The obituary is data now — build the page so machines quote yours

In 2021, medical researchers built a system that read 12,407 ordinary obituaries and reconstructed the families inside them — names, residences, birth and death dates, and seventy-one kinds of kinship across eight generations, at 93% accuracy. An earlier team hit 97% on the kinship alone, building family-relationship data for genetic research out of the same raw material. Neither system was exotic. They worked because an obituary is structured data in prose clothing — and every machine that matters, from Google’s crawler to ChatGPT to the genealogy platforms, reads your obituary pages the same way. The only open question is whether the copy that gets parsed and quoted is the one on your website.

What does a machine see in an obituary?

A record it already has a vocabulary for. Schema.org’s Person type — the shared data dictionary the search engines maintain — carries fields that map onto an obituary almost line for line: birthDate and deathDate, birthPlace and deathPlace, given name, family name, middle name, honorifics, and the survivors as first-class relations — spouse, parent, children, sibling. The vocabulary might as well have been designed from a funeral program.

That’s why the obituary punches above its weight in machine-read visibility. AI answers reward fact density — the one peer-reviewed study of generative-engine optimization found sourced facts and statistics lift a page’s AI visibility by as much as 40% — and an obituary is fact-dense by nature. When a family asks an assistant about a person, or searches a name with “obituary” and a town attached, the machine’s answer is assembled from pages exactly like the one your site publishes. For an ordinary person there’s no Google knowledge panel and no biography card — the obituary page is the record that surfaces.

The mechanics that make your copy the quotable one

Every item below is something to hand your web person, and each exists because a machine can only quote what it can parse.

Real text, never text inside an image. A scanned obituary or a designed memorial graphic looks the same to a family and reads as nothing to a parser. The accessibility standard behind this — WCAG’s rule against “images of text” — exists because assistive technology can’t read them; the practical extension is that a date, a maiden name, or a service time locked inside a JPEG is not reliably machine-readable, by a screen reader or a search engine. Every fact on the page should be selectable with a cursor. That’s the test.

Full names, with the maiden name, in the visible text. “Mary (Sullivan) O’Brien” is findable by the granddaughter searching Sullivan in 2040 and by the parser matching a genealogy query today. There’s no dedicated maiden-name field in the data vocabulary — the visible text is where it lives, so write it there.

Dates and survivors written out plainly. “March 4, 1938 – July 19, 2026” as text, and the survivors listed the ordinary way — spouse, children, siblings, grandchildren by name. The kinship extractors hit 93% on exactly this kind of plain listing; what they can reconstruct, a family historian can too.

A photo near the text, with alt text that names the person. Google reads images through their alt text, the surrounding page, and computer vision — name the person in the alt attribute and keep the photo beside the obituary it belongs to.

An honest Person markup layer underneath. JSON-LD carrying the same names, dates, and relations the visible page states. Google is plain about what structured data is for: “to understand the content of the page” — you add it so the machine reads the record correctly, both channels saying the same true thing.

One honesty note, because vendors sell the opposite: there is no obituary rich result. Google’s supported list of special search features has no Person or obituary type, so markup will not earn a badge or a card — and a funeral service doesn’t qualify for Google’s event cards either, which require publicly bookable events. Anyone promising “obituary schema” as a magic SERP feature, or quoting a “3× more AI citations” multiplier for it, is selling a number nobody has ever measured. The win is comprehension. That’s enough.

One person, one URL, forever

The mechanics above make a page quotable. Permanence makes it the record — and permanence is where obituary archives quietly die. Pew Research measured the baseline: 38% of the webpages that existed in 2013 were gone by 2023, and a quarter of everything published across that decade has already vanished. An obituary archive with no permanence plan is on that clock.

The killer is the website change. The documented version of the story: a funeral home leaving its website vendor was handed its own obituary data as “a nearly unreadable” dump — with two days’ notice before the old site went dark. A rescuer managed to script all 2,000 obituaries into a new system and warned that most vendors “will not be so flexible.” The industry term worth knowing is closed versus open obituaries: closed archives only work on the vendor’s platform, which means the archive is the vendor’s leverage; open ones export and follow you. It’s the single sharpest question to ask before signing any website contract — are the obituaries mine, in a form I can take with me?

Two rules keep the record alive through any change. Give every deceased person one permanent page of their own — one person, one URL, never reorganized away. And if the site ever moves, redirect every old obituary URL permanently; Google’s own migration guidance says 301s held for at least a year carry the page’s earned ranking with them. A migration done right is invisible. Done wrong, it hands the record to the scrapers and aggregators already republishing your obituaries — because when your URL dies, theirs is what’s left to quote.

There’s a second audience with a long memory here: the genealogists. FamilySearch’s own research guidance treats the obituary as one of the richest records a family leaves — death date, birth date, parents, spouse, children, occupation — and one partner collection alone indexes over six million of them. Genealogy platforms have also logged, for years, the pattern of funeral homes redesigning websites and watching memorial pages disappear. A century from now, for most of the people you serve, the obituary your home published will be the only narrative record of their life that anyone can still find — if the URL held.

That’s the frame that makes the checklist worth the hour it takes. The obituary was always the town’s record of a life; the machines just made the record legible, liftable, and easy to lose. Build the page that does this work as real text on a permanent address, and every parser and every great-granddaughter that comes looking gets the same answer — yours.