Chapter 5 — What the Dashboard Cannot See

What the Dashboard Cannot See

The board meeting began with a green number.

It appeared in the upper-right corner of the first slide, large enough to be visible from the far end of the table. Beneath it was a smaller arrow pointing upward.

The number represented people served during the quarter.

The executive director had reviewed it several times before the meeting. She knew how it had been calculated, which programs were included, which people had been counted more than once, and which forms staff had completed late enough that their work would not appear until the next report.

She also knew the number was good.

The organization had reached more people than it had the previous quarter. Costs per participant had declined. Attendance had increased at two sites. A new partnership with a school district had produced a steady stream of referrals.

The board members smiled. Someone said, “This is exactly the kind of momentum we need.” The executive director advanced to the next slide. It showed program completion. Another green number. Then volunteer hours. Green again.

The dashboard was doing what it had been designed to do. It translated a complicated organization into a set of signals that people who did not live inside the work could understand.

The board met six times a year. Most members had full-time jobs elsewhere. Some had visited program sites, though not recently. One knew fundraising. Another understood municipal grants. One had spent decades in finance and could see risk in a budget before anyone else noticed it. They cared about the organization. They also depended on summaries.

Without summaries, the meeting would have required days.

The board could not read every case note, accompany every staff member, listen to every family, or observe each program as it unfolded. The dashboard made governance possible by reducing the work to a scale the board could hold.

That reduction was not dishonest. It was incomplete. The executive director knew that the quarter had also been difficult.

A field coordinator had taken leave after months of absorbing conflict between staff and a partner organization. Two experienced volunteers had stopped coming after a program expansion changed the character of their work. One site had reached more people partly because sessions had become shorter. A family who appeared in the count as successfully referred had spent six weeks moving between agencies before receiving help.

None of this appeared in red. The dashboard had no color for we reached more people because the work became thinner. It had no field for the person holding this relationship is close to leaving. It had no arrow for the number improved while trust declined.

The board was not hiding these realities. The executive director had mentioned staffing pressure in earlier updates. The site managers had submitted narrative reports. A board member had attended one of the difficult meetings with the partner organization.

But the green numbers gathered attention. That is what numbers do when a room is uncertain. They offer something solid.

The green number gives the board permission to move on. The organization reached the people. The arrow points up. The board can return to the rest of the agenda with a clearer conscience, and a solid number can end the asking before anyone asks what the number had to leave out in order to travel. We often say that organizations measure what matters. More often, they measure what can be made consistent enough to compare. People served can be counted. Dollars spent can be counted. Hours worked, applications processed, cases closed, and projects completed can be placed in rows.

The numbers are not trivial. They can reveal patterns that stories conceal. A program may feel successful because one family’s story is moving while quietly failing most of the people who enter it. A team may believe it is improving because meetings feel better while delivery continues to slow.

Measurement can interrupt flattering narratives. It can show that memory is selective. It can reveal that an organization’s favorite story is not the most common one. The problem begins when the measure stops being a window and becomes the room.¹ A dashboard does not merely report what happened. It teaches people what kind of happening deserves attention.

When the board looks first at people served, the organization learns that reach must be narrated. When leadership asks weekly for percentage complete, the team learns to translate uncertainty into progress. When a university rewards publications, faculty learn that mentoring counts most clearly when it produces someone else’s publication.

This does not require anyone to manipulate the system. People adapt to visibility because visibility affects survival. What appears on the dashboard can be defended. What remains outside it must be explained. Explanation is slower than a number. It is also more vulnerable.

A number arrives without tone. It does not sound defensive. It does not appear tired. It does not pause before answering. It can be compared to last quarter and placed beside a target.

A person saying, “The work is more fragile than the report suggests,” introduces judgment. The room must decide whether to trust them. This is one reason organizations often reach for measurement when trust thins. A board that does not regularly see the work asks for clearer metrics. A manager who cannot tell whether a project is progressing asks for more frequent updates. A funder who worries about impact asks for standardized outcomes. A leader who has been surprised by failure asks for earlier indicators. Each request is reasonable. Together, they can create an organization that spends increasing amounts of time proving that work is happening while becoming less able to hear when the work is becoming thinner.

The proof may become more visible as the work becomes harder to see. The nonprofit’s dashboard began after a difficult year.

A major donor had asked why two programs with similar budgets reported different outcomes. The board had struggled to answer. One program counted attendance. Another counted completed plans. A third relied on stories from staff. The organization could not describe itself coherently.

