Three Dashboards. Three Truths.

Part 1. When Everyone Was Right

Every executive likes dashboards until the day they disagree.

For years, business intelligence has been sold as the single source of truth. Vendors promise that modern data platforms eliminate uncertainty by bringing information from dozens of systems into one place. Dashboards become cleaner, reports become faster, and executives gain confidence that every strategic decision is backed by data rather than intuition. The promise is compelling because it sounds logical. If everyone looks at the same numbers, everyone should reach the same conclusion.

Reality is considerably less cooperative.

One Tuesday morning, the executive committee of a rapidly growing financial company gathered to review quarterly performance. It was not an emergency meeting. Revenue was growing, customer acquisition remained healthy, and operational metrics looked stable. The board expected a routine discussion about expansion plans for the following quarter. Instead, the meeting became a lesson that every CTO should experience exactly once in a career—and preferably never twice.

The CFO opened the financial dashboard first.

According to the finance department, quarterly revenue had reached €82.4 million, comfortably exceeding the original forecast. Cash flow looked healthy, collection rates had improved, and the company appeared ready to accelerate investment in new markets. From a financial perspective, the quarter had been an undeniable success.

The Chief Marketing Officer projected a completely different dashboard onto the meeting room screen. Marketing attribution, built on Google Analytics 4 and AppsFlyer data stored inside BigQuery, reported €86.1 million in attributed revenue. That difference was too large to ignore. Marketing immediately celebrated the number because higher attributed revenue meant significantly better campaign efficiency and a stronger return on advertising investment.

Before anyone had time to compare the reports, the Head of Business Intelligence quietly opened the company’s executive dashboard.

Revenue.

€79.8 million.

Nobody spoke for several seconds.

Three dashboards.

Three different answers.

One business.

The room became noticeably uncomfortable because every dashboard had an excellent reputation. None of them was built by inexperienced analysts. Finance trusted its ERP system because every transaction could be traced to actual payments. Marketing trusted attribution reports because campaigns were optimized using those metrics every single day. Executives trusted the corporate dashboard because it was built on top of the centralized data warehouse that had taken nearly two years to design.

If one report had obviously looked unreliable, the meeting would have ended within minutes.

Instead, every report appeared completely credible.

That is precisely what made the situation dangerous.

Technology leaders often assume that conflicting numbers indicate poor engineering. Surprisingly, that is not the most common reason. Mature organizations rarely suffer from broken dashboards. They suffer from something far more subtle: perfectly functioning systems answering completely different questions.

The CTO resisted the temptation to ask which dashboard was correct.

Instead, he asked every department to answer a much simpler question.

“What exactly does your revenue number represent?”

The finance team answered first.

“Our report includes only completed payments that have been successfully settled through the banking system. Refunds are deducted immediately, and every value matches accounting records.”

Marketing gave a different explanation.

“Our dashboard measures revenue generated by customers whose purchases can be attributed to advertising campaigns. We don’t care when accounting recognizes the payment. We measure marketing performance.”

The BI team smiled.

“We report confirmed business revenue after applying the company’s official transformation rules. We exclude duplicate transactions, cancelled applications, fraudulent orders and incomplete payment events before publishing executive metrics.”

Each explanation sounded perfectly reasonable.

Each represented a valid business definition.

Each measured something valuable.

None measured exactly the same thing.

For years, every department had been optimizing its own processes using its own interpretation of revenue. Nobody had deliberately created conflicting definitions. They had simply evolved independently as the company expanded. Finance optimized financial reporting. Marketing optimized advertising performance. Business Intelligence optimized executive reporting. Each team solved its own problem extremely well, but nobody had stopped to ask whether everyone was still speaking the same language.

Ironically, the company had invested millions of euros building a modern cloud platform while leaving one of the simplest questions unanswered.

What exactly does the word “Revenue” mean?

The engineering investigation began that afternoon. Data engineers compared ETL pipelines, analysts reviewed transformation logic, and architects traced the metric from source systems to executive dashboards. BigQuery tables matched perfectly. Dataform transformations completed successfully. Cloud Run jobs executed without failure. Data quality tests reported no anomalies. From a technical perspective, the platform behaved almost flawlessly.

That discovery eliminated what many people initially suspected.

This was not a Google Cloud problem.

It was not a BigQuery problem.

It was not even a data quality problem.

The technology had faithfully implemented the business rules it had been given.

The uncomfortable possibility was beginning to emerge that the platform had never been wrong.

Perhaps the organization itself had defined three different versions of the truth.

That hypothesis sounded almost absurd.

Unfortunately, it also explained every number displayed in the boardroom.