiological diagnosis.

How can I invite a more exploratory discussion?

Clarify what is genuinely open, who decides and what evidence could change your view. Ask a question you can learn from, then show how the answer changes the next version or decision.

Does joint inquiry require consensus?

No. Exploration can inform an authorized decision, a refusal or a hold. Explain the decision process in advance and preserve unresolved concerns rather than describing participation as approval.