Skip to main content
Steady at the Top
← All articles

How to tell whether a company turnaround is working

Measure turnaround progress by connecting customer and operating outcomes to the decisions and routines expected to produce them. Use leading indicators as hypotheses about improvement, then check whether the promised consequence follows. Report unresolved failures alongside gains so the dashboard can change the recovery plan.

Job feedback is direct information from carrying out work about the effectiveness of that work. J. Richard Hackman and Greg R. Oldham distinguish this feature in their 1980 work-design treatment, Work Redesign, chapter 4: Motivation Through the Design of Work. A dashboard can summarize feedback, but a green status label is not necessarily feedback from the job. Praise also differs from information showing whether a customer commitment was fulfilled.

A private moment that makes the issue visible

Composite scenario, not a client case: You refresh the recovery dashboard from bed before the house wakes. Every workstream is green. A customer email arriving overnight describes the same broken handoff as last month. The color is recording that managers completed their updates, while the customer is describing a result nobody has yet measured.

Use the research within its limits

A recovery measure should help the person who must choose the next action. Vroom and Jago (2007): The Role of the Situation in Leadership discusses situational requirements in leadership and participation. It supports attention to the decision context, without validating any particular turnaround dashboard. The evidence-chain review below proposes a way to examine whether a measure informs the recovery work.

State the decision before discussing personalities

The dashboard needs to answer whether recovery work is changing delivery, not whether leaders are busy. Choose a small chain from a leadership behavior to an operating consequence and a customer outcome.

Use the recovery evidence-chain review

  1. Name the recovery outcome. Choose a result the recovery must achieve, such as dependable delivery of agreed commitments. Define the population, time window, and record used to measure it. Specify whether the measure counts customers, orders, or commitments. Changing the unit midway can create improvement that reflects a different denominator.

  2. Select the behavior to test. Identify the routine expected to improve that outcome. For delivery, this might be checking available capacity before confirming a date to a customer. Write why this behavior should affect the outcome. The explanation should identify the handoff it changes, rather than rely on a general label such as discipline.

  3. Connect the intermediate evidence. Track whether the routine occurs and whether it changes the next handoff. Record exceptions rather than treating a completed meeting as proof of effective coordination. Distinguish a review completed before commitment from one completed afterward. Identical activity counts can describe very different controls.

  4. Inspect a contrary example. Trace a failed commitment despite a green leading indicator. Ask whether the proposed causal link is wrong or whether the indicator measures the routine poorly. Include a failure from an ordinary case, not only an exceptional crisis. Test whether the indicator misses the conditions most customers actually encounter.

  5. Revise the recovery decision. Keep indicators that inform a change in action. Retire status measures that consume reporting time without changing an owner, resource allocation, or operating choice. Give every retained measure a named decision use. If nobody can explain what a different reading would change, reconsider collecting it.

Build one chain before expanding the dashboard

Begin with a recovery outcome customers or operating owners can recognize. Dependable delivery, for example, needs an agreed definition of the commitment and its completion. Decide whether a revised date counts as success and how customer-approved changes are recorded. Without that definition, two managers can report different results while both believe they are following the measure.

Next identify the process change expected to improve that outcome. A capacity check before accepting an order is a testable proposal. A general promise to improve accountability is harder to connect to evidence. Ask what the check must contain, who can reject an impossible date, and where that rejection appears in the order record. The measure should describe the proposed control closely enough to expose a failure.

Finally connect the process to a later result. Did checked orders encounter fewer avoidable conflicts? Did the next team receive the information it needed? These are questions for the organization’s records, not claims that one routine always causes improvement. Staffing, product mix, demand, and supplier conditions can change at the same time. Keep those changes visible when interpreting the result.

Separate completion, quality, and consequence

A completed review establishes that an activity occurred. Its quality concerns whether the relevant evidence was available and used. Its consequence concerns what changed afterward. A reporting system that collapses all three into one checkmark can show perfect compliance while leaving the operating problem untouched.

