Content Decay: Detecting Declining Pages and the Refresh Signals That Matter

Pages don't fail all at once. They slip, a position here, a lost featured snippet there, a competitor refresh that quietly siphons clicks, until a page that once drove steady traffic is a ghost of itself. The trick is catching that slide while it's still a slope and not a cliff. This is a monitoring system built on Google Search Console trend deltas that flags decaying pages early and tells