Server rendering and the crawlers that do not wait
Crawlers operated by the large model vendors fetch the raw response and stop there, so text injected after hydration never reaches them.
Every story below is summarised in its own article element. There is no main element on this page, which is how a great many real homepages are built.
Crawlers operated by the large model vendors fetch the raw response and stop there, so text injected after hydration never reaches them.
Running the suite against the merged state rather than the branch catches the twelve percent of failures that only appear after integration.
A pricing page that lists numbers without naming what changes between tiers leaves the reader to reconstruct the comparison themselves.
It is the only part of a page an engine does not have to interpret, which is what makes it the cheapest way to be understood correctly.