In a hypothetical example, managers record a capacity review for every order. Inspection shows that the reviews contain no information about a constrained installation team. The meeting occurred, yet it could not resolve the relevant conflict. The corrective choice is to improve the review’s inputs and authority, rather than demand a higher completion percentage that is already at its ceiling.

Decide which layer each indicator represents. Attendance might help locate a missing participant, but it should not be presented as proof of customer recovery. An exception resolved before confirmation can provide stronger evidence of a functioning control. The outcome measure remains necessary because a sound-looking process may still fail to produce the intended result.

Inspect green cases as well as red ones

Select some cases that the dashboard calls successful and trace them through the underlying records. Ask what makes them count as successful. If a promised date was changed several times before delivery, determine whether the measure recognizes the original failure or quietly treats the last date as the only commitment. Define the rule before debating a particular manager’s score.

Also review a failure that passed every leading check. Such a case can reveal that the proposed chain is incomplete. Perhaps the capacity check covered production but omitted installation. Perhaps the relevant information arrived after approval. Repair the link that failed instead of adding another status meeting unrelated to the missing information.

Invite the people who perform the handoff to explain misleading readings. They may know that a green field reflects a software default or an administrative deadline. Verify the account against the actual workflow. Avoid treating dashboard skepticism as resistance to recovery; the useful objection identifies where the displayed measure and the work diverge.

Retire measures when their decision use ends

During containment, a daily count of unresolved commitments may determine immediate attention. After ownership stabilizes, the same count may need a different review frequency. Ask what decision still depends on the daily report. Retaining every crisis measure indefinitely can consume the time needed to perform the recovered process.

Keep a short record when you change a definition or retire an indicator. Explain the previous use, the reason for the change, and what will detect deterioration afterward. Otherwise, a lighter dashboard can look like an attempt to hide poor results. Preserve the ability to compare relevant periods without pretending that differently defined numbers are interchangeable.

At a recovery review, describe the evidence chain in ordinary language: the behavior changed, the next handoff improved or did not, and the final outcome remains under observation. Name the remaining gap. This gives the board and operating team a decision they can discuss, rather than a color they must trust.

If the chain fails, revise the proposed mechanism before demanding more data. More rows cannot rescue a measure that has no plausible connection to the result. A smaller set of clearly defined indicators, supported by case inspection and an adverse-result route, can make the recovery discussion more useful without claiming certainty about causation.

Put the boundary into words

This indicator is an early signal, not proof that the recovery is complete. Show me a customer result that supports it and a case that challenges it. Then identify the operating decision the comparison changes.

Work through a bounded example

Hypothetical application: The team reports that every large order received a capacity review. A sample shows reviews happened after sales confirmed the date. The indicator counted the activity but missed its timing. The owner changes the control point and tests subsequent commitments before declaring the handoff repaired.

Check the failure mode before scaling

Do not move a target or change the denominator without showing the change. Avoid a score that averages away the most consequential failures. Have finance and operating owners verify the source records before the board relies on the recovery picture.

When an indicator reveals an immediate safety, liquidity, or contractual problem, route it to the responsible specialist before the next dashboard meeting. Keep the operational response separate from any claim that the recovery system now works. One contained incident cannot establish that the underlying cause has been removed.

Review the work, then revise the arrangement

At each checkpoint, ask which early signals predicted the expected consequence and which did not. Preserve the prior measure long enough to explain any change in the reported trend.

The team must be able to report an adverse result alongside an improving indicator. Edmondson (1999): Psychological Safety and Learning Behavior in Work Teams examines interpersonal risk and learning in work teams. A dashboard review that punishes the messenger can restrict the evidence available; the paper does not establish that adding an exception column alone changes that climate.

Connect this decision to the next one

Use these measures within Returning as CEO: how to lead a company turnaround. Findings about ownership connect to How to choose leaders who can carry a company turnaround. Explain the recovery commitments through What to say to employees when you return as their CEO. For survey evidence, see What to do after an employee engagement survey disappoints.

Sources

Steady at the Top