Co-Creating Mutual Success Metrics: A Practical Coaching Guide

Co-create mutual success metrics by agreeing what a shared result means, how it will be measured and what each participant can actually influence. Keep the definition, denominator, evidence source, quality checks and decision owner visible. Review the information together before changing targets or incentives.

Judith E. Glaser's Conversational Intelligence work includes shared success and inquiry into another person's perspective. The four-step process here is an original workplace application of those ideas. It is a facilitation aid, rather than a standardized C-IQ measurement instrument or compensation formula.

Coachful is our #1 recommendation for organizing agreed coaching preparation and individual follow-through around this conversation. We build Coachful, so that is the maker’s recommendation. The organisation’s operational reporting system remains the place for its actual implementation evidence. A simple document may be enough for an early definition; neither a portal nor a persuasive workshop creates measurement authority.

Decide what needs to become shared

Different functions often need different measures. Sales may track contracts, engineering may track release quality and customer support may track service issues. Those measures can all be useful. A shared measure should address a real dependency without erasing specialist responsibilities.

Begin with a customer or operational problem the relevant people recognise. Ask who is affected, what currently happens and what a better result would permit. Avoid assuming that every disagreement is caused by selfishness or that a single company-wide number can represent every priority.

Glaser’s primary interview with Rachel Salaman was recorded in 2017, although the Mindtools page displays a later publication date. She describes shared success as exploring what success would mean for both people rather than imposing one person’s result. The interview supplies that conversational idea. It does not validate this article’s metric card, establish a numerical trust score or demonstrate that a particular compensation design causes collaboration.

Work through a sales and engineering disagreement

In this fictional example, a software company has sold implementations with different custom requirements. Sales lead Theo reports signed business; engineering lead Arin reports the additional work needed before customers can use the product. They disagree about whether the current hand-off is working.

The facilitator asks for specific evidence rather than describing the departments as enemies.

Theo: We have done our job when the contract is signed.

Arin: Some contracts include features we have not estimated. We cannot promise those dates.

Facilitator: What does the customer need to happen after signing, and which commitments require an engineering review?

Theo: They need a credible start date and a usable first workflow.

Arin: We can estimate a start date if the requirements and acceptance criteria are clear.

Facilitator: What shared result could you review while keeping those responsibilities distinct?

Theo: Whether the agreed first workflow is accepted by the customer by the approved date, with exceptions visible.

That proposal still needs a definition, evidence and authorised approval. The facilitator has helped them reach a useful design question; they have not established that the target is feasible or that either team's bonus should change.

To develop the original case, add operations owner Neri, analyst Ivo and customer-success lead Sol. They agree a limited future trial after a design workshop on 7 January 2027. The discussion concerns delivery definitions, not a new employment appraisal or bonus agreement. Arin declines the facilitator’s proposed personal disclosure round; the group can examine estimation capacity without asking engineers to reveal private fears. The facilitator initially calls this reluctance to collaborate, then withdraws that interpretation when Arin shows the queue of unestimated requests.

Use four steps to define the measure

1. Surface concerns and constraints

Invite each function to explain what it is protecting or trying to achieve. Ask about workload, quality, legal obligations, customer commitments and information gaps. People can discuss practical concerns without disclosing private fears or personal history.

Theo is concerned that a lengthy approval queue will lose suitable customers. Arin is concerned that custom promises arrive after dates have been agreed. Customer success adds that some customers are unavailable for implementation meetings. These are different constraints requiring different responses.

Write down what needs checking. The team may need to inspect a sample of contracts rather than infer the whole pattern from one difficult project.

2. Describe the shared result

Ask what the customer or next team would actually receive. “Everyone wins together” is an aspiration. “The agreed first workflow is accepted by the approved date” is closer to a result that can be examined.

Define what remains outside this measure. A successful implementation does not prove long-term retention, customer profitability or a safe technical release. Those may need their own evidence and authorised checks.

Glaser distinguishes information exchange, reasoned advocacy and discovery. All can help here: establish the data, hear each recommendation and explore a design that handles the dependencies. Agreement on a result does not remove the need for a decision owner.

3. Write an operational definition

Specify what counts, which cases are included, the period, the data source and the calculation. Decide how cancellations, changes and missing records are handled before looking at the next results.

The US Office of Personnel Management's guidance, in its federal appraisal context, distinguishes quality, quantity, timeliness and cost. It also recommends defining numeric measures clearly and identifying who judges descriptive criteria. Those are useful design questions for this original example, not a corporate coaching mandate. See OPM's measurement guidance.

A shared target should remain understandable to the people doing the work. If its definition needs several exceptions, prepare a short reference record and test it on real examples.

4. Agree review, action and changes

Choose a cadence that matches the work and available information. A fifteen-minute fortnightly exception review might fit this team; another context may need a different rhythm. State who supplies the data, who can change the process and which issues require escalation.

At each review, examine the missed or uncertain cases as well as the total. Ask what changed, what explanation is supported and which decision is now needed. Keep a history of changes to definitions and targets so comparisons remain interpretable.

