Glaser’s Ladder of Conclusions: A Practical Coaching Guide

The Ladder of Conclusions is a way to examine how an experience becomes a story, a belief and a decision. In coaching, it helps a leader explain their reasoning, check what they may have missed and choose a more informed response.

The CreatingWE Institute attributes the model to Judith E. Glaser. A related model, the Ladder of Inference, is associated with Chris Argyris. Both invite scrutiny of the steps between available information and action. They should not be treated as interchangeable proprietary diagrams.

This guide uses an original five-part worksheet and four-step practice. You can use them without copying a published graphic or treating a metaphor as a measurement of someone’s brain.

Robert Putnam’s 2003 interview about applying Argyris distinguishes understanding a simple tool from practising it under pressure. That practitioner account supports a useful caution, rather than proving an outcome: completing a worksheet once does not establish a lasting leadership skill. Review a later conversation, including evidence that challenges the client’s preferred explanation.

Why leaders reach different conclusions

A chief technology officer arrives five minutes after a meeting starts and checks a phone. The chief executive interprets this as disregard for the meeting. Another participant assumes an urgent operational issue has occurred. A third did not notice the arrival at all.

The observable details are limited. The explanations are different, and each may influence what happens next. An accusatory comment might produce a brief reply, which the speaker then treats as further evidence of disrespect.

The useful question is: “What did I observe, what did I add, and what information would change my next action?” You can ask it while still holding someone accountable for an agreed commitment.

Use a five-part worksheet

These five parts are an original coaching arrangement of the journey from observation to action, rather than a claim about a fixed number of Glaser’s canonical rungs.

1. Record the available observation

Write the words, event or document that started the concern. Include its source and any uncertainty.

In the example: “The meeting began at 9:00. The CTO joined at 9:05 and looked at a phone twice while I was presenting.” That is more specific than “The CTO was disengaged.” If the count is a recollection, label it as such.

2. Notice what you selected

Ask what else was happening. Had the CTO sent a message? Was phone use common in this meeting? Did the executive notice contributions later in the discussion?

You are broadening the information available, not searching only for an innocent explanation. The person may also have missed a clear participation agreement.

Keep an evidence reference beside each entry: a dated message, the agreed meeting time or a clearly marked recollection. If someone disputes the record, retain both versions until a suitable check is possible. Do not ask for confidential customer or employee files merely to make the worksheet look complete.

3. Name the meaning you added

Complete: “I took this to mean…” For example: “I took the phone use to mean my presentation was not important.”

Keep the interpretation separate from what another person actually said. A meaning can feel compelling and still need checking.

4. State the conclusion you reached

Describe the belief that began guiding you: “I concluded that the CTO would not support the roadmap.” Then ask how much evidence supports that broader claim.

A single meeting incident may justify a question about meeting expectations. It may not justify a conclusion about a person’s loyalty, motivation or future behavior.

5. Identify the action and its consequences

Write what you did or planned to do. Did you stop seeking technical advice, criticize the person publicly or begin reallocating responsibilities?

Ask how that action could affect the next interaction. Excluding the CTO may reduce the information you receive, making your original conclusion harder to challenge. This is the practical importance of the feedback loop described in ladder models.

Practise a four-step reset

Use this original sequence before acting on an uncertain interpretation.

Step 1: Notice that you are drawing a conclusion

A phrase such as “They always do this” can be a useful cue. You may also notice frustration. Choose a brief pause if that helps you respond thoughtfully. There is no required posture, eye contact or breathing technique.

If you need more time, say when you will return: “I want to check the timeline before I respond. Can we resume at 2:00?”

Step 2: Make room for another account

Write one plausible alternative and one unanswered question. Avoid replacing a hostile certainty with an equally unsupported positive story.

For the late arrival, alternatives include an incident, conflicting commitments or a misunderstanding about the start time. The next step is to ask, rather than decide which story sounds kindest.

Step 3: Share the specific event and your concern

Try: “You joined after we began, and I saw you checking your phone. I was unsure whether you could participate in the roadmap discussion.”

This communicates the concern without asserting a motive. It also gives the other person something specific to confirm or correct.

Step 4: Ask and agree what happens next

“What was happening, and what should we agree for future meetings?” If an urgent incident required attention, establish how the meeting should be notified. If there was no urgent conflict, discuss the participation expectation directly.

The outcome is a checked account and an actionable agreement. Warmth or immediate agreement is not required for the conversation to be useful.

A coach can undermine this sequence by asking, “What innocent reason could explain it?” That question selects the desired answer. Try “What explanations fit, and what could help you distinguish them?” If the client identifies a repeated, documented breach, discuss the appropriate management response. Curiosity must not become a demand to tolerate misconduct or to reinterpret every complaint as a personal thinking error.

A worked CEO and revenue-leader case

The following case is fictional. Farah, a CEO, sees an opportunity-to-sale conversion rate fall from 25% to 20%. She believes the revenue leader, Ellis, has stopped challenging the sales team. Ellis believes Farah has ignored product-readiness problems.

