A data unit can build the right architecture, answer queries in milliseconds and still fail. The tool exists and works, and nobody uses it. The usual diagnosis (“people need training”, “we need one more dashboard”) treats the symptom as the cause. In this case the organization did not trust the numbers, and no feature was going to change that on its own.
This series documents how that credibility gap closed in a regional commercial operation. It reads the case from angles that rarely meet in one place: product engineering, the sociology of deviance, and what the experience teaches about building a technical career.
What the series covers
Where it starts
Before convincing anyone, you have to decide what is being measured and why. The Commercial Compass and the single number (in Spanish) argues that one defensible metric beats fifteen dashboards that contradict each other, and describes what you lose by picking it badly.
The whole story
How I made them trust our data tells the case from beginning to end. It covers the internal business school, the move from report provider to technical reference, and what we did while trust still did not exist.
The adoption mechanism
Adoption engineering, not software takes apart the persuasion stack that moved adoption. Friction was removed, value became visible in the first minute, status signals appeared, and users could verify the numbers themselves. None of those levers is technical, and every one of them was designed.
The criminological reading
Useful organizational deviance borrows labeling theory to explain why an organization treats the first users of a new tool as deviants, and why that deviance is exactly what needs protecting.
The career lesson
Building evidence before consensus generalizes the pattern. Build the proof before asking for permission, and learn to tell when that order moves a large organization and when it turns into governance debt.
Who it is for
For people who build data products inside a company and find that the hard part begins once the code works. Technical leads, data architects and analytics owners who need their work to be used will recognize most of it.
The case is real and anonymized. The numbers and sequences are the ones that happened; the names are not.