The dashboard solved a real problem.

It created common definitions. It forced programs to record participation consistently. It revealed that one site had been counting every contact as service while another counted only completed engagements. It gave the board a way to ask better questions.

For a while, the numbers improved understanding. Then the numbers began shaping the work they were meant to describe.

A program director noticed that short workshops reached more people than longer sessions. The shorter workshops also produced better quarterly numbers. Staff continued offering deeper support where they could, but the calendar gradually filled with activities that counted cleanly.

A coordinator learned that following one family through a complicated process might consume several days and still produce no completed outcome. Hosting a group session could add twenty-five people to the quarterly total.

No one instructed her to choose the group session. The dashboard did not need to issue commands. It only needed to decide which contribution would be visible later. Measurement changes behavior even when nobody is trying to game it.² The coordinator can still choose the slower case. She cannot choose it without feeling that the visible story of the quarter will make less room for her judgment. The measure enters judgment.

Sometimes that is the point. Organizations use measures to coordinate behavior because direct supervision cannot scale. A target tells thousands of people what the institution is trying to increase or reduce.

But targets do not carry all the reasons behind them. They travel farther than context.

This is not proof that measurement is corrupt. It is evidence that every measure is a compression.³ It takes a wide field of activity and chooses one dimension that can travel. The more complicated the work, the more remains outside the chosen dimension. A dashboard is a map made for a purpose. The danger is not that the map leaves things out.

Every map does. The danger is forgetting why those things were left out and who must continue living in the territory. People closest to the work usually know the difference. They know when an improved number represents real improvement. They also know when it represents a transfer. The program reached more people because access improved.

Or it reached more people because each person received less. The project is on schedule because a difficult dependency was resolved. Or the project is on schedule because the work was moved into a later phase. Volunteer hours increased because more people were invited into meaningful work. Or they increased because volunteers quietly replaced staff capacity.

The team closed more incidents because systems became more reliable. Or it closed them because unresolved problems were divided into smaller tickets. The dashboard may not distinguish these cases. Someone must. This is why measurement depends on trust even when measurement is introduced because trust is weak.⁴ The number requires a person who can say what the number means here.

A board member sees that volunteer hours increased. A site coordinator knows the increase came from volunteers taking on work previously done by a staff member who left.

Without interpretation, visibility can become a form of blindness. The number is visible. The conditions that produced it are not. Organizations often respond by asking for more numbers. Break the total down by site. Add quality measures. Track employee sentiment. Include customer satisfaction. Create a risk indicator. Measure the measure’s unintended consequences. This can help.

A richer dashboard may reveal that reach increased while sustained support declined. It may show that one site improved because staff absorbed work another site could not. It may surface differences hidden inside the average.

But complexity creates its own pressure.

A dashboard with forty indicators becomes difficult to interpret. Leaders ask for a summary. The summary returns to a few colors.

Green. Yellow. Red. Compression reappears because the need for orientation remains. The solution is not a perfect dashboard. There is no perfect reduction of living work. The question is whether the organization still knows that a reduction has occurred. Healthy measurement keeps a visible distance between the indicator and the thing it represents. It says: This number tells us something important. It does not tell us everything important. It should change what we ask, not end the asking. That posture is harder to maintain when accountability arrives through the number.

A funder needs evidence that money produced value. A regulator needs consistent reporting. A board has legal duties. An executive cannot govern entirely through stories from people they happen to trust.

Legibility matters.

People affected by an institution deserve to know what it is doing. Staff should not be protected from scrutiny by claiming that their work is too complex to measure. Invisible labor can conceal devotion, but it can also conceal neglect. Professional judgment can be wise, but it can also become a shield against correction.

The alternative to bad measurement is not romantic trust. Trust without visibility can become permission. Visibility without trust can become surveillance.⁵ Collaboration must live between them. That middle is uncomfortable because it does not offer a single rule. Sometimes the dashboard is revealing what staff would rather not see. Sometimes staff are revealing what the dashboard cannot see. Sometimes leaders ask for clarity because a real failure has been hidden behind complexity. Sometimes they ask for clarity because they cannot tolerate a reality that remains uncertain. The form of the request may sound the same.

“Can we make this more measurable?”

“Can you show progress more clearly?”

“What would success look like?”

“How will we know whether this is working?”