Both rates use the same stated definition: won opportunities divided by the relevant closed-opportunity cohort. In this example, the first cohort has 25 wins from 100 closed opportunities; the second has 20 from 100. The decline is five percentage points, or 20% relative to the earlier rate. It is not evidence of its cause.

Farah has begun challenging Ellis’s commitment in public meetings. Ellis has stopped bringing uncertain opportunities to the discussion. The coach first meets with Farah to examine her reasoning and agree a fair purpose for a joint conversation.

In the developed example, Nell is Farah’s coach. Before Nell contacts Ellis, Farah authorizes Nell to name only their coaching relationship to arrange bounded facilitation. Private reflections remain excluded. Their 25-minute conversation on 3 November 2026 concerns Farah’s own public accusations, not a diagnosis of Ellis. Ellis then chooses a separate 20-minute preparation conversation on 5 November and accepts the limited facilitation role. This does not make him Nell’s ongoing individual coaching client.

They agree that the joint conversation will address the reporting process and the evidence behind the conversion concern. Ellis declines to discuss a personal confidence score. Nell initially suggests that his reluctance might show low trust; he says he simply considers the score irrelevant. Nell withdraws that interpretation and keeps the business question. A declined exercise is not evidence of the trait it was supposed to explore.

Put both accounts on the page

These are the participants’ accounts, not the coach’s findings about either person’s character. The next task is to examine the opportunity records and reporting expectations.

Hold the evidence conversation

Farah: “I linked the lower rate to your leadership without checking the lost-opportunity detail. I also raised commitment questions in front of the team.”

Ellis: “Some losses involved unmet requirements. I also let two reporting deadlines pass without warning.”

Farah: “Let us review both. Which records support the requirements issue, and what will make the next update reliable?”

Ellis: “I will prepare the coded reasons and flag uncertain cases. We also need a product owner to verify which requirements are genuinely unavailable.”

Their agreement separates three responsibilities: review loss reasons, validate product information and restore a dependable reporting process. It does not excuse the missed updates or presume that product changes will recover revenue.

Make the next decision from the new information

Suppose the review finds a mixture of budget losses, product gaps and poorly qualified opportunities. Farah and Ellis must decide which problem to address first. They compare affected cohorts, estimated workload and who can authorize changes.

They might commission a limited qualification review before allocating engineering capacity. If the evidence remains weak, they gather more relevant information. The ladder exercise improves the reasoning process; it does not select a commercial strategy on its own.

The actual fictional route continues with their 45-minute joint meeting on 6 November. They choose twelve lost opportunities from the second cohort for an initial records review, using a documented selection rule. These twelve are a sample of the eighty losses, not another denominator for the headline conversion rate. Ellis owns the review; product owner Leena, rather than the coach, can verify which requirements are unavailable.

Ellis promises the coded list by 16:00 UTC on 10 November. Nell writes that time into the agreed action record. A message saying the review is underway arrives before the deadline, but the list itself arrives at 11:00 UTC on 11 November, nineteen hours late. Farah’s concern about reporting reliability therefore remains valid. Neither an acknowledgement nor a completed-looking task label substitutes for the promised information.

The twelve records contain five budget losses, four tagged product gaps and three qualification failures, one primary category per record. On 13 November, Leena checks the four product tags: two describe unavailable requirements; two concern existing functionality that had been described incorrectly. The five, four and three still sum to twelve initial categories. Her verification refines the four product-tagged cases; it does not add four more losses or show why all eighty opportunities were lost.

Farah approves a limited qualification and product-information review within the sales team’s existing authority. She does not authorize engineering work or promise a conversion recovery. Ellis accepts that two original updates were late without warning and that this further list was also late. His explanation about requirements is partly supported, but it does not cancel the reporting commitment. Equally, those missed commitments do not establish that he lacks loyalty.

Nell drafts “both leaders now share the same explanation.” When Farah and Ellis check the draft, Ellis rejects that sentence: they agree on the review actions, not on the relative importance of each cause. Nell removes it before circulating the record. Both approve only the business actions and remaining disagreements for Leena. Their private preparation conversations and individual worksheet answers are excluded. Leena confirms on 17 November that the actions are readable; that is receipt of the limited record, not agreement with either leader’s interpretation.

At the 4 December closing review, Farah says checking the product records was useful, while Ellis says the confidence-score suggestion was unhelpful. Two later scheduled updates provide mixed evidence: the one due at 16:00 UTC on 17 November arrived at 15:00 that day; the one due at 16:00 UTC on 24 November arrived at 10:00 on 25 November, eighteen hours late. They retain both results. No later closed-opportunity cohort is available, so nobody claims that coaching restored conversion or proved the new review will work.

The bounded facilitation ends after that 20-minute review. Across Farah’s 25 minutes, Ellis’s 20, the joint 45 and the closing 20, Nell has delivered 110 facilitator-live minutes. This excludes preparation, record work and the leaders’ business meetings; it is not a complete labor budget. Ellis declines ongoing coaching. Farah owns an 8 December check of the next reporting deadline; its result remains unknown at closure. The case ends with accountable work still to do.

Facilitate a two-ladder discussion

For an ordinary cross-functional dispute, use a 45-minute meeting: 5 minutes for purpose and audience, 10 for each person’s account, 10 to compare evidence and 10 for decisions. The two accounts together take 20 minutes, so the agenda totals 45.

