The compounding returns of refreshing old content

Every content program has a quiet balance-sheet problem. The team's attention — and the entire editorial calendar — points at new posts, while the actual organic traffic lives in old ones: articles written two years ago, ranking decently, decaying by a percent or two a month as competitors publish fresher answers and the SERP drifts. Left alone, a content library is an asset depreciating on schedule. Maintained, it's the closest thing content marketing has to compound interest.

We've run refresh programs for clients across media, SaaS and e-commerce, and the pattern holds: a disciplined refresh cadence returns more traffic per hour than net-new publishing, typically by a wide margin, because you're renovating ranked assets instead of betting on unproven ones. Here's the system.

Step 1: Detect decay before it's fatal

The data source is Google Search Console, compared across time. Our monthly decay report flags any URL where clicks or average position have declined over a trailing three-month window versus the prior three months — not week-over-week, which is weather; quarter-over-quarter, which is climate. Segment the flagged URLs into three buckets:

• Rank decay — impressions stable, position slipping. A competitor published something better or fresher. Highest refresh ROI: the page still has authority; it's losing relevance.
• Impression decay — the query universe itself is shrinking (the topic is cooling, or SERP features ate the clicks). Refresh may not help; sometimes the honest answer is consolidation or retirement.
• Click-through decay — positions held, CTR dropped. The problem is the snippet: title and description lost a fight against a livelier SERP. Cheapest fix in the program: rewrite titles and metas, no body edits.

One honest caveat, in the spirit of attribution that admits what it doesn't know: month-level Search Console data is noisy. We don't refresh on a single bad month; we refresh on a trend.

Prioritise within the decay list by (current traffic × conversion value of the topic) ÷ (estimated refresh effort). An article ranking #4 for a buying-intent query beats a #2 ranking for a curiosity query every time. Traffic without intent is decoration — the same argument we make in funnel metrics.

Step 2: Refresh or rewrite — the decision rules

Not every decaying page deserves the same surgery. We triage into four outcomes:

Refresh (estimate: 2–4 hours). The core argument is still right and still ours. Work: update examples and dates, replace dead links and stale screenshots, tighten the intro against the current intent, add a section covering the subtopic that emerged since publication (visible in the queries the page now earns impressions for), and improve internal links to newer related pieces. This is 70–80% of the program.

Rewrite (a day). The ranking is decent but the piece was written for a different intent than the query now carries — the SERP has redefined the question underneath you. Keep the URL, keep what's earned, restructure around the current intent. A rewrite with the same URL preserves authority; a new URL spends it.

Consolidate (half a day). Three posts each ranking #11–20 for variants of the same query are three weak answers where one strong one would rank. Merge into the strongest URL, 301 the others, and fold their unique material in. Consolidation is the refresh program's most underused move — content libraries accumulate overlapping posts the way codebases accumulate dead feature flags, and for the same reason: addition is easy, deletion requires a meeting.

Retire. Some pages can't be saved and shouldn't be: outdated advice in a YMYL-adjacent space, announcements with no residual value. Noindex or remove with a clear conscience. A library that keeps everything eventually ranks for nothing.

Step 3: The refresh checklist

When a URL goes in for refresh, the pass is the same every time:

1. Re-check the SERP first. What's ranking now, what's the dominant intent, which SERP features (video, forums, AI overviews) changed the click economics? Refreshing without re-checking intent is repainting a house whose street moved.
2. Update what's dated — examples, statistics, screenshots, tools mentioned, version numbers.
3. Expand into earned impressions. Search Console shows queries the page surfaces for but doesn't answer. Missing subtopics the SERP already associates with your URL are the cheapest expansion you'll ever write.
4. Rework the intro. Two sentences, the reader's situation, the promise. Intros written before you had data are always vague; you now know exactly who arrives and why.
5. Strengthen internal links both ways — from the refreshed page to your newer relevant pieces, and from high-authority pages to the refreshed one (clusters make this systematic).
6. Rewrite title and description against the current SERP. It earns the click the rank deserves.
7. Update the displayed date honestly. Republish with the true revision date and, on substantial rewrites, an editor's note. Fake freshness is detectable, penalised by readers before any algorithm.
8. Re-index through Search Console — request indexing, and let the updated date do its work in the snippet.

Step 4: Cadence and capacity

The compounding comes from rhythm, not heroics. Our standard cadence for a mid-size library (100–300 posts): one refresh day per week, forever. Four to six URLs a week, chosen from the decay queue by the priority formula. That rhythm refreshes the top quartile of a library roughly twice a year — fast enough to hold position against normal competitive churn.

Program hygiene matters more than volume. We track refresh history per URL (a refreshed-twice page that's still decaying is telling you the topic, not the article, is the problem), and we measure at the portfolio level: refreshed-URL cohort clicks vs the un-touched control. Across client programs the refreshed cohort consistently outperforms — the effect shows up at four to eight weeks and persists. (Those numbers are illustrative of pattern, not a promise; any agency quoting you exact refresh-uplift percentages is rounding up its own mythology.)

Governance, because every refresh program is really an editorial calendar problem: the decay queue lives in the same tracker as new content, refreshes count as shipped work (a refresh that recovers a declining page is worth more than a new post that ranks for nothing), and ownership is named. "The whole team owns the archive" means the archive owns itself. Nobody owns it.

Key takeaways

• Segment decay by symptom: rank decay (refresh), impression decay (consolidate or retire), CTR decay (rewrite the snippet). Same report, three different surgeries.
• Refresh beats rewrite for most URLs: update the dated material, expand into earned impressions, rework intro and title, strengthen internal links both directions.
• Consolidate overlapping mid-ranking posts into one strong URL with 301s; retire what can't be saved. Libraries die of accumulation.
• Display dates honestly and request reindexing; fake freshness erodes reader trust before any penalty does.
• Compounding comes from cadence: one refresh day a week, tracked at cohort level, governed inside the same editorial system as new content.

FAQ

How soon after publishing should a post enter the refresh queue? Twelve months at the earliest for most topics — before that, a post hasn't finished earning its position and "refreshing" is just rewriting on impulse. The exception is fast-moving topics (software versions, regulation), which enter the queue on the industry's calendar, not ours.

Do we change the URL when the topic evolves? Almost never; the URL is where the authority lives. If the slug genuinely misdescribes the content after a rewrite, change it once, 301 permanently, and accept a wobble while the redirect settles. Changing slugs for aesthetic reasons is how sites lose a year of equity in an afternoon.

Should we delete underperforming posts in bulk? Bulk deletion is usually panic, not strategy. Run the triage: many "underperformers" are one consolidation away from being the strong half of a merged URL. Delete deliberately — genuinely harmful content, zero-traffic pages with zero link equity and no strategic role — and document what you removed and why.

How do we refresh when the original author left? Treat it as editing, not ghost-writing: the refresh owner verifies claims, updates examples, and adds their name as editor where the changes are substantial. Institutional knowledge of why the piece says what it says matters less than whether it's still right.

What's the single highest-ROI refresh action? The title-and-meta rewrite on a CTR-decayed page that holds its rank. Ten minutes of work, no body edits, and it recovers clicks the page was already earning impressions for. It is embarrassingly effective, which is why it heads every checklist we run.