Structured data has to be generated, not written by hand

Most guidance about structured data ends where the real work begins. Add the JSON-LD, validate it once, tick the box. But Google's requirement is not that the markup exists — it is that the markup is still true about a page people keep editing. That is a very different job, and it is why we generate the machine-readable layer instead of maintaining it.

Google asks for agreement, not for markup

Google's general structured data guidelines are unusually direct about the standard. Under relevance, your structured data must be a true representation of the page content. Under content, you are told not to mark up anything that is not visible to readers of the page, and to provide up-to-date information.

Read those together and the bar is not the presence of a script tag. It is agreement between two descriptions of one thing: the page a person reads, and the record a machine reads. Passing a validator proves the second is well-formed. It says nothing about whether the two still match.

That distinction changes what "done" means. Markup you added correctly in March is not finished work. It is a standing claim about the page that has to keep being accurate in September.

The downside is narrower than most people are told

The stakes are worth knowing precisely, because this field attracts overstatement. Google's documentation says a structured data manual action means the page loses eligibility for appearance as a rich result, and that it does not affect how the page ranks in web search. The same page notes that structured data enables a feature to be present but does not guarantee it will be.

So this is not a ranking lever. Anyone selling it as one is selling something else. What it buys is eligibility and legibility — the chance of a richer result, and a clean description of your page for anything that parses rather than reads.

It is the same argument we made about why Google doesn't penalize AI content. The ranking systems are aimed at whether a page was worth showing. The plumbing decides how legibly the page presents itself, not whether it deserved to be shown at all.

Hand-maintained markup drifts, and nothing tells you

Here is how it actually breaks. Never sabotage; always ordinary editing.

Someone rewrites a headline and does not touch the headline property. A writer leaves and their name stays in the author field of forty posts. A piece is substantially rewritten and dateModified never moves. Occasionally the reverse happens and datePublished is nudged forward so an old post looks fresh — the one version of this that is deliberate, and the one Google's guidelines describe as misleading rather than merely stale.

None of it throws an error. That is the entire problem. A broken link returns a 404 and somebody eventually files it. Wrong structured data renders perfectly, validates cleanly if the shape is right, and stays wrong for years. The page and its machine-readable twin quietly stop being the same page, and your first signal is a rich result you used to have and no longer do.

Feeds and sitemaps fail the same way for the same reason. A hand-curated RSS file is a second copy of your archive, and second copies disagree with the first.

Generate it from the same field the page reads

The fix is architectural rather than editorial, which is exactly why it holds.

If the h1 on the page and the headline in the markup read the same stored field, they cannot disagree. If dateModified is written by whatever saves the post rather than by whoever remembers, it is right by construction. If the sitemap and the feed are queries against the archive rather than files someone maintains, they cannot fall behind it. Correctness stops being a discipline problem and becomes a property of the system, which is the only kind of correctness that survives a busy quarter.

This is also the honest reason the layer is cheap for us and expensive for most teams. It is not that we care more. A generated artefact costs nothing to keep true, while a hand-kept one levies a small tax on every edit anyone makes, forever — and that tax stops being paid somewhere around month four.

Two smaller things from the same guidelines are worth doing once and then forgetting: put the structured data on the page it describes, and do not block those pages from crawlers with robots.txt or noindex. Both fall out of generation for free and are easy to get wrong by hand.

What the layer is worth, stated plainly

It will not rescue a page nobody needed, and it will not move you up a results page. What it does is make your archive legible to the things that never read it — answer engines deciding what to cite, aggregators deciding what is new, crawlers deciding what changed.

Do it, then stop thinking about it. The test of whether you did it properly is not that the markup is correct today. It is that nobody has to remember it tomorrow.