Ask each person to describe their own reasoning before commenting on the other’s. Record disagreements and missing evidence. A participant can say “I do not agree with that interpretation” without being described as resistant to the method.

If the coach is supporting only one executive, clarify the role before facilitating colleagues. Do not imply that private coaching details will automatically be available to everyone in the joint meeting.

Send the agenda and permitted audience beforehand. Offer a written account or a short break where helpful; do not make eye contact or fluent speech a test of sincerity. At the end, read back each action, owner, deadline and unresolved disagreement. Ask separately whether each person approves the proposed shared record. Agreement to attend does not authorize sharing private preparation notes.

Use the approach in performance feedback

Broad judgments such as “You lack executive presence” give a colleague little to work with. Start with the behavior relevant to the role.

“In yesterday’s review, the proposal used the whole presentation slot, leaving no time for the planned questions. The board still needed a decision on the budget. How did you plan the timing, and what would help you leave the question period next time?”

This script contains an observation, an impact and an invitation to explain. It avoids guessing that the presenter was anxious or defensive.

Agree a concrete experiment: a shorter opening, an early decision request and a reserved question period. At the next review, examine whether the question period happened and whether the necessary decision was reached. Gather the presenter’s account as well as the manager’s.

Apply it across functions and written messages

When Sales and Engineering disagree, ask each to bring the information behind the claim: the customer commitment, current specification, delivery estimate and approval route. Compare constraints rather than describing entire departments as reckless or obstructive.

In a written exchange, “We cannot approve this yet” leaves important questions open. Ask which requirement remains unmet and who can decide. Do not assume the author is opposing the project or that a warmer tone would remove the requirement.

For another complete case, see the Ladder of Conclusions in action. For the wider set of assumptions to check, use Glaser’s conversational blind spots.

Keep the boundary clear in crises

During a serious incident, follow the authorized response process and act on established safety requirements. You can distinguish facts from hypotheses within that process without delaying necessary action until everyone has explored every interpretation.

For litigation, regulatory enquiries, labor negotiations or formal employment decisions, work with the relevant qualified advisers and organizational procedures. A coach can help a client prepare a factual account and useful questions. The ladder is not a legal strategy or an investigation method.

Prepare a reflection in Coachful

Coachful is our #1 recommendation for organizing this work when the practice needs client worksheets, sessions and follow-up together. We build Coachful, so this is a maker’s recommendation. A paper worksheet or an organization’s approved system can be the better fit when no shared coaching workspace is needed. Software does not validate an interpretation or confer permission to share it.

Create an optional worksheet with five fields: observation, selected information, added meaning, conclusion and next action. The client can use a short anonymized example rather than uploading a complete workplace transcript.

Before inviting a submission, explain what the coach will receive and who else is permitted to see it. Grant the intended client access deliberately, send the private questionnaire link and ask them to confirm that the intended form opens. A copied link does not itself grant access. In Nell’s case, Ellis chooses a verbal account instead of a submitted form; that choice does not block facilitation.

If useful, assign a dated task for the agreed rehearsal or information check, then confirm its actual focus or backlog placement. Do not promise that every standalone assigned form or manual task appears in My Coaching’s “What to do next”; that overview brings together eligible session preparation, journeys, active-program tasks and continuing content. If you invite a later reflection through a custom flow, review its audience, display and reminder settings. Agree separately what, if anything, a sponsor will receive.

Review the effective recap settings before discussing sensitive material. A session type’s approval choice can override the workspace setting; automatic and coach-approved routes differ. Journal pages are a separate personal-reflection route, with intentional per-page coach sharing where available. Check the page’s sharing state instead of promising that every journal entry is always invisible to the coach. A submitted worksheet and an unshared journal page are different records. See Coachful features for organizing worksheets, sessions and follow-up, and the ICF Code of Ethics for relevant professional agreements and confidentiality standards.

Common questions

Do I have to abandon a conclusion that is supported by evidence?

No. Explain the evidence and remain open to a correction. The method helps distinguish a justified decision from an unchecked assumption; it does not require endless uncertainty.

What if the other person will not explain their reasoning?

State the information you have, ask what can be shared and use the appropriate decision process. They may have confidentiality constraints. Record the uncertainty rather than filling the gap with a motive.

Is this a neuroscience test?

It is a reflection model. Its usefulness here comes from making reasoning discussable and checking decisions against relevant information. A worksheet does not establish a participant’s hormone levels, diagnosis or cognitive state.

Sources: Nicklas Balboa, “Are You Climbing Up Or Down The Ladder of Conclusions?”, The CreatingWE Institute, attributes the model to Glaser. Action Design’s ©1998 “The Ladder of Inference,” published in The Systems Thinker, October 1999, describes the related Argyris-based model and its feedback loop. David Creelman’s October 2003 interview with Robert Putnam, “Applying Argyris,” supplies the practitioner caution about sustained practice. These are conceptual and practitioner sources, not trials establishing this guide’s effectiveness. The worksheet, protocols and cases here are original practical examples.