There is a bias built into every editorial operation: publishing feels like progress and maintenance feels like overhead. The data says the opposite. A page that has ranked for two years has accumulated history, links and behavioural trust — an asset that a light refresh reactivates at a fraction of the cost of building a new page from nothing. The discipline of content refresh is the systematic harvesting of that asset.
Decay is normal, and diagnosable
Pages decline for identifiable reasons, and the first skill is telling them apart. Query drift: the intent behind the query moved — the results page now fills with a different format — per our intent guide. Fact ageing: prices, versions, statistics, the very benchmarks on our costs page — all rot on schedule. Competition: rivals updated, you didn’t. Consolidation: three of your pages now compete for one cluster and split the signals. Each diagnosis has a different cure, which is why the audit precedes the work.
The audit: classify everything
Pull every page with its sessions trend, its head rankings, and its age. Then sort into four buckets. Performers — ranking, traipsing along: refresh on schedule, change little. Decayers — were ranking, now sliding: the priority queue; diagnose which decay cause applies. Underperformers — never ranked, no links, no clear intent match: candidates for merge or prune. Sleepers — modest traffic, positions five to fifteen, recent drift: the highest-leverage bucket on the list, because a nudge can move them onto page one.
The update–merge–prune decision
- Update when the page’s premise is right and its facts, examples or completeness have aged. Keep the URL, keep the proven ranking, renew the substance.
- Merge when two of your pages answer one cluster. Fold the weaker into the stronger — redirect, absorb the good paragraphs, and consolidate every signal onto one URL. The clustering method tells you which pairs qualify.
- Prune when nothing is left to defend: no traffic, no links, no intent, no purpose. Redirect if it has any equity at all; otherwise let it go. Deletion is maintenance too — a site is judged by its worst pages as much as its best.
The refresh workflow
- Re-search the query. Read today’s results page — has the expected shape changed? Fix format first, prose second.
- Renew the perishables. Figures, versions, screenshots-in-words, examples. Date the update honestly.
- Close the gaps. Add the subsections the new intent demands; cut the ones nothing demands.
- Sharpen the job. Re-run the one-sentence job test from the intent guide. Pages drift away from their own premise.
- Keep the URL and its history. Refresh is not republishing. The address is part of the asset.
Publishing is buying traffic at full price. Refreshing is buying it back at a discount, from yourself.
The over-refreshing tell
The practice has a failure mode of its own, and it is worth naming. A refresh that changes dates and shuffles adjectives teaches the engine to expect churn without substance — and teaches readers that returning was unnecessary. The tells of over-refreshing: changelogs that list edits but no new information, titles that change faster than their rankings, and update stamps that survive untouched content for years. The rule of thumb this desk uses: if nothing in the page would surprise a reader who saw the previous version, it was not a refresh; it was a ritual.
Cadence and the compounding effect
The standing rhythm is part of the weekly schedule described in traffic-building activities: a slice of every month’s production time goes to the decayers and sleepers, not to new pages. The effect compounds twice — refreshed pages recover traffic immediately, and the audit keeps teaching you which publication patterns decay fastest, which quietly improves everything you publish next. One caution to keep the practice honest: a refresh must change what the reader gets, not just the date on the page. Search engines measure the difference, and so do readers.