Your dashboard says everything is fine. Where do those numbers come from?

Contents

• In 1 minute
• Can a company improve on everything and still lose value?
• What did Kaplan and Norton propose, in two lines?
• Why does a good dashboard end up as a spreadsheet nobody believes?
• What does the right order look like in a Mexican company?
• What is different about measuring in Mexico?
• How far off is your company? A diagnostic with thresholds
• What breaks in the systems, and what is best decided beforehand?
• Which mistakes cost the most?
• What should you take to the meeting? The sheet for each indicator
• Where do you start on Monday?
• Key takeaways
• What Hábil does

Example scenario (it does not describe a real company): it is the 12th and the monthly results meeting is starting. Sales says they sold 18.4 million. Finance says 17.1. Operations brings a third number, because it counts what left the warehouse.

All three are right. The first hour goes into working out which number is "the right one." The decision that was going to be made, whether or not to hire two salespeople, is put off until next month.

This article is not about pretty dashboards. It is about something less eye-catching and more expensive: where the numbers you decide with come from.

One clarification before we start: Hábil did not take part in the cases mentioned here and has not worked with those companies. We tell them using what the model's authors, the SAT (Mexico's tax authority) and INEGI (Mexico's national statistics institute) published.

In 1 minute

• The risk: deciding with indicators that arrive late, that each area calculates differently and that someone puts together by hand in a spreadsheet. You think you have control, and what you have is an opinion with decimals.
• What the model says: in 1992, Kaplan and Norton proposed measuring the company with a few indicators on four fronts, not just the financial one. And they warned that an information system that is not responsive can be its weak point [1].
• The case: an electronics company improved its quality tenfold between 1987 and 1990, and its share price fell to a third [1]. It measured a lot; it did not measure what connected with the money.
• What inaction costs: the VAT (IVA, Mexico's value-added tax) you cannot credit because the receipt or the payment complement is missing comes out of your cash. At the general rate of 16%, one million pesos of poorly documented purchases means 160,000 pesos of VAT at risk [8][9]. Add the hours of whoever puts the package together by hand, at their hourly cost.
• Your decision for Monday: stick to a few indicators (the authors describe scorecards of 15 to 20 per business unit [3]; you should start with five) and, before buying any dashboard, require each one to have its source system, its owner and its definition in writing.

Can a company improve on everything and still lose value?

Yes: a company can turn its operating indicators green and still not make more money.

Between 1987 and 1990, an electronics company listed on the New York Stock Exchange did almost everything right. That is how Robert Kaplan and David Norton tell it in their 1992 article [1]:

Operations improved tenfold in quality. The financials barely moved, and the share price collapsed [1]. According to the authors, the company was very slow to bring out new products and did not go looking for new, perhaps more demanding, customers; that kept it from taking advantage of the improvements [1].

In the 1992 article that company has no name. Kaplan later told a similar story, this time with a name: that of Analog Devices, a semiconductor manufacturer [5].

In 1990, Analog's vice president of quality and productivity, Arthur Schneiderman, told Kaplan and Norton that all the quality indicators on his dashboard had improved markedly. And that, at the same time, the share price had fallen by about 70% in three years [5]. Analog served as the prototype for the model [3].

The lesson is not "quality does not matter." It is something else: measuring a lot is not the same as measuring what connects with the money. And to connect it, the numbers from the plant, from sales and from finance have to talk to each other.

What did Kaplan and Norton propose, in two lines?

The model fits in two lines: a few indicators, on four fronts, that tell you today how you will do tomorrow.

The balanced scorecard, known in Spanish as cuadro de mando integral, measures the company with a handful of indicators on four fronts, not just the financial one. The financials say how things went; the other three fronts warn how they are going to go [1].

Robert Kaplan, of Harvard Business School [5], and David Norton published it in Harvard Business Review in the January–February 1992 issue [1][2]. It came out of a one-year research project with 12 companies [1][5].

They compare it to the dials in an airplane cockpit: a set of related indicators that, together, give a more complete view [1]. The four fronts:

In 1992 the fourth front was called Innovation and Learning; in later texts, the authors called it Learning and Growth [1][4]. The example indicators in the table are ours.

