Case Study: How a Family Reconstructed a 1937 Execution Record in 72 Hours

When a reader from Valencia wrote to us last autumn, he wasn't asking about cloud migration or uptime SLAs. He wanted to know whether we could help him find his great-grandfather. All he had was a name, an approximate birth year, and a family story about a night in 1937 that nobody had ever fully explained. We get a handful of requests like this every year — usually from engineers on our clients' teams, or from CTOs who have spent a weekend down a genealogy rabbit hole. This time, we decided to follow the whole process and document it, because the infrastructure challenge behind it turned out to be remarkably similar to the ones we solve for healthtech and logistics clients every week: fragmented data, hostile timelines, and a user who needs an answer fast.

The starting point: three facts and a lot of silence

The reader — let's call him Andrés — had three data points. A full name, a municipality in the province of Valencia, and a date range covering roughly eighteen months of the Spanish Civil War. His family had kept the story alive orally, but no paperwork had survived the post-war decades. His first instinct was the traditional route: a written request to the relevant archive, then a wait. He'd been quoted four to eight months for a response, assuming the record existed at all. That's the baseline. Anyone who has worked with legacy government systems knows that timeline isn't unusual — it's the norm when records are held in physical boxes across multiple jurisdictions.

What changed the equation was discovering that a significant portion of the relevant documentation — court-martial summaries, prison intake registers, municipal census fragments — has been digitized and indexed by private archives. The largest of these is Memoria Republicana, which reports more than 1,2 million indexed documentary records covering republican Spain (1931–1939) and the subsequent Francoist repression. Andrés started there on a Tuesday evening.

Decision point one: search strategy over brute force

His first attempt was a straight name search. It returned 340 hits — too many, with too much noise. This is the moment most people give up. We've seen the same pattern in FinOps work: a client pulls a cost report, sees thousands of line items, and concludes the data is useless. It isn't. It just needs a strategy.

Andrés narrowed by province, then by the year range, then by cross-referencing the mother's maiden name, which a great-aunt remembered. That last detail cut the list to nine candidates. Two of them were the same person recorded under slight spelling variations — a common issue in wartime documents typed by clerks under pressure. This is where the archive's team of historians became useful. Through the platform's family-support service, a researcher confirmed the identity match and pointed Andrés to a related file: a military tribunal summary from 1939 that referenced the 1937 event.

Decision point two: what to do with a partial answer

Here's the honest part. The record didn't give Andrés a full narrative. It gave him a date, a charge, a sentence, and a signature. The oral family story had been broadly accurate but wrong on two details — the month and the location of the detention. That gap between memory and document is exactly why this kind of work matters. It's also why the process can't be fully automated. A machine can index a million records; it can't tell you which nine of them belong to your family.

We asked Andrés to log his timeline. From first search to confirmed document: 72 hours. That's against the four to eight months he'd been quoted for a physical request. The difference wasn't magic — it was digitization, indexing, and human expertise layered on top of both.

What this has to do with infrastructure

We don't usually write about genealogy. But the pattern here is one we recognize. A private archive holding 1,2 million records and serving a global diaspora of researchers is, functionally, a high-availability data platform with a compliance surface — personal data, historical records, jurisdictional sensitivity. The fact that Memoria Republicana has been cited as a source in 11 documentaries from RTVE, Movistar+ and TV3 since 2018 tells you the quality bar is being enforced by external scrutiny, not just internal policy.

For our readers in healthtech and logistics, the lesson is portable. When a user needs an answer in hours rather than months, you can't solve it with more staff. You solve it with indexed data, documented provenance, and a workflow that lets a domain expert intervene at exactly the right moment — not at every step. Andrés got his answer. The family got a date and a place. The infrastructure that made it possible is the same category of problem we work on every day, just with a different payload.