Control and Management Dashboard

From the risk register to a management dashboard

Take out your Week 15 risk register and your Week 13 cash calendar. The register lists the main risks in the 12-week test of Option A · Expansion (expand cheerleading and sports-entertainment talent management). Each row has an owner, a trigger and a stop condition. The cash calendar shows when money leaves the company and when the home team's NT$60,000 per appearance arrives, 30 days after each game. It also shows how close cash comes to the NT$100,000 floor.

Now go back to the Monday at the start of test week 8, six days before game 2. Judy opens the chat group, and the news looks good. The video from game 1 had many views. Last week, in test week 7, a third team contacted Judy and asked for an appearance at its own home game on the Saturday of test week 8. It feels like momentum.

Other facts are also true that Monday. Cash is NT$155,000, and NT$25,000 for game 2 must go out this week. The game 1 fee is due in a week, and nobody at the home team has confirmed it. Accepting the third team's request would have taken Mina to 11 of her 10 test hours, so in test week 7 Judy sent the third team a written offer for a date after game 2 (Week 12). Its reply is due this week. Mina's backup has never run a game week.

All of this information exists, but it sits in different places: the register, the cash calendar and the chat. Nobody sees it together, so nobody asks the one question that matters on Monday: what should we do this week? A week later the fee did not arrive, and two people waited a week for their pay.

A management dashboard is a short, regularly reviewed page that brings together the few signals a manager needs for the next decisions. Each line on it is useful only when six things are connected. Take one line from the practice scenario, "days since the latest game with no payment confirmation", and look at the six links.

• Promise. The home team pays NT$60,000 no later than 30 days after the game.
• Decision. Can Judy sign an agreement with another team?
• Source. The bank record kept by the bookkeeper, and the written confirmation from the home team's invoice approver.
• Owner. Judy, who holds the payment conversations.
• Action. On day 25 with no confirmation, Judy calls the approver.
• Review. At the end of the test, Judy's staff ask whether this line helped, and keep, change or drop it.

If one link is missing, the line is only a number on a screen. This week's question is: which few signals should Judy's company watch during the test, and what should happen when one of them moves?

Where this week fits

Week 15 produced a risk register in which every risk has an owner, a trigger and a stop condition. This week turns each row into a line on a management dashboard and adds the cash line from the Week 13 cash calendar. In Week 17, the dashboard's leading indicators become the stop signals in your final decision memo.

Report, measure and control

Three words are often used as if they meant the same thing. A report describes what happened. A measure records a defined condition or result, such as "hours Mina worked on the test last week". A control connects a measure to a decision, a responsible person and a response.

Suppose the chat shows that a photo approval is overdue, but nobody is responsible for following it up. The company then has information, but no control.

A practical control loop asks six questions in order:

1. What promise or risk is at stake?
2. What decision might change?
3. What evidence can we collect for a legitimate purpose?
4. Who can act?
5. What action follows when the signal moves?
6. When will we check whether the measure and the action still make sense?

The loop matters more than the look of the page. A beautiful display cannot replace a missing owner.

A measure can also be used in two ways. Henri (2006) studied how companies use their performance measures and separated diagnostic use from interactive use. Diagnostic use checks whether work stays inside an agreed range, such as "is forecast cash at least NT$130,000?" Interactive use brings the measure into a discussion about uncertainty and strategy, such as "the game 1 fee was late; should we ask for a deposit before the next game?" The same cash line can do both jobs. It triggers a phone call in one meeting and a strategic question in the next.

Control also includes boundaries: points at which work pauses until a condition is met. The stop conditions from your risk register are boundaries of this kind. A boundary works when people understand its purpose, have the time to follow it and can report an exception. If a rule can only be obeyed by hiding work, people will hide the work.

Start from a decision, then choose the measures

You may meet the Balanced Scorecard, a well-known format that places measures from four areas side by side: financial results, customers, internal processes, and learning and growth. Its main idea is that a company should not judge its strategy by money alone. De Geuser, Mooraj and Oyon (2009) studied organizations that used the scorecard. Its link to better performance seemed to work through connecting strategy to daily work and involving people, rather than through the number of measures.