These are good questions. They become dangerous when the group is allowed to answer only in advance. Important work often becomes clear after people enter it.

A community program may begin with the goal of increasing participation and discover that the deeper barrier is whether residents believe the institution will remain. A research project may set out to test one hypothesis and find that its most useful result is a question the original design could not have anticipated. A team may begin building a feature and learn that the problem cannot be solved without changing the surrounding process.

Premature clarity protects orientation by narrowing discovery. It tells people what the work must prove before the work has had time to reveal what matters. This can feel efficient. Everyone knows the target. Funding can be allocated. Progress can be tracked. The project can be evaluated. But some forms of clarity are purchased by excluding the possibility that the work will teach the organization something inconvenient.

The nonprofit’s board wanted to know whether the programs were effective. The dashboard answered by defining effectiveness as reach, completion, cost, and satisfaction. Those dimensions mattered. They did not include whether participants developed enough trust to return before a crisis. They did not include whether staff relationships with local institutions became more credible. They did not include whether a person learned how to navigate a system independently after the program ended.

Some of these outcomes could be measured eventually. Others would appear unevenly, in stories, behavior, and relationships that did not belong neatly to the organization alone. The more the board asked the dashboard to settle, the more the dashboard narrowed what the organization could recognize as success. This is how tools of accountability can quietly become tools of control.

Control does not always arrive as an order. Sometimes it arrives as a required field. A staff member cannot move forward until the form is complete. The form requires a category. The situation does not fit the available categories. The staff member selects the closest one. Later, leadership sees a clean distribution and designs policy around it. The ambiguity did not disappear. It was pushed downward into the judgment of the person completing the form. Templates do this constantly.

They help people remember essential information. They create consistency across handoffs. They make omissions visible.

They also decide which kinds of information can enter easily. Anything outside the structure becomes extra explanation. Extra explanation becomes work. Under time pressure, the extra disappears first. The form fills. The dashboard turns green. The organization becomes legible by becoming less faithful to the work.

This does not happen because forms are inherently dehumanizing. It happens because the living situation has more dimensions than the form can carry, and the organization forgets who absorbs the loss.

Usually, the person closest to the work does. They translate a complicated family into a service category. They compress a technical risk into a status color. They turn uncertainty into a date. They convert an unfinished relationship into a percentage complete. They make reality look manageable enough to travel upward.

Narrating the work for people who cannot see it is labor. Shared work requires stories that travel. The problem is that the people creating those stories are often judged by whether the stories make the system feel under control.

The update must be clear. The trend must be explainable. The plan must appear intentional. Uncertainty must be bounded. This creates an incentive not necessarily to lie, but to smooth. A delayed project becomes “at risk.” A fragile partnership becomes “requiring continued stakeholder engagement.” An exhausted team becomes “focused on prioritization.” A program that reaches more people with less depth becomes “scaling access.” The language remains defensible. The experience recedes.

Organizations often accuse leaders of using vague language to hide reality. Sometimes they are. But vagueness can also be a survival adaptation inside systems that demand certainty before certainty exists.

The report must travel through several levels. Each level needs a shorter version. The detail becomes a paragraph. The paragraph becomes a bullet. The bullet becomes a color. By the time the information reaches the board, the quarter is green. The process resembles the way a landscape becomes a map and then an icon on a phone.

Each stage removes information so action becomes possible. The executive director’s responsibility is not to refuse compression. The board cannot govern from thousands of unfiltered details. Her responsibility is to know when the compression has changed the moral meaning of the work. That moment often arrives when the summary begins denying the people inside it.

The staff say they are overwhelmed. The dashboard says people served increased. Participants say the program feels rushed. The dashboard says satisfaction remains above target. A partner says trust has weakened. The dashboard says referrals are up. These facts can coexist. The contradiction is not necessarily evidence that one side is wrong. It may reveal that the organization is observing itself on different clocks.

Numbers often respond quickly to formal change. Trust responds slowly. A new intake process may increase volume immediately and exhaust staff over several months. A reorganization may remove duplicated roles on paper while gradually eliminating the relationships that allowed problems to move across sites.

The dashboard sees the first clock because the first clock produces data. The second appears later as turnover, conflict, failure, or silence. By then, the original decision may have been declared successful. This is why the absence of negative data should not be confused with the absence of cost. People often stop reporting before they stop experiencing.

