Content Refresh Case Study

Content refreshing is the audit playbook's favourite prescription, so here's the prescription tested at portfolio scale: fifteen decaying posts on one site, systematically refreshed over one month, tracked for four months against fifteen matched decliners left alone. The result validates the tactic emphatically — with a distribution lesson (which refreshes paid and which didn't) that changes how the work should be triaged.

Selecting the patients

All thirty pages shared the decay signature: ranked page one historically, slid to positions 6–20 over the prior year, traffic down 40%+ from peak — the classic "was winning, stopped" profile, which matters because refreshing never-ranked content is a different (and worse) bet. Diagnosis per page before touching anything: why did it decay? Competitor coverage passed it (9 of the 15 treated), information aged visibly — dead examples, old data, stale screenshots (4), or the query's intent had shifted under it (2, flagged as the hard cases going in).

The treatment protocol

Per post, matched to diagnosis, ~4 hours average: coverage rebuilt against the current SERP (new sections for the subtopics winners now carry — the statistical read as gap-finder, judgment as editor), facts and examples re-verified and re-dated, internal links added both directions (each refresh linked from two newer relevant posts — the experiment's lever bundled in deliberately), titles/metas rewritten where CTR lagged position, FAQ blocks added where question-queries clustered, and the update date honestly changed (content genuinely updated — the date-bump-alone myth stayed untested on principle). Republished at original URLs, recrawl requested.

The results at four months

1. Treated: average position 11.2 → 6.1; controls: 11.5 → 12.4 (decay, untreated, continues — the control group's quiet lesson).
2. Traffic to treated pages: +82% aggregate, concentrated in the eight pages that re-crossed onto page one.
3. The distribution — the real finding: the 9 coverage-decay pages delivered nearly all the gain (average +6.3 positions; several full recoveries to former peaks); the 4 aged-info pages gained modestly (+2.5 — freshness helped, but their queries hadn't grown, so the ceiling was the old ceiling); the 2 intent-shift pages: one recovered after effectively being rewritten for the new intent (that's not a refresh, that's a replacement at the same URL), one didn't move — its query now rewards a format (tools/comparison) the page can't be.
4. Timeline: movement from week 3, most gains landed by week 10 — slower than the title test's CTR effects, faster than external links, consistent with re-evaluation-on-recrawl mechanics.

The triage rule this produces

Refresh ROI follows the diagnosis, so diagnose before scheduling: coverage-passed pages are the gold tier (proven query, proven page, fixable gap — schedule them first, always); aged-info pages are maintenance (worth the smaller hours for the smaller, real returns — batch them); intent-shifted pages need the honest fork (rebuild-as-new-format or merge/retire — a "refresh" line in the calendar underestimates them every time). And the meta-lesson the controls paid to teach: decay untreated compounds — the do-nothing group lost another position on average while the test ran, which prices procrastination properly in the maintenance-slot argument.

Frequently asked questions

Does changing the visible date matter?

With genuine updates behind it, the honest current date belongs on the page (and in dateModified) — readers and SERP display both reward visible currency. Date-bumping unchanged content is the version that earns nothing and teaches Google your dates lie, per the lastmod-honesty rule.

Refresh or write new — how do I split the calendar?

The audit decides annually, but the experiment's arithmetic generalises: 4 hours refreshing a coverage-decayed page returned more traffic than 4 hours of new content on all but our best new topics — because age, links and history are pre-paid. The standing one-in-five maintenance slot is a floor, not a ceiling, on sites with a back catalogue.

Why did recovered pages hold their gains this time?

Because the decay cause was addressed, not cosmetically painted — and because the refreshes entered the maintenance cycle (quarterly trend checks, annual re-review, per the contract). Pages decay again when competitors move again; the moat against that is the pairing this whole series ends on — maintained content plus accumulating authority (our half).