For a small company, the lesson is to start from the decision and the promise, not from a template with fixed boxes. During the Option A test, the next decisions are practical: is game 2 safe to hold as planned, can the company afford it, and can Judy accept another team? The dashboard needs only the lines that inform those decisions.

Each line should pass three tests. Its definition is clear enough that two people would count the same event. Its source is named, such as the bank record or Mina's time log. Its owner can act, or can pass the problem up to Judy.

The owner is the person responsible for checking the source and responding, not the person blamed when the number moves. If Mina owns the photo line but the home team sends its posting plan late, the line records a gap in Mina's authority, not in her effort.

Keep the main page short. Four strong lines are more useful than ten weak ones, and details can sit in the register behind the page. Some lines carry more weight than their space on the page suggests. The photo line takes one row, like the others, but it can stop the work on its own. If a photo is published without written approval, all promotional posts pause, whatever the views, bookings or cash show.

Leading and lagging indicators

A lagging indicator measures a result after it has happened. In the test, examples are "fees received within 30 days" and "photos published without written approval". Lagging indicators show whether a promise was kept, but they arrive too late to prevent the failure they record.

A leading indicator measures a condition that may affect a future result. Examples are "lowest cash forecast in the coming weeks", "posting plan received 48 hours before the game" and "Mina's test hours". Each is a warning, not a guarantee. A posting plan that arrives on time makes a photo mistake less likely, but it cannot prove that no mistake will happen.

Both kinds are easy to over-read. A low number of complaints may mean that people are satisfied, or that they have stopped reporting. A fast reply time may mean clear work, or rushed work. Write the likely meaning and its limit next to each line.

In this course, every value on a dashboard carries one of three labels. Known means it comes from the case or from a record you were given. Invented means you made it up for practice. Unknown means nobody has the information yet, and you write the reason next to it. A blank with a reason is more useful than a precise number nobody can defend.

Turning risk-register rows into dashboard lines

The link between Week 15 and this week is direct. Each column of the risk register becomes a column of the dashboard.

The cash calendar from Week 13 adds one more line. It forecasts the lowest cash point in the coming weeks and compares it with the NT$100,000 floor. Judy's staff also add a learning line from the Week 5 goal tree, because the test is meant to find out whether other teams will pay.

Worked example: the dashboard at the start of test week 8

To test the dashboard, Judy's staff fill it in for a Monday that has already passed. It is the start of test week 8, six days before game 2 and one week before the game 1 fee was due. If the dashboard works, it should have warned them before anyone was asked to wait. The cash figures come from the Week 13 cash calendar.

The cash line follows one plain rule. If cash is forecast to fall below NT$130,000, Judy decides before any new booking. The threshold sits NT$30,000 above the floor, so she has time to act.

One line has crossed its action threshold this Monday: cash. The payment line is two days from its threshold, and Mina's hours show how close last week came to the limit. The lagging fee line is still empty, because it can record a late fee only after it has happened. Here is how Judy's staff read the cash line.

• Cash at the start of test week 8: NT$155,000. That is the NT$180,000 the test started with, minus NT$25,000 paid before game 1.
• Paid before the game 1 fee is due on day 57: NT$25,000 for game 2 this week (the performer's first half NT$6,000, travel NT$11,000 and logistics NT$8,000).
• If the fee arrives on time: NT$155,000 − NT$25,000 = NT$130,000. Payday is the next day, day 58, and carries NT$20,000, so cash becomes NT$130,000 + NT$60,000 − NT$20,000 = NT$170,000.
• If the fee is a week late and everyone is paid on payday: NT$130,000 − NT$20,000 = NT$110,000. That is below the NT$130,000 threshold, though still NT$10,000 above the floor.

So Judy decides on Monday of test week 8, before any new booking. She asks the home team's approver to confirm the payment date in writing and postpones spending that is not essential. She also decides in advance that if the fee is late, everyone is still paid on payday, because NT$110,000 stays above the floor. In the Week 13 and 14 story, nobody made that decision in advance: when the fee was late, Judy held NT$10,000 of pay and cash stayed at NT$120,000. The dashboard would have turned a last-minute choice into an open decision made a week earlier.

