Immunity to Change Assumption Test: SMART Design

An Immunity to Change assumption test asks what happens when a client acts differently in a particular situation. The immediate purpose is learning about a prediction, rather than proving that the person has become confident, courageous or better at the goal.

A useful design identifies what the client expects, what they will actually do, which consequences are acceptable and what evidence could change their understanding. It also allows an awkward result, a flawed attempt or a decision not to proceed.

This guide explains the published SMART criteria, then follows a fictional planning manager through three attempts and a completed review. The emphasis is experiment design and interpretation. For the earlier mapping work, see the Immunity to Change coaching process.

Know which SMART criteria you are using

Robert Kegan and Lisa Laskow Lahey’s Immunity to Change uses S-M-A-R-T in the HarvardX book excerpt, printed pages 261–262. S-M concerns safe, modest scope; A concerns an action feasible soon; R-T concerns a research stance and a test of the assumption. Near-term timing sits under actionable.

This differs from the familiar goal-setting acronym. The original excerpt does not require an additional final S. If a practitioner uses “SMARTS,” ask which adaptation they mean rather than presenting it as the book’s exact lettering.

The distinction is about purpose: action generates information relevant to a particular assumption. A completed task alone does not show what has been learned. The book’s later guidance distinguishes observations from interpretations and treats conclusions as provisional.

The case and practical records below are original illustrations. They do not reproduce the authors’ cases, establish a treatment method or demonstrate that this framework causes developmental change.

Connect the assumption with the goal without collapsing them

In a Harvard Graduate School of Education interview, Kegan describes big assumptions as beliefs held as certain truths, and testing as a way to inquire whether they need modification. That is the developers’ account of the approach, rather than proof that every stalled goal has this cause.

A manager may want to clarify work earlier but also want to avoid looking uninformed. The assumption “If I ask, they will stop trusting me” can help explain that tension. It does not establish that the workplace is safe, the question is appropriate or the manager already has every relevant skill.

Keep three different statements visible:

• Goal: what the client wants to do more effectively.
• Protective concern: what the client wants to avoid losing.
• Assumption: the relationship the client predicts between an action and that feared consequence.

A technical knowledge gap, an impossible workload or an actual punitive environment requires its own response. Do not use the model as a universal explanation for behavior or infer a hidden commitment that the client does not recognize.

Turn a broad verdict into an observable prediction

“I will be exposed as incompetent” is difficult to examine because almost any uncomfortable moment can be interpreted as exposure. Ask what the client expects to hear or see, from whom, in which situation and over what period.

Then check whether the proposed action genuinely differs from the usual response. Reading a question silently at home does not test what happens when the client asks it in a meeting. Conversely, asking a colleague privately does not establish what a large public group would do.

Some predictions contain an observable part and an unknowable part. “Rhea will think less of me” cannot be read directly from a pause. “Rhea will tell me the question is obvious and remove my planning responsibility that week” supplies events the client can record, while still leaving other interpretations open.

Do not replace an untestable fear with a dangerous test. If the feared consequence cannot responsibly be risked, choose another question, another context or no experiment. A coach’s enthusiasm does not determine what is acceptable for the client.

Complete the first design before the opportunity arrives

Arun is a fictional programme manager. He wants to clarify ambiguous plans instead of quietly choosing an interpretation. He often avoids asking because he wants to preserve colleagues’ confidence in him.

Coach Mara and Arun agree three sessions on 2, 9 and 23 November. They will design and review modest learning tests within ordinary work. Mara is not assessing Arun’s competence for the employer, and his private assumption log is not part of a sponsor report.

Arun’s narrow prediction is: “If I ask Rhea a relevant scope question that the briefing has not answered, she will say I should already know and stop asking me to own that week’s planning item.” Rhea leads a routine internal planning meeting, has previously answered practical questions constructively and is not conducting a performance review.

All people, dates, workplace events and results are invented. No real colleague is being observed or contacted through this article.

Check safety and information value separately

Arun judges this routine question and the possible ordinary discomfort acceptable. Mara asks him to explain why rather than declaring the meeting safe for him. He can pause if the meeting’s purpose changes, a confidential matter appears or the question no longer belongs there.

Safety is not the same as informativity. An entirely reassuring private rehearsal might be comfortable but produce no evidence about the actual prediction. A reckless public challenge might produce a strong reaction but still tell little about a normal relevant question.

Arun does not ask Rhea to act friendly so his test will succeed. He does not conceal relevant information, manipulate a colleague or seek an opportunity to embarrass someone. This is a change in his own ordinary work behavior, not a covert rating system for colleagues.

He also decides what would challenge the prediction: a practical answer followed by continued planning responsibility. A neutral or mixed response is allowed. If every possible outcome would count as rejection, he and Mara need to revise the design before trying it.

They agree that personal discomfort is information about Arun’s experience, but not direct access to Rhea’s judgment. A tense stomach does not establish that the feared loss of trust happened.

Run the first test and record the actual result

Arun reads the current briefing before the 4 November meeting. It names a pilot but does not identify its customer group. He asks: “Does this pilot cover existing account customers, new enquiries, or both?”

Rhea pauses, looks at the document and says: “Existing account customers for this version. Please make that explicit in Friday’s plan.” She keeps Arun as the owner. He submits the revised planning item on 6 November, and no one else is assigned to replace him that week.

Arun writes his first note immediately after the meeting:

“I asked the planned scope question. Rhea paused and checked the document, then specified existing account customers and asked me to make that explicit. I felt exposed during the pause. I interpreted it as annoyance at first. I do not know why she paused. I remained the planning owner.”

After completing Friday’s item, he adds the ownership observation. The initial note and later update remain distinguishable. He does not write that Rhea respected him more, that all colleagues welcome questions or that the coaching changed the team’s culture.

The test supplies one relevant counterexample to the exact prediction in this setting. It does not examine a different colleague, an already answered question, a board presentation or future promotion prospects.

Use the first review to improve the next design

At the 9 November session, Mara asks what Arun did, what occurred and what he inferred. They check the briefing’s unanswered scope and the actual ownership window before interpreting the result.

Mara: Which part of your prediction occurred?

Arun: Neither the statement that I should know nor removal of my planning item. I did feel uncomfortable.

Mara: What does that suggest, and what does it leave open?

Arun: In that routine meeting, a genuine unanswered question helped define my work. I cannot conclude that timing and preparation never matter.

Mara: What would another useful question examine?

Arun: Whether I can ask about a genuinely unresolved date while the meeting is moving quickly, without assuming a brief reply means rejection.

They retain the same limits: routine work, relevant unanswered question, current reading, no crisis and no withheld information. The next possible meetings are 11 and 18 November. Arun can decline either opportunity; two more attempts are not a compliance target.

The review has changed the question. It is no longer “Can I get Rhea to reassure me?” It concerns what actually happens when a prepared, relevant clarification is needed. A second test should answer that new question rather than merely collect another pleasant example.

Keep a flawed attempt in the record

On 11 November, Arun asks which day the draft is due. Rhea says: “Wednesday, as set out in the briefing.” Arun later checks and finds the due date clearly stated in a section he missed.

He initially writes, “The assumption was true after all.” His same-day design check identifies the problem: the planned condition was an unanswered question after reading the current information. That condition was not met.

He corrects the entry without deleting the event: “I asked an already answered question. Rhea referred me to the briefing. I had missed the section. This attempt does not test the planned prediction about a genuine unanswered question.”

There is a practical responsibility to address. Arun reads the full briefing and posts the required planning update on time. He does not blame Rhea, ask Mara to make the result positive or pretend the attempt never happened.

The event may inform a different question about preparation, but it cannot be silently combined with the valid test as equivalent evidence. This is why the actual action and context belong in the record, not only the original plan.

Examine a useful result with an awkward response

Before 18 November, Arun reads the updated briefing and confirms that the handover date between two internal groups is still unresolved. The routine meeting is running late, but it has not become an emergency or confidential review.

He asks: “Which handover date should the two groups plan against? The current briefing leaves it open.” Rhea replies: “We haven’t settled it. Bring the two workable dates on Friday so we can choose.” Arun experiences the delivery as sharp. She has not said that he should already know the date.

Arun records the words, his impression and the requested next action separately. He remains responsible for the planning item through 20 November and supplies the two options that Friday. Rhea schedules the date decision with the actual group owners. At the final coaching review, that decision is still open.

The response was not especially reassuring. It also did not meet the predicted competence statement or removal of responsibility. The result supports a narrower conclusion: a relevant question can receive a brief, uncomfortable answer while still producing a useful next step and continued involvement.

It does not prove Rhea’s private opinion. Nor does an options request mean the handover date has been decided. The company’s planning record keeps that decision open with its actual owners.

Complete the evidence and interpretation record

At the 23 November session, Arun describes three attempts, two of which met the planned condition. He does not report “two out of three successes” or a reliability percentage. These are selected everyday observations with different conditions, not a representative experiment or a measure of leadership ability.

His revised assumption is: “In these routine discussions with Rhea, asking a relevant unanswered question does not necessarily remove my responsibility. Preparation and timing matter, and a brief answer is not enough to infer loss of trust.”

He chooses an ordinary practice of checking the current briefing and asking necessary clarifications. He is not committing to ask a question in every meeting or to feel calm. The handover decision remains company work, and the three-session coaching engagement closes.

Choose a conclusion proportionate to the evidence

Arun cannot tell whether Rhea privately revised her opinion of him. Keeping him as the planning owner might reflect her confidence, an existing allocation or several considerations together. The record does not distinguish those explanations. It does distinguish continued responsibility from his prediction that she would remove it within the stated week.

That narrower conclusion is enough for his next choice. He does not need certainty about every private judgment before asking a necessary work question. He can prepare, ask and respond to the actual answer, while retaining the option to seek support if an explicit adverse consequence occurs.

He also does not treat careful preparation as a new rule that he must understand everything before speaking. The condition was to check the current briefing for the particular answer. If the document itself is confusing, that confusion can be the relevant clarification. The point is to define the question honestly rather than arrange conditions that can only confirm a preferred belief.

A short worksheet for a new test

1. State the assumption and observable prediction. Include the context, person and observation period. Separate a feared feeling from an external consequence.
2. Describe the changed action. Write what you will actually do differently and the conditions that must be present for the test to mean anything.
3. Check consequences and consent. Identify what you are willing to risk, what is outside scope and what would make you stop. Do not make colleagues responsible for producing a preferred reaction.
4. Plan a brief record. Capture your action, others’ words or actions, your experience, missing information and any departure from the plan. Do not record people covertly.
5. Compare with the prediction. Ask what is supported, challenged or still unknown. Identify a plausible alternative interpretation before deciding whether another test is useful.

This original worksheet is an application aid. It is not a diagnostic instrument or an instruction to keep escalating the stakes. For the separate historical inquiry into why an assumption became convincing, see the biography of a big assumption.

A contrasting result can require protection rather than another test

Fictional analyst Jo asks an ordinary clarification and receives an explicit threat that further questions will affect her assignment. That event differs from Arun interpreting a pause or a brief tone. The words and the actual context matter.

Jo and her coach stop proposing further workplace experiments. She keeps an appropriate factual record and seeks advice through a suitable organizational or qualified route about her actual situation. The coach does not decide the legal meaning of the statement or encourage her to provoke another response.

A model of internal assumptions must leave room for external constraints. Repeated exposure is not a duty owed to the method, and caution is not automatically evidence of a hidden resistance to growth.

Prepare and review the learning in Coachful

Coachful is our #1 recommendation for organizing an agreed assumption test and its review. Forms and worksheets can hold a permitted learning summary, while client tasks and conversations keep the selected next step visible.

Mara delivers Arun a short worksheet through the client resource workflow. It includes the prediction, conditions, actual action, observations, interpretation and next question. Arun submits a nonconfidential summary. Mara works alone and is the sole authorized coaching staff audience; the log is not placed in a group or sent to his employer.

They discuss each submitted record at the agreed session. If they choose an action, they explicitly record it through the client plan workflow. A submitted answer does not automatically become a task, notify Rhea or prove that an assumption was tested.

Arun uses the private Coachful conversation for scheduling and brief preparation updates. Mara replies about those arrangements there, with interpretation reserved for the booked sessions. Work questions and formal planning decisions remain in the company’s channels. The agreement includes no emergency monitoring or between-session workplace advice.

At closing they retain the mixed learning summary, including the flawed attempt, and end the service as agreed. The demo My Coaching image illustrates preparation and next actions, not a scientific test result. Explore Coachful’s features for the supporting workflow.

Troubleshoot before repeating

The client changes the action. Record the actual action and check whether it still addresses the prediction. A different action may be useful, but it should not inherit the original design’s meaning without examination.

There is no clear response. Keep the result uncertain. Another person’s silence is not automatically rejection, approval or deliberate concealment. A direct work clarification may be more appropriate than another psychological interpretation.

The coach wants a disconfirming result. Return to learning. A supported concern or a narrower conditional assumption is a legitimate result. Do not edit out an inconvenient event.

The first test feels easy. Ask what remains worth learning before increasing scope. More risk does not necessarily yield better information, and an ordinary practice may now serve the goal better.

Frequently asked questions

Does the original source use SMART or SMARTS?

The inspected book excerpt uses S-M-A-R-T, grouping safe and modest and placing near-term timing under actionable. Ask which adaptation is intended if another label is used.

Must a test disprove the assumption?

No. It may support, qualify or challenge the prediction, or remain inconclusive. The client needs an honest account of the evidence rather than a prescribed positive story.

Does completing the action mean the test worked?

Completion alone is insufficient. Check the actual conditions, the information collected and what it suggests about the particular assumption. Arun’s already answered question did not meet his planned condition.

Can a coach guarantee a test is safe?

No. Discuss appropriate scope and acceptable consequences with the client. Avoid important risks and stop or change the approach when circumstances make it unsuitable.

When should testing end?

When the learning serves the current purpose, another test has no useful question or the context is unsuitable, the client can stop. Ending the agreed coaching service does not require the complete disappearance of discomfort.