Two details of the original article matter more than the diagram. The first: the authors insist on limiting the number of indicators [1]. A year later, they described scorecards of 15 to 20 indicators per business unit, and said that reading them should reveal the unit's strategy [3]. It is neither a ceiling nor a minimum.

They also tell of an executive who mocked the habit of adding an indicator every time someone had a good idea, comparing it to a program for cutting down one more tree at his company [1].

The second detail lies in the systems the numbers come from.

Why does a good dashboard end up as a spreadsheet nobody believes?

The scorecard can fail not because of the indicators it chooses, but because of the systems it pulls them from.

In the 1992 article, Kaplan and Norton illustrate their method with a semiconductor manufacturer they call, using an invented name, Electronic Circuits Inc., or ECI [1]. Two scenes from ECI will sound familiar.

Scene 1: "on time" did not mean the same thing to anyone. ECI counted a delivery as on time if it arrived within a seven-day window. Some customers used five days; others, nine. ECI disappointed some and made extra effort for others [1]. The indicator existed. The definition did not.

Scene 2: the dashboard reached the meeting late. The biggest concern of ECI's executives was that the information did not arrive on time: the reports were generally a week behind the management meetings [1].

ECI, the authors say, lacked an operational information system that would provide that data [1].

If the information system is not responsive, they warn, "it can be the Achilles' heel of performance measurement" [1].

In their implementation guide, one of the steps is exactly that: linking the indicators to their databases and information systems [3].

The warning of an executive who used it. In an interview with Kaplan published in 1993, Larry Brady, of FMC Corporation, said the mistake was to treat the scorecard as a new measurement system that ends up calling for hundreds and thousands of indicators and a big, expensive executive information system. For him, it had to be the core of the way the company is run, not a reporting system [3].

Translated to your company: the problem may not be the lack of a dashboard. It is that each number on the dashboard is put together by a person, by hand, cross-checking three systems that do not talk to each other. That is why it arrives late, why it does not add up and why nobody believes it.

What does the right order look like in a Mexican company?

First you decide what to measure and where each number comes from; the screen comes last.

We looked for a public case of a Mexican company that told, with figures and in order, how it connected its indicators to its systems. We did not find one: the annual reports we reviewed mention new dashboards or systems, but not that path with numbers. That is why what follows is a scenario, marked as such.

Example scenario (fictitious company; the figures without a reference number are illustrative, not an average or a promise): an electrical supplies distributor with 140 people and two legal entities. It invoices in one system, keeps its accounts in another and reconciles the bank in spreadsheets. It is the company from the meeting on the 12th.

Use it as a template: replace the illustrative figures with your own in three formulas.

• What the package costs you today = hours of data entry per month × hourly cost of whoever does it. In the scenario, at 250 pesos an hour: 60 × 250 = 15,000 pesos a month before, and 15 × 250 = 3,750 after.
• VAT at risk = VAT on the month's payments without a CFDI (Mexico's electronic invoice) + without a payment complement + to suppliers on the final 69-B list (the SAT's list under that article of the CFF, the Código Fiscal de la Federación or Federal Tax Code). At the general rate of 16% [8], the scenario's 240,000 pesos are equivalent to 1.5 million pesos of poorly documented purchases.
• What the project costs = internal hours × hourly cost, plus the software. In the scenario: 209 hours × 250 = 52,250 pesos of internal time, recovered in less than five months of data entry saved.

Two clarifications. The 12th or the 5th is when the package reaches management. Do not compare it with APQC's median of 6.4 days, which measures something else: from the trial balance to the consolidated statements [12].

And the timeline. At National, the insurer the authors describe, it took 30 months [4], but there the whole management system was rebuilt. Here only five indicators are connected. That is why it takes 12 months.

Note the order. The screen arrives in month 12, not in month 1. What is expensive in the scenario is not the software: it is the definitions, the owners and the connections between systems.

It is the same order the authors suggest: a few indicators, linked to the strategy and to their information systems [1][3]. And it is exactly what Brady asked not to reverse: first the way of running the company, then the tool [3].

What is different about measuring in Mexico?

In Mexico, much of the raw material for your financial front already exists at the SAT; what is missing is cross-checking it against your bank and your accounting system.

