Before
A nonprofit had noticed something reasonable. A national endurance event publishes its finishers, and the people who finish it are, by definition, people who will do something extremely difficult for a reason that matters to them. Many of them do it for a cause. Some of those causes are close to this one.
The ask was straightforward: turn that public list of names into a list of people who could be contacted.
I said I would find out whether that was possible before quoting anything to build it.
What the work actually was
Not a scraper. The extraction was the easy half and took an afternoon — the event’s own results endpoint is public and unauthenticated, and folding a decade of results into one record per person is ordinary work.
The real work was testing every route from a name to a way to reach that person, and writing down what each one did when probed rather than what its marketing said it did.
That distinction turned out to be the whole project.
What came back
Commercial data appending does not work on a name alone. Every consumer data vendor tested requires a postal address or a region as a mandatory input. A file of names fails validation before cost is even a question — and the cost was never the obstacle. Their terms separately prohibit appending contact details onto people who have no prior relationship with the buyer, which would have ruled it out even if the inputs had been there.
A name with no public footprint stays that way. Eight ordinary participants were searched by name against the open web. None resolved. A well-known athlete was searched as a control and resolved instantly, which confirmed the method worked and that the answer was simply no.
Roughly a third of the list could not be contacted at all. A large share of participants raced at Canadian venues. Canada’s anti-spam law requires express consent before a commercial message, with penalties into eight figures, and the charity exemption requires a Canadian registration a US nonprofit does not have. That cohort was identified and excluded before any list existed, rather than after somebody sent something.
Several other routes — photo galleries, sponsor activations, books, documentaries — were checked and closed for ordinary reasons. They did not contain what people assumed they contained.
The recommendation
Not the list.
What the research did find was that a small number of these people had already made themselves public — written first-person accounts naming the cause they climbed for, local newspaper profiles, fundraising pages they created themselves. Filtered down to those whose stated reason overlapped with this nonprofit’s actual work, the population was a few dozen people rather than several thousand.
A few dozen people who have publicly written about why they care about this specific thing are worth more than every cold name that was originally asked for, and the entire difference is that they said it first.
The recommendation was to work that list by hand, plus one relationship route that inverted the ask entirely: the event sells entries that sell out, partner organisations receive allocations of them, and holding something scarce is a better position than requesting money.
One more thing, because it mattered
A specific fundraising figure had been circulating in earlier research summaries and had started to be repeated as fact. I went back to the article it was attributed to. The article does not contain the claim, and the name it was built on almost certainly referred to a different person entirely.
It was removed, with a note explaining why, so it would not get repeated into a grant application or a press release. Finding a number that is wrong is worth about as much as finding one that is right, and it is a great deal cheaper than finding out later.
After
The work product was a document: every route, what was tried, what it returned, and a verdict of pursue, blocked, or dead, with the evidence for each.
Nobody built a system, and that was the correct outcome. The cost of the research was a fraction of the cost of building the thing it recommended against, and considerably less than the cost of sending several thousand messages into a jurisdiction that fines you for it.
I would rather be the person who says an idea does not work than the person who bills for building it anyway.