Content Freshness

Content freshness is simply whether a page says when it was written or last changed — and says it in both of the places that matter. One is visible: a line a reader can see. The other is a date label inside the page's machine-readable business card, which no visitor ever sees. Same fact, written twice, because two different kinds of reader look in two different places.

This mattered less when the job was ranking pages. It matters a great deal now, because most of what people ask an assistant is implicitly a question about the present. What does it cost. Are they open. Do they still do this. Faced with two sources that disagree, a system reaching for an answer leans toward the one it can date — and an undated page is the easier of the two to pass over.

There are three ways to get this wrong, and only one of them looks like neglect. The first is no date anywhere. The second is a date that contradicts you — a copyright year three years behind, which reads less like an oversight than like a business that may have quietly closed. The third is the opposite failure and the one people talk themselves into: a date bumped automatically every night on a page that has not changed since spring. That is a claim you cannot back, and a machine comparing the date against the words is well placed to notice.

What good looks like is narrower than it sounds. Date the pages that actually change — prices, hours, availability, anything a customer would want to be current — and put that date where a person can read it, not only in the code. Leave genuinely evergreen pages alone rather than dressing them up. And move the date when the content moves, which is the whole of the discipline: the date is not decoration, it is a statement about the words beneath it.

The plain version: a price list taped to a wall with no date on it. You would read it, wonder whether it still held, and ask someone. A machine cannot ask. It either trusts the page or reaches for one that answered the question the page left open.