Kaplan and Norton wrote in 1992. They knew nothing of the CFDI. In Mexico, every invoiced sale and purchase already exists in digital form: in 2023 more than 10,323 million electronic invoices were issued, an average of 327 per second, according to the SAT [6][7].

The advantage: the raw material is already digitized. The catch: cross-checking those invoices against what you collected, paid and recorded is up to you, and it does not happen by itself.

Five tax rules turn that cross-check into money:

• VAT has several requirements before it can be credited. Among them, the Ley del IVA (LIVA, Mexico's VAT law) requires the tax to be shown separately on the tax receipt and to have actually been paid in the month. There are others: that the expense be strictly indispensable and, in certain cases, withholdings paid over or additional documentation [8].
• Each deferred payment carries its own receipt. If an invoice is not paid in a single installment, or is paid after it is issued, another CFDI is issued for each payment: the so-called payment complement [9].
• If you collect and do not issue the complement, the offense is yours. Failing to issue a CFDI that the law requires is an offense, if the authorities discover it in an audit: a fine of 22,300 to 127,530 pesos for a company under the general regime and, for a repeat offense, possible closure of the business for 3 to 15 days [9]. And your customer is left without the receipt for the payment on which its VAT depends.
• A deduction needs its receipt on time. The Ley del ISR (LISR, Mexico's income tax law) requires the deduction to be supported by a tax receipt, obtained no later than the day you must file your return [10].
• A supplier on the final 69-B list puts your receipts at risk. Its transactions are deemed to have no tax effect. Whoever gave them effect has 30 days from publication to prove that they did receive the goods or services, or to correct their situation with amended returns [9].

This summarizes the law for measurement purposes. It is not a substitute for review by your accountant or tax advisor.

From there come three very Mexican indicators for your financial front. The construction is ours, based on those rules:

None of them comes from a single system. If someone puts them together by hand at every close today, that is your Achilles' heel.

How far off is your company? A diagnostic with thresholds

If an indicator arrives a week late or one person puts it together by hand, treat it as a red flag.

Use this table at your next meeting. The thresholds marked as indicative are our own judgment.

A figure from Mexico for calibration: according to INEGI, with 2017 data, 28.2% of Mexican SMEs tracked three to five key indicators and only 8.8% tracked ten or more [11].

Few companies measure a lot. If yours tracks few indicators, the challenge is not to cut: it is to choose those few well and make each one hold up. If you are in the 8.8%, start by cutting.

What breaks in the systems, and what is best decided beforehand?

Each front breaks in a different system, and each one calls for a technology decision before buying any dashboard.

The table is our own judgment, not the model's.

Note that none of these decisions is "buy a business intelligence tool." First, the systems pass the data to each other on their own; then, if needed, the screen.

Which mistakes cost the most?

The expensive mistakes are not technical: they are management decisions made in the wrong order.

1. Buying the dashboard before having the source. Consequence: a new screen that someone feeds with the same old spreadsheet. Brady warned about it as early as 1993 [3].
2. Measuring everything. Consequence: dozens of indicators and no decision. It is the habit that executive mocked as a program for cutting down one more tree at his company [1].
3. Letting each area define its own indicator. Consequence: three sales figures in the same meeting. At a bank the authors call Metro Bank, its 25 executives agreed on the wording of the strategy, but each had their own definition of what superior service was [4].
4. Believing that green operations ensure the money. Consequence: the company from 1987 to 1990, with its quality multiplied and its share price at a third [1]. The authors say it bluntly: not even the best set of indicators ensures a winning strategy [1].
5. Paying bonuses on data nobody can trace. Consequence: arguments about the number instead of about performance. Describing an oil company that tied 60% of its executives' bonus to four financial indicators, the authors ask whether the company has valid and reliable data for those indicators [4].
6. Delegating it only to the controller's office. Consequence: one more report. When the controllers and finance vice presidents in the original research took it to their companies, they found they could not implement it without senior management [1].
7. Expecting results in a quarter. Consequence: abandoning it before it works. In the case of an insurer they describe, the complete system was built step by step over 30 months [4].
8. Leaving tax out. Consequence in Mexico: VAT at risk of not being creditable because the receipt is missing or the payment is not documented [8][9], or deductions with a supplier on the final 69-B list that you will have to defend or correct within 30 days [9].

