Publish a good page and its traffic climbs, peaks, and then, if nobody touches it, fades. That fade is content decay, and it is not failure. It is the default physics of a results page where everyone else keeps publishing.

The economics are the point: catching decay early turns a rebuild into a refresh. Here is the why, the detection routine and the playbook.

The lifecycle every page lives

climb weeks 0-12 peak and plateau maintained: plateau holds untouched: decay the quarterly check lives here
The curves split where maintenance happens or does not. The check is cheap; the gap between the curves is not.

The four causes

CauseWhat happenedThe tell
StalenessYour specifics aged: prices, screenshots, years, toolsContent reads fine, dates read wrong
OutcompetedSomeone published a better answerNew names above you in the SERP
Intent driftThe query started meaning something elseThe SERP's format changed around you
SERP reshapingFeatures and AI answers absorb the clicksPositions hold, clicks fall anyway

The fourth cause matters more every year: a page can hold position 3 while the results page grows an answer box above it, and the decay shows in CTR rather than rank.

Detection: the quarterly compare

The routine from the reports guide, expanded: in Search Console, Pages tab, compare the last 3 months against the previous 3, sort by click decline. Everything falling more than about 15 percent joins the triage list.

Before treating anyone, rule out the impostors. Seasonal pages sag every off-season, so judge them year over year, never quarter over quarter. Demand collapse, the keyword itself shrinking, shows in impressions falling while position holds. And ordinary volatility never sustains a slope for a full quarter.

page clicks fell 15%+ seasonal dip? YoY says fine → wait demand died? impressions fell → consolidate SERP reshaped? rank holds, CTR fell → adapt outcompeted? new winners above → refresh Only the last two are yours to fix with content. Diagnosing first stops you refreshing pages whose problem is the calendar.
Four checks, two minutes each, before any refresh work starts.

The refresh playbook, matched to cause

Stale: update every specific: numbers, screenshots, examples, the year in the title if it carries one. Substance first, date second, and only then the date.

Outcompeted: read the pages that passed you and out-answer them: fresher data, a missing subtopic, a better artifact (table, checklist, calculator). Match their format where the SERP rewards it.

Intent drift: rebuild the page for the query's new meaning, or accept the loss and retarget the old angle to a query that still means it. Fighting a drifted intent with the old page shape loses slowly and always.

SERP reshaped: chase the format that now owns the clicks: structure answers for the feature box, add the FAQ, and where an answer box genuinely ends the query, shift the page toward the follow-up questions that survive it.

After any refresh, close the loop: re-link the page from your newest relevant content, and note the refresh date in your tracking so next quarter's compare reads clean, alongside the registry habits that stop refreshes colliding with newer pages.

Budgeting maintenance honestly

A useful planning rule: every four to five new pages, one refresh slot. Sites past year two often flip the ratio, because a large back catalog decaying at a few percent a quarter loses more traffic than new pages gain.

Refreshes also pay faster: the page has age, links and history, so recoveries land in weeks where new pages take months.

The one-line takeaway: decay is default physics: stale facts, fresher rivals, drifting intent, reshaping SERPs. Compare quarters, rule out seasonality and dead demand, then refresh to the cause, and give maintenance a standing slot in every content month.