Reviewing the dashboard as a conversation

A dashboard meeting works best when it starts from the question each line was built to answer, not from a demand to explain every change. A review sequence in five moves keeps the meeting useful and fair. Here it is applied to Mina's hours line:

1. Name the change without blame. "Last week, accepting the third team's request would have taken Mina to 11 of her 10 test hours."
2. Check the source and the definition. Mina's time log shows about 3 test hours in a normal week and about 7–8 in the week before a game. The Week 11 estimate says a new team's request adds about 4 hours of checks and messages.
3. Ask what the number cannot show. The line leaves out the work the performer liaison did on the move to Sunday. It also cannot show how tired Mina was after a week of schedule changes.
4. Choose an owner, an action and who to pass it up to. Judy has already sent the third team a written offer for a date after game 2 (Week 12). From now on, a request that arrives in the week before a game waits until after the game, and Mina passes any exception up to Judy.
5. Record what was learned, and set a date to review the measure itself. At the test week 12 review, Judy's staff will ask whether "hours" is the right measure, or whether "game-day tasks the backup has practiced" would warn them better.

This sequence supports the psychological safety you studied in Week 8. Mina can say that she is near her limit without being called slow, and Judy still writes down what will change and checks it the next Monday.

The meeting should also look at silence. A clean dashboard may reflect good work, or a reporting route that nobody uses. Ask who could not report, who had a reason to stay quiet and whether an earlier concern led to any change. A dashboard that never shows an exception may be one that nobody trusts.

Two design choices keep the dashboard in proportion. First, the review frequency should fit the risk. Cash and Mina's hours are checked weekly, while the learning line is checked at the test week 6 and test week 12 reviews. Second, every line needs a retirement rule. If the company stops booking appearances, the permission and rest lines can be retired, because they no longer inform a decision.

When targets reward the wrong behavior

Kerr (1975) described a common design error: an organization rewards one behavior while hoping for a different result. Bevan and Hood (2006) studied targets in the English public health care system. They showed how targets can lead to gaming, in which the reported numbers improve more than the real service. Both sources point to the same risk. A target can be measured accurately and still be a bad control.

Here is how that can happen in the test. Judy praises Mina in a staff meeting for replying to every message from the home team within one hour. Over the next weeks, Mina starts sending quick replies such as "Received, I will check!"

The reply time looks excellent. Meanwhile, the home team's question about who owns the photo rights stays unanswered for four days. The measure captured speed. It hid whether questions were answered.

Two terms describe what went wrong. Metric distortion is any change in behavior caused by aiming at the number instead of the purpose behind it. Mina did not plan to cheat; the system made speed visible and answers invisible.

Gaming is narrower: changing records or behavior on purpose to improve the number without improving the real result. The difference matters. A manager should first look at the design of the measure before accusing a person, because an accusation makes future reporting less honest.

The remedy is a countermeasure and a conversation. Judy's staff pair the reply-time line with a second line, "home-team questions fully answered within three days", and check a sample of conversations each Monday. Every target also needs an exit rule: an owner, a review date and a statement of what the target must never be used to decide. The reply-time line, for example, is never used to judge Mina's commitment.

Exploration and exploitation need different lines

March (1991) described two kinds of organizational learning. Exploitation means improving and using what the organization already knows, such as running a game-day routine more smoothly. Exploration means searching for new possibilities, such as finding out whether other teams would pay for the service. March showed that the returns from exploitation come sooner and are more certain, so organizations tend to let exploitation crowd out exploration. The organization then becomes efficient at today's work and less able to adapt.

The Option A test contains both kinds of work. Delivering the two games is exploitation: the company has made promises, and the dashboard lines protect them with thresholds. Judy's talks with two more teams are exploration: the company is testing whether a future promise is worth making. The two kinds of work need different evidence.

Mixing the two kinds of work into one score causes confusion. A useful experiment that ends early can look like poor delivery. Routine delivery can look like innovation.

In Week 12, the second team agreed to pay NT$60,000, but only 45 days after the game. That is a useful result for the learning line, because it suggests that cash timing, not demand, may limit growth. The label "learning" should not protect a repeated mistake either. Each learning line needs a review date and a decision.

Written rules that help people do the work

In Week 15, Judy's staff moved photo approvals, schedule changes and payment dates from the chat group into a written log. Written rules can help or harm. Adler and Borys (1996) separated two kinds of formal rules. Enabling formalization helps people understand the work, handle exceptions and improve the process. Coercive formalization pushes people to comply, limits their judgment and treats their local knowledge as a problem.

