Playbook: the content refresh and decay system

Decayed winners outperform new pages more often than teams expect. The system: detect decay on schedule, triage by recoverable value, refresh with discipline, and verify the recovery.

PlaybookBY THE ASTROFABRIC TEAM · AUG 13, 2026 · 8 MIN READ

Every content library over a year old contains buried treasure: pages that once earned traffic and citations, slid quietly, and would recover with an afternoon's disciplined work - because the URL still carries the accumulated authority a new page starts without. Most teams never collect it, because decay is invisible without detection and refresh work feels like admitting failure. This playbook is the system: scheduled detection, honest diagnosis, value-ranked triage, and refreshes that verify their own results - the maintenance loop of content operations, run as missions.

Why refresh beats net-new

The arithmetic favors refresh for one structural reason: authority accrues to URLs. A page that ranked for two years holds links, engagement history and crawler familiarity a fresh URL earns over quarters - so the same improvement applied to the decayed page starts from its floor, while the new page starts from nothing. The corollary the content brief pipeline should respect: before commissioning net-new on a topic, check whether a decayed page already owns the cluster - the refresh is usually the cheaper win, and the double-page alternative splits the cluster's equity into cannibalizing halves.

Step one: detect decay on schedule

Decay hides because dashboards show totals, not per-page trajectories. The monthly sweep computes, per page: ranking deltas on the page's target cluster, estimated traffic trend, and - the newer column - citation presence in AI answers per the measurement method, since a page can hold rankings while losing the answer seats that matter. Flag pages whose trend crosses thresholds (sustained decline over two-plus periods, not single-month noise), and the sweep's output is the decay register: every declining page with its trajectory attached. This is pure sandbox computation across the library - exactly the mission shape that runs monthly without anyone remembering to run it.

Step two: diagnose the decay type

THE FOUR DECAY TYPES
TypeSignatureTreatment
StalenessFacts aged; fresher pages now rank aboveRe-verify and update substance; honest new modified date
OutcompetitionBetter pages arrived; yours unchangedGap analysis vs. current SERP; match and exceed
Intent shiftThe query now means something elseRealign to current intent, or retire the target
CannibalizationYour own pages splitting the clusterConsolidate: merge, redirect, one canonical page

Diagnosis is a read of the current SERP against the page: who displaced you and why, whether the ranking pages answer a different question than they used to, whether the displacing page is your own. Skipping this step produces the classic wasted refresh - polishing a page whose query moved out from under it.

Step three: triage by recoverable value

The register will exceed capacity, so rank: recoverable value (the traffic and answer presence the page held at peak, priced with current demand) times fixability (staleness refreshes are afternoons; losing a rewritten SERP war is a project). The top of that ranking is almost always former winners with staleness decay - maximum equity, minimum work - which is why the playbook's slogan is "finish the page that is four points short before writing the one that is forty". Cannibalization cases also rank high: consolidation recovers value from both halves at once.

Step four: the refresh itself

Substance, not cosmetics
The refresh that works: re-verify every fact and number against current sources; fill the gaps the current SERP exposes (what the newly-ranking pages cover that yours lacks); retrofit the LLM SEO formats - answer-first sections, tables, real FAQs - since decayed pages usually predate them; update internal links to the cluster's current shape; and set the modified date honestly, because it changed honestly. The refresh that fails: date-bumping with cosmetic edits - detectable, counterproductive, and the reason "refresh" has a bad name in some rooms.

Step five: verify the recovery

Every refresh gets a follow-up read at four and eight weeks: rankings on the cluster, traffic trend, answer presence. Recovered pages close their register entry with the win logged; unrecovered ones get re-diagnosed - the usual finding is a type-two decay treated as type one - and either re-treated or consciously retired. The verification loop is what makes the system compound: diagnosis accuracy improves against its own recorded outcomes, exactly the evidence discipline the rest of the program runs on. The runnable version - sweep, register, triage, refresh drafts staged for review - lives on the SEO and Content shelf.

The working checklist

THE REFRESH SYSTEM, END TO END
0/10 · 0%

Frequently asked questions

When is refreshing better than writing new content?

Whenever a decayed page already holds the cluster: its URL carries accumulated authority a new page starts without, so the same improvement recovers faster. Check the register before commissioning net-new.

How is content decay detected?

A monthly computed sweep per page: ranking deltas on the target cluster, traffic trend, and AI-answer presence. Sustained multi-period declines flag into a register; single-month dips are noise.

What are the types of content decay?

Four: staleness (facts aged), outcompetition (better pages arrived), intent shift (the query changed meaning), and cannibalization (your own pages splitting a cluster). Each wants a different treatment.

Is updating the published date enough?

No - date-bumping without substance is detectable and counterproductive. A real refresh re-verifies facts, fills the gaps the current SERP exposes, retrofits answer-ready formats, and then dates honestly.

How do you know a refresh worked?

Follow-up reads at four and eight weeks on rankings, traffic and answer presence. Log outcomes against diagnoses - misdiagnosed decay (usually outcompetition treated as staleness) is the common failure and the system learns from it.

Sources

⟨ RUN IT INSTEAD OF READING IT ⟩

Every playbook on this blog ships as a runnable mission.

Open a workspace and the playbook library is waiting - describe the outcome and the agents carry it end to end, on your plan's monthly credits.

⟨ KEEP READING ⟩
GuideContent operations

AI content operations: the complete guide

Content as a production system: demand-priced planning, evidence-grounded drafting, QA gates that catch what models get wrong, refresh loops, and distribution that compounds.

Aug 13, 2026 · 12 min read
ArticleContent operations

Content briefs from demand data: encoding the gap

The brief is where content programs win or lose. How to price the opportunity, read the SERP and the coverage, find the falsifiable angle, and write the specification a draft cannot dodge.

Aug 13, 2026 · 8 min read
GuideContent operations

Programmatic SEO with agents: scale without thin content

When one brief becomes five hundred pages: finding real patterns, the data-plus-authored architecture that defeats thinness, the quality floor per page, and the maintenance loop.

Aug 13, 2026 · 8 min read