What 1.3 million payments revealed about a Wednesday in February

The pagoPA dataset from Regione Lombardia is an open and continuously updated record of public sector transactions. Every digital payment to the public administration flows through it: fines, healthcare fees, taxes.

We analyzed 1,331,545 transactions between January 1 and March 30, 2026, for a total value of €264,617,039.66. For most of the period, the system behaves as expected: fairly stable daily volumes, a few million euros per day, no significant deviations.

Then February 26.

On that single day, €18,919,986.54 were processed, more than four times the average. This is not a gradual increase, but a sharp jump that breaks the pattern of the previous days.

Daily trend: transactions vs. total amount (EUR), Jan–Mar 2026

At that point, the question is simple: what happened that day?

Looking into the data, the signal appears immediately. The largest transactions are not spread across time or categories. They are concentrated. They share the same type (RL_IST_TASSA_UNIV), pass through a small number of PSPs, and point to the same beneficiary.

This is not a general increase in activity. It is alignment.

Most of the volume on that day is generated by a small number of high-value transactions, all belonging to the same category: university taxes. The dataset does not include an academic calendar, but the pattern is clear: same type of payment, same intermediaries, same moment.

There is nothing anomalous, and nothing suspicious. The system is not flagging an issue. It is simply reflecting a deadline that materializes all at once.

The interesting part is not the spike itself, but how it gets explained.

We did not start from a hypothesis. We did not write manual queries or follow a predefined path. The system automatically isolated the outlier day, identified the relevant dimensions, and followed the most informative branch of the analysis: type, amounts, concentration.

From there, the explanation emerges.

We enjoy working on datasets like this one: public ledgers that nobody really reads, but that contain precise, verifiable stories. Starting this month, we are publishing one analysis like this every month, always drawing from open data. If you have a dataset you think deserves a closer look, or an idea worth exploring, we would love to hear from you.