A staff member raises a concern once and receives an explanation of why the metric matters. They raise it again and are asked for evidence. The evidence is difficult to produce because the concern involves work the system does not count. The staff member continues for a while, then stops.

The dashboard remains green. The organization calls this alignment.

Seen from the boardroom, the quiet can look like agreement. The concern was raised. An explanation was given. Nobody continues to object. The metric still matters, and the room has moved on. The staff member may have stopped because they learned that their evidence would never count as evidence—not because the concern disappeared. What looks like alignment may be the moment when experience stops traveling upward.

Silence is especially easy to misread when the measurement system appears objective.

People do not feel entitled to argue with a number. They may distrust the way it was produced, but the number arrives with institutional authority. Their own experience feels anecdotal.

A field coordinator says, “Families are falling through the cracks.” A board member asks, “How many?” The question is legitimate. It may also end the conversation if the organization has no process for counting what the coordinator can already see. The absence of a number does not make the observation meaningless. It means the organization has not yet made that reality legible.

Sometimes the correct response is to build a measure. Sometimes it is to investigate. Sometimes it is to trust the person long enough to look. Relational trust is not the suspension of evidence. It is the willingness to treat someone’s perception as a reason to seek evidence rather than as a failure to provide it.

This matters most when people are reporting early signals. By the time every concern is measurable, the cost may already be large. A safety culture depends on near-misses that never became incidents. A healthy team depends on tension noticed before people leave. A durable partnership depends on changes in tone that do not yet appear in formal outcomes.

A responsive institution needs ways to hear information while it is still too incomplete for the dashboard. This is one reason informal conversations persist even in highly measured organizations. An executive director asks a coordinator, “What am I missing?” A manager notices that the usually talkative person has stopped speaking. A board member visits a program site without a presentation.

These actions can become performative. Proximity is not automatically understanding. A brief visit can produce more confidence than knowledge.

But direct contact interrupts the fantasy that the dashboard is the work. It returns texture. The board member hears how long a “successful referral” actually took. The executive sees that a supposedly streamlined process requires staff to maintain two unofficial spreadsheets. The manager learns that the project’s apparent progress depends on one person answering messages late at night. The number remains true. Its meaning changes.

At the nonprofit board meeting, the executive director paused on the slide showing people served. The green number still occupied the corner.

She could have continued. The report was accurate. The organization had met the target.

Instead, she said, “We need to talk about what this number cost.” The room shifted.

She explained that the increase came partly from shorter engagements. She described the coordinator who had gone on leave. She told the board that two sites were meeting volume goals while losing experienced volunteers. She did not claim the program was failing. She did not ask the board to disregard the dashboard.

She widened the meaning of green. One board member asked whether the organization should lower its targets. Another asked whether the staffing problem was temporary. The finance chair wanted to know how much deeper engagement would cost. The questions did not become less numerical. They became more connected to the work.

The board eventually requested a revised report. It would still include reach and cost. It would also distinguish brief contact from sustained support. Staff turnover would appear beside program growth. Narrative reports from site leaders would be discussed before the dashboard rather than appended after it.

No one believed the new report would capture everything. That was the change. The old dashboard had been treated as the organization made visible. The new one was treated as one view through which the organization could be questioned. Several months later, one of the numbers turned yellow. Volunteer hours had declined.

In the earlier system, the decline might have been interpreted as reduced engagement. The site report told a different story. Staff had stopped using volunteers to cover work that required paid expertise. Fewer hours reflected a healthier boundary.

The number had worsened. The organization may have improved. A dashboard cannot know the difference on its own. It can count. It can compare. It can warn. It can reveal patterns no individual would otherwise see. But it cannot decide what the pattern means, who paid for it, or what remained outside the frame. Those questions still belong to people. The board meeting ended with fewer green numbers than it had begun with. The room felt less certain.

It may also have understood more.

A dashboard is valuable because no one can see the whole.

It becomes dangerous when it persuades us that the whole is what we can see.

Footnotes

1. See James C. Scott, Seeing Like a State: How Certain Schemes to Improve the Human Condition Have Failed (New Haven, CT: Yale University Press, 1998). ↩

2. See Jerry Z. Muller, The Tyranny of Metrics (Princeton, NJ: Princeton University Press, 2018), on performance measurement redirecting effort toward what can be counted. ↩

3. See Scott, Seeing Like a State. ↩

4. See Luhmann, Trust and Power. ↩

5. See March and Simon, Organizations. ↩