The report that only looks backward

A packaging manufacturer, 200 people, three plants (a mid-size Brazilian company, figures in Brazilian reais). Every 12th of the month, the controller sends the prior month's close: revenue, margin by product line, delinquency. It's a good report, carefully built, and leadership discusses each line for an hour and a half.

What nobody can explain in that meeting is why the flexible-packaging line's margin dropped two points. Hypotheses come up — customer mix, rework, a new operator, the price of film — but none of them are measured. The report reports the result down to the cent and says nothing about the cause. And, most importantly, nothing about December.

01The idea: four perspectives, not four reports #

Robert Kaplan and David Norton published The Balanced Scorecard — Measures That Drive Performance in Harvard Business Review in 1992. The argument grew from a simple observation: financial indicators are the result of decisions made months earlier, so they work well for accountability and poorly for steering.

The proposal was to track the company through four simultaneous perspectives: financial (how we look to shareholders), customer (how the customer sees us), internal processes (what we need to excel at), and learning and growth (how we keep improving). The idea wasn't to add up reports, but to link the four into a cause-and-effect chain: skilled people run better processes, which deliver to the customer what sustains the financial result.

In 1996, the same authors went further and began treating the scorecard as a strategic management system, not just a measurement tool: a way to translate strategy into objectives each area could recognize in its own work. That second part is usually what gets lost in implementation.

02The four perspectives without a BI team #

The packaging company has no data analyst. It has a controller, an ERP and several spreadsheets. That doesn't rule out a scorecard — it rules out a 40-indicator scorecard. With two indicators per perspective, eight total, they can start next week:

• Financial. Contribution margin by line and average collection period. Both already exist in the ERP; they just need to be looked at alongside the other six.
• Customer. Percentage of orders delivered on the promised date and number of formal complaints per month. Both come from what the operation already logs.
• Internal processes. Rework rate and time between order approval and start of production. These are the two points where the margin leaks before it shows up in the close.
• Learning and growth. Number of critical roles with more than one trained person, and training hours actually delivered. These measure whether the company is becoming less dependent on any one individual.

Notice what the list does to the December discussion. Rework and delayed production starts rise before the margin drops; single-person dependency explains why one absence takes down a plant. These are indicators that precede the result, not alternate versions of it. It's the same logic of looking at the bottleneck before the whole that we discussed in improving everything is improving nothing.

03The cost of waiting for the close #

Stated assumptions, at the scale of the 200-person company — redo this with your own numbers:

The first line: a problem that occurs on the 1st is only discussed on the 12th of the following month, averaging 42 days between event and conversation. The third: the controller and two analysts spend about 16 hours a month closing spreadsheets, which adds up to 192 hours a year; at a fully loaded R$ 120 an hour, plus rework on corrections, that's close to R$ 96,000. The fourth is the actual figure from the example company, and each margin point on that line was worth about R$ 140,000 a year.

Want to see how this looks inside a real operation? Explore the platform.

04The classic mistake: the bureaucratic scorecard #

The version that fails is always the same. Someone proposes a dashboard, every director wants to make sure their own area is represented, and what was supposed to be eight indicators becomes 40. Nobody looks at 40 indicators. The dashboard starts getting updated out of obligation, goes stale, and within eight months becomes a tab only the intern opens.

1. Indicator with no owner. If no one person is responsible for explaining the number's movement, the number becomes decoration.
2. Indicator with no associated decision. Before adding it, answer: what would we do differently if this number got worse? No answer, it doesn't make the list.
3. Manual collection. Any indicator that depends on someone filling in a spreadsheet on Friday dies by the third month.
4. Disconnected perspectives. Four groups of numbers with no cause hypothesis between them don't form a scorecard; they form four stapled-together reports.
5. Target as a stick. As soon as the number becomes a bonus target, it starts being managed instead of improved — the problem we cover in when the target becomes the goal.

The academic critique is worth noting. Hanne Nørreklit, in 2000, argued that the cause-and-effect relationship among the four perspectives is assumed by the authors, not demonstrated, and that confusing correlation with causation can lead a company to chase indicators that don't move the result. The critique is fair and useful: the links in your scorecard are your company's hypotheses, and should be treated as testable ones.

05What management with data answers #

What makes the scorecard unworkable in a mid-size company is almost never the theory: it's the data collection. When rework, delivery time, customer incidents and training are logged in different systems — or in none — building the dashboard costs more than the decision it would inform.

When work is logged where it happens, the indicator becomes a consequence, not a project. Delivery time comes from the order flow itself; rework comes from the incident log; single-person dependency comes from who runs each critical task. The close still happens — but it stops being the only moment the company knows itself. This connects directly to the gap between strategy and execution.

At Relevanti, the operations, process and intelligence modules keep those numbers alive with no intermediate spreadsheet — see the platform, the breakdown by area in solutions, or bring your case to a conversation.

A report that only looks backward is honest and insufficient. Kaplan and Norton's point was never to add more numbers: it was to measure what there's still time to change.