Which tool to choose
A general-purpose AI assistant in its most recent version does all the rewriting work. For texts where it matters that the information is up to date (dates, data, industry news) use a tool connected to the web, so it doesn't put back facts as old as the text you're freshening up. For the rest, copy-pasting the text into the AI is more than enough.
How to do it
- Find the texts worth it: the ones still read, the ones on topics that bring you customers, the ones with information that's now wrong.
- Paste the old text into the AI saying what's wrong with it: outdated information, too formal or dated tone, sentences you wouldn't write today.
- Ask for an update that saves the good parts and freshens up the rest.
- Reread the facts: the AI may have left in (or reintroduced) information that's no longer true.
The operational syntax:
I'm pasting an article from my site from a few years ago. Update it without rewriting it from scratch: keep the structure and the parts still valid, modernize the tone where it sounds dated, flag any information that might be outdated today and propose how to correct it. Don't add new facts you can't verify: where an updated figure is needed, write "[verify]" in brackets. Preserve my way of writing, don't make it more formal.
Follow-through: compare the before and after looking for the points where the AI changed the meaning, not just the words. If it "improved" a piece of information by inventing it, you've caught it: ask "I didn't write that figure: where does it come from? Remove it if it's not verified."
A concrete example
Roberto has a cooking blog and one of his 2021 articles on "the best air fryers" was still being visited, but it mentioned models that had gone out of production. He gives it to the AI asking it to update it: the AI keeps the structure and the buying advice (still valid), flags the models to review and modernizes the tone. Roberto checks the current products, replaces them, republishes. The article starts climbing again, because it already had years of accumulated "trust" that a new article wouldn't have had.
When it does NOT work (and how to fix it)
If the AI overhauls the text instead of updating it
You gave it too much freedom. Constrain it: "Change only the sentences I point out, leave the rest identical". Proceed in pieces, paragraph by paragraph, on the texts you care about. Surgical updating beats total rewriting when the old text already worked.
If you don't know which old texts are worth freshening up
Start from two categories: the ones still read (you see them in the site's statistics) and the ones with information that's now false, which makes you look bad. Ignore the texts that no one reads and that won't bring customers: updating them is wasted effort. Ask the AI to help you decide only after you've looked at the real traffic data.
If after the update the text loses your voice
The AI tends to flatten things toward a neutral and somewhat cold tone. Give it an example of your style ("I write like this: [a paragraph of yours you like]") and ask it to respect it. If you don't tell it, it gives you back its voice, not yours.
A tip from someone who actually uses it
Updating the old often pays off more than the new, and almost no one does it. Everyone rushes to write the next article and lets thirty old pieces rot that, with an hour of work, would go back to bringing visits. Before writing something new, check whether you don't already have a text on the topic to freshen up: the work is half and the result often double.
Frequently asked questions
Update an article or write a new one on the same topic?
Update, almost always. A new article on the same topic competes with the old one and splits the visits; updating the existing one concentrates all the value on a single page. Create a new article only if the angle is really different, not if it's the same topic redone.
How often should I review old texts?
One pass a year over the texts that matter is enough for most businesses. Mark a day on the calendar for the "check-up" and have the AI update the ones with information that ages (prices, regulations, products). The texts with no expiry, like a guide of concepts, hold up much longer.
Is updating the date of an old article an unfair trick?
It is if you change only the date without touching the content: the search engines understand it and don't reward it. It becomes fair and useful if you really update the information and then reflect the new date. The trick isn't the date: it's actually updating the text and letting the reader know.