Any site that has been publishing for a few years is carrying pages that nearly work. They rank at the bottom of page one or the top of page two, collect impressions, and convert nobody because almost nobody sees them.
Improving those is faster and more certain than starting a new subject, because the hard part — being considered relevant at all — has already happened.
Sort the archive into four piles
Pile one: refresh
Pages ranking between roughly position five and twenty for a term you want. The page is close. It usually needs the introduction sharpened, two sub-questions added that competitors answer and it does not, updated specifics, and better internal links pointing at it.
Keep the URL. Keep whatever is working. This is editing, not replacement.
Pile two: rewrite
Pages on a subject you want to own where the existing article is genuinely poor — thin, outdated in substance, or written for a different business than the one you are now.
Same URL, new page. The address keeps its history; the content earns its keep again.
Pile three: merge
Three or four pages circling the same question. Pick the strongest URL, fold the best material from the others into it, redirect them, and update every internal link that pointed at the old ones.
This is the highest-value pile on most sites, and the one teams avoid because deleting feels like destroying work.
Pile four: leave or remove
Pages with no impressions, no links, no relevance to what you sell now. Leaving them costs little; removing them tidies the crawl. What matters is not spending money improving them.
How to decide quickly
Export the last twelve months of Search Console data by page. For each page you care about, three numbers decide the pile: impressions, average position, and whether it relates to something you actually sell.
- High impressions, position 5–20, commercially relevant → refresh first.
- Some impressions, position beyond 20, commercially relevant → rewrite.
- Overlapping with other pages on the same query → merge.
- No impressions, no links, no commercial relevance → leave it alone.
What refreshing is not
It is not changing the year in the title and republishing. Search engines compare the substance of a page against its competitors, not its timestamp — and readers notice immediately when nothing else changed.
A refresh that does not answer at least one question the page previously ignored is not a refresh.
Worth remembering
- Position five to twenty is the cheapest traffic on your site.
- Refresh keeps the URL; merging consolidates competing pages onto the strongest one.
- Never delete a page carrying external links — redirect it.
- Changing the date and republishing is not a refresh.