What should you take to the meeting? The sheet for each indicator

No indicator goes onto the dashboard without a nine-field sheet; if a field is left empty, that is the next project.

This sheet is ours. It sums up what Kaplan and Norton ask for (a few indicators, linked to the strategy and to their systems [1][3]) and what Mexico requires.

Example scenario (illustrative figures, not from a real company): this is what a completed sheet would look like.

Fill in the sheet for your five most important indicators. The empty cells in rows 4, 5 and 6 are your list of systems projects, in order.

Where do you start on Monday?

Start with five indicators, not twenty, and with the one that costs you the most hours of manual entry.

1. Ask for the list of all the indicators that reach management today. Count them. If there are more than 20, you already have more than the authors describe for an entire business unit [3]: your first decision is to cut.
2. Choose five, at least one per front. Ask for each one: "what decision did I make last month with this number?" If nobody remembers one, it comes off the list.
3. Fill in the sheet for those five with whoever puts them together today. Not with whoever presents them.
4. Measure the delay: how many days after the close does each one arrive? Compare it with the diagnostic table.
5. Rank the projects by hours of manual entry and by tax risk. The first on the list is the first one worth connecting.

Key takeaways

Three questions for your next executive committee:

1. How many indicators do we review here, and what decision did we make with each one last quarter?
2. Of our five main indicators, how many does a person put together by hand, cross-checking two or more systems?
3. How much VAT paid is at risk of not being creditable, and do we know it every week or find out at the close?

What Hábil does

Hábil has been engineering critical systems since 2006. We do not build KPI dashboards: we work on what comes before, getting the systems your numbers come from to talk to each other, so that the indicators arrive on time and with the same definition.

As far as this article goes, that means two things:

• Cuadre, for the tax front. Cuadre continuously reviews what your suppliers invoice you. It downloads from the SAT the CFDI your company issues and receives, including payment complements, and reconciles on three fronts: complements against invoices, with the balance of each one; what the SAT says against what your accounting system recorded; and your bank deposits against your CFDI, using the bank statement you upload. It shows you which complement you still need to request to document the payment on which the VAT you credit depends, and leaves in plain view what does not reconcile so you can resolve it before the close. Learn about it at habil.mx/en/cuadre.
• Integrating your systems. We connect your ERP, your invoicing, your banks and your channels, with each operation between systems recorded with its response, its time and the reason if it failed. That way, sales, invoicing and collections can stop being three numbers. See the Integration and APIs page.

If you want the details of why a payment complement throws off your close, read Payment complement and cancellations.

If you filled in the sheet and were left with empty cells, let's start there. Request your first conversation, at no cost. We review with you which systems your five main indicators come from, and you take away the map of the connections between your systems and the points where money, time or the audit trail is probably being lost. Write to us on WhatsApp.

Frequently asked questions

What is the balanced scorecard?

It is the framework that Robert Kaplan and David Norton published in Harvard Business Review in January–February 1992: a small set of indicators on four fronts (financial, customer, internal processes and learning) that complements the financials with signals of what is coming [1][2].

Is it the same as a business intelligence dashboard?

No. The scorecard is a way of deciding what to measure and what for. A dashboard is just a screen. An executive who used it warned precisely against confusing it with a big, expensive information system [3].

How many indicators should it have?

The authors describe scorecards of 15 to 20 per business unit [3]; it is a description, not a ceiling or a minimum. To start, five well-sustained indicators are worth more than twenty put together by hand.

What is a strategy map?

It is a visual framework, proposed by the same authors in 2000, that organizes the objectives of the four fronts [13]. It helps explain the strategy; it does not replace the data.

Do I need an ERP to have a scorecard?

Not necessarily. What you need is for each indicator to have a reliable source system and for the systems that have to be cross-checked to exchange data without manual entry. Sometimes that is a common ERP; sometimes it is connecting the systems you already have.

Sources consulted on October 22, 2026. The paraphrases of Kaplan and Norton's articles are our own; the only direct quotation is a brief excerpt, in English, from the 1992 article.