The same photo checklist can work either way. The enabling version states the purpose (the performer controls her image), the rule (written approval before any posting) and who approves exceptions. It also gives a route for reporting a problem, such as a home team that posts too early. The coercive version is a list of boxes to tick. Mina is blamed when the home team sends its plan late, even though she does not control its timing.

Every formal control needs an exception route, because an exception is often information rather than failure.

Judy's staff agree on three routes. Anything touching consent, safety or legal permission goes to Judy immediately. A small delay in a routine task stays with the role owner. A pattern, such as home teams repeatedly sending plans late, goes to the next review. Nobody should have to pretend to be certain in order to get attention.

Measuring the conditions people work in

People lines on a dashboard should measure the conditions of the work, such as workload, rest and pay timing. They should not rank people's commitment. Mina's hours line is a capacity signal. It tells Judy when the test is running beyond what one coordinator can carry, as you found in Week 11. It would become harmful if it were used to compare Mina with someone else or to question her loyalty.

Collect only the people information that a decision needs. Mina's family-care reason stays in a private note, and the dashboard shows only that she is unavailable after 6 p.m. Information shared for safety or planning should not later be used for pay or evaluation. A person who shares a scheduling limit for safety should never discover that the same note was used to judge commitment.

A dashboard should also make invisible work visible where it can. Translation, checking and follow-up rarely appear in public results. In the practice scenario, the performer liaison's work on Korean approvals appears only because Judy's staff added it to the log. Week 14 showed that silence can mean fear, a belief that speaking will change nothing, or a wish to protect a relationship, so a quiet people line needs a question as well as a number.

Comparing across borders by mechanism

Week 4 showed how to compare settings by mechanisms, such as who approves and which language counts, rather than by national character. For a dashboard, the question is which mechanism changes the meaning of a number. The performer's reply time to a schedule change includes the time the performer liaison needs to send her a Korean version. A line that counts her replies from the moment Mina sends her message would measure the language route, not her willingness. The fix is to start the clock when the liaison sends the Korean version.

Where AI adds risk to a dashboard

An AI tool could draft the Monday summary or flag the lines that moved. Kellogg, Valentine and Christin (2020) describe how software can record some work and miss the rest, turn complex work into one rating and recommend what to look at first. Studies of chess players (Krakowski et al., 2023) and radiologists (Lebovitz et al., 2022) suggest that AI output is most useful when people can question it and combine it with their own judgment. So each line's owner reads the source before acting, and no AI output decides a stop line.

Dashboards for the three options

The worked example is for the Option A test. Each option would need different lines, and each has a tempting number that can mislead.

For every option, ask the same questions: what decision each line supports, who owns it and what behavior it might encourage. This week does not choose between the options; in Week 17 you choose one and defend it.

Class practice: a three-line dashboard

This practice is done without AI tools. Your teacher will tell you which worksheet to use and how to submit it. Bring your Week 15 risk register and your Week 13 cash calendar.

1. Choose three lines. Take one line for cash, one for people and working conditions, and one from another row of your register.
2. Complete each line. For each line, write its owner, its action threshold and the action that follows. Label each value known, invented or unknown.
3. Check each other's work. Swap with a classmate, who names one thing your three lines cannot see and one behavior a line might encourage.

Keep your dashboard for Week 17.

Key points and what you bring to Week 17

• A dashboard line is useful when it links a promise, a decision, a source, an owner, an action and a review of the line itself.
• Each risk-register row becomes a dashboard line: the owner stays, the trigger becomes the action threshold and the stop condition becomes the stop line. The cash calendar adds the cash line.
• Leading indicators warn before a result; lagging indicators record whether a promise was kept. Write the limit of each line next to it.
• Targets can reward the wrong behavior. Pair a target with a countermeasure, and look at the design before blaming a person.
• Delivering promises (exploitation) and testing new ones (exploration) need different lines. A learning line is judged by what was learned.

For Week 17. Bring your management dashboard. In Week 17 you write a final decision memo that compares the three options as whole systems. The leading indicators on your dashboard become its stop signals: the conditions under which the chosen option would pause or change.