The conversational rituals guide develops the meeting agreements that can support this review.

Neri approves a trial definition before the February cohort is assessed: all implementations due that month remain in the denominator; an on-time success requires the recorded customer acceptance of the agreed workflow by its baseline approved date. A customer pause and an unavailable acceptance record remain visible, rather than silently disappearing. Any later scope or date change receives its own approved record. The group cannot revise an original deadline simply because a case is about to miss it.

A scorecard version therefore has two dates worth preserving: when the definition applies and when the evidence was last checked. A newer file does not automatically make its definition applicable to earlier work. Ask the analyst to run the proposed rule on a success, a late acceptance, a customer pause and an unknown acceptance before the number is used outside the pilot.

Complete a ten-field metric card

Here is an original draft for Theo and Arin to test with the authorised operating owners.

Add an agreed target only after checking the baseline, capacity and risks. A stretch target may be inappropriate if it encourages people to conceal a delay or pressure a customer to accept incomplete work.

Build a balanced scorecard

The team chooses a small set of complementary measures rather than one headline number.

These measures should not override product safety, contractual rights or other specialist criteria. Keep the actual approved release and acceptance requirements in place.

Name the individual contributions too. Sales supplies the agreed scope; engineering estimates the relevant work; customer success checks customer preparation; the authorised owner approves dates. A shared result can coexist with clear responsibility for each step.

Keep quality evidence alongside acceptance timing. A customer’s on-time acceptance and a later material defect answer different questions. The team records one reopened workflow among the trial’s seventeen on-time acceptances, without treating that as a reason to erase its original acceptance event. The quality review still has to establish the defect, its materiality and the response through the authorised product process. A faster hand-off is not permission to relax a release criterion.

Check the arithmetic and its meaning

Suppose forty of fifty eligible implementations met the definition in the baseline period. The rate is 80%. An illustrative next-period target of forty-six out of fifty would be 92%.

The difference is twelve percentage points. Relative to the 80% baseline, the increase would be 15%, because twelve divided by eighty is 0.15. Six additional successful cases out of the same fifty explain the change. Label this as a target example, not an achieved coaching result.

If the next period contains only twenty implementations, compare counts and case mix as well as percentages. A changed denominator can make a rate look more impressive while representing different work.

Likewise, a nominal monetary value assigned to each completed implementation is not automatically realised profit. Financial evaluation needs actual costs, benefits, timing and defensible attribution assumptions agreed with the relevant finance owner.

For the detailed trial, the earlier fifty cases include forty standard implementations with thirty-two on-time acceptances, and ten custom implementations with eight. Both subgroups are at 80%. The next twenty include eighteen standard cases with sixteen on time and two custom cases with one on time: seventeen of twenty, or 85% overall. That is five percentage points above the earlier overall rate, while custom work is one of two, or 50%, and the custom share has fallen from ten of fifty to two of twenty.

The smaller cohort has seventeen successful cases, not the earlier forty or the illustrative target of forty-six. Standard cases are sixteen of eighteen, about 88.9%; the two custom cases are too few to support a confident general performance judgment. These comparisons expose a question about case mix and capacity. They do not demonstrate that the workshop caused the overall difference or that custom work has a stable new failure rate.

HM Treasury’s policy-evaluation guidance distinguishes monitoring a before-and-after change from establishing attribution. With no suitable comparison, other influences can explain a change. That methodological caution is useful here; it does not make this company a government policy evaluation or require a coach to run an experiment on employee pay. Choose an appropriate evaluation design with the people qualified to do that work before making a causal or financial claim.

Facilitate a seventy-five-minute design workshop

An original agenda allocates ten minutes to purpose and authority, fifteen to the current evidence, twenty to functional concerns and dependencies, fifteen to the draft definition, ten to quality and unintended-incentive checks, and five to owners and next steps. Total: seventy-five minutes.

Send a short preparation request: “Bring one example where the current hand-off worked and one where it did not. Identify what information was available and which decision was needed. Use approved, proportionate records.”

At the close, read the definition aloud and test it against two cases. A customer who is unavailable and a product that fails acceptance should not be treated as interchangeable without examining the agreed rules. Record uncertainty for the data owner to resolve before publishing a score.

A full-day offsite may be appropriate for a wider problem, but the duration should follow the work. The deliverable is a usable definition and decision process, not a promise that a day together creates permanent alignment.

Six people, including the facilitator, attend the seventy-five-minute workshop. That is 450 participant-minutes, not six workshops. Three later fifteen-minute exception reviews each have four attendees, including the facilitator: another 180 participant-minutes. The selected meeting total is 630 participant-minutes, while the facilitator’s live contribution is 120 minutes nested within it. Preparation, analysis, report correction and the wider implementation effort are separate work, not zero-cost because they occur outside a meeting.

Review governance and compensation separately

The board or authorised management may need to approve targets, reporting or compensation changes. The coach can facilitate clarification within competence, while HR, legal, finance and relevant governance owners assess the actual arrangements.

A shared operational measure does not require a fixed percentage of every executive's bonus. Before linking pay to a measure, examine control, data quality, unintended incentives, time horizon and applicable obligations. A metric suitable for learning may be unsuitable for an individual payout.

Keep developmental coaching material separate from performance-management decisions unless the relevant arrangements have been explicitly agreed. Company funding does not provide unrestricted access to private session content.

Theo asks whether the 85% headline can immediately trigger an individual bonus adjustment. Neri says no: a pilot learning measure has not become an authorised pay rule. The relevant management, HR and finance owners still need to examine that separate proposal. Arin’s request to retain the custom-case caveat is recorded as a substantive concern, not converted into agreement by everyone else’s attendance.

Respond to the second decision

After a trial, the headline delivery rate improves. The analyst notices that the new cohort contains fewer custom projects. Theo and Arin therefore compare standard and custom implementations separately before claiming that the new hand-off caused the improvement.

They also find that early review has created a queue for estimates. The next decision concerns estimate capacity and which commitments genuinely need review. They keep the useful definition, inspect the bottleneck and propose a narrower approval rule for the authorised owner to consider.

This is what a review process is for: new evidence can change the next action. Repeating a shared-success slogan would not resolve the queue.

The trial’s first report reveals another error. Ivo’s copy, promised by 16:00 UTC on 3 March, arrives at 10:00 on 4 March, eighteen elapsed hours late. It shows seventeen successes out of eighteen after excluding the paused case and the case with no acceptance evidence. That would be about 94.4%, but it does not use the approved twenty-case denominator. Arin identifies the exclusion and asks for the original rule to be restored.

At 14:00 Ivo acknowledges the missed promise and the denominator error. The corrected organisational report arrives at 09:00 on 5 March, and Neri confirms readable receipt at 11:00. It restores seventeen of twenty and separately names the three other states: one late acceptance, one customer pause and one unknown acceptance. Eighteen workflows have eventually been accepted if the late one is included; that does not make eighteen on-time successes. The late first report remains part of the history.

The queue also remains real. Early review catches unestimated commitments but leaves a bottleneck; the facilitator cannot clear it by asking for greater trust. Neri owns a proposal to reserve estimation capacity and distinguish routine commitments from genuinely custom work. Theo does not receive authority to promise the dates still awaiting review. Arin declines another coaching session about willingness to cooperate, while agreeing to supply the bounded capacity information required for the operational decision.

They close this limited facilitation after the third exception review on 8 March. Theo and Arin receive the agreed definition and their own stated actions; Neri receives the organisational report with its caveats. Neri owns a proposed 10 March capacity decision, and Ivo owns an 11 March check of the unknown acceptance record. Both future outcomes remain unknown. No revised bonus, extra coaching appointment, customer acceptance or solved queue is implied by the close.

Keep the records in appropriate systems

The operational scorecard belongs in the organisation's approved reporting system with suitable source permissions. A coaching record can contain the agreed definition, the client's action and the question to review, without copying customer contracts or confidential technical details.

In Coachful, use an appropriate worksheet for preparation and explain the audience for submitted responses. Review an agreed response with the client and record actions through the normal client-plan workflow. A custom flow can gather a short reflection when its audience, display and reminders are configured appropriately.

Before sharing recaps or sponsor summaries, check the effective approval, session-type and delivery settings. Approval is not universally required by default; a session-type override can change the workspace rule. Personal journal takeaways are not automatic coaching feedback: inspect applicable page-sharing controls and the client’s deliberate sharing choice. The crisis trust-dashboard guide considers a different context where decisions and information boundaries need particular care.

For a custom reflection flow, check its actual entry routes as well as the selected-client list. Journey or program entry can have separate availability, and channel settings do not prove delivery. An editable text response can also feed a weekly reflection record under its configured behavior. Confirm the intended destination and audience before requesting an answer; do not copy customer contract evidence into a coaching reflection merely to keep the scorecard in one place.

Frequently asked questions

Do shared metrics replace departmental measures?

Usually they add a view of a dependency. Keep specialist measures needed for quality, safety, finance or other responsibilities.

Must success be a number?

Some criteria need a defined descriptive judgment. Identify the authorised reviewer, evidence and criteria rather than assigning an arbitrary score.

Should we change bonuses immediately?

Review the measure and its consequences first, with the relevant qualified owners. Developmental inquiry and pay decisions have different purposes.

What if the teams still disagree?

Clarify whether the issue is evidence, resources, priorities or authority. Record the unresolved choice and take it to the person who can decide.

Can a portal prove that coaching caused improvement?

Activity and performance records can inform evaluation. They do not establish causal attribution or financial return on their own.

Sources and further reading

Judith E. Glaser, primary Expert Interview with Rachel Salaman, Mindtools. Shared success, inquiry and the useful roles of different conversation types. The four-step protocol and examples here are original adaptations.

US Office of Personnel Management, “Developing Performance Standards.” Measurement-design context; federal appraisal requirements are not presented as corporate coaching rules.

Explore Coachful's features to organise preparation, agreed coaching actions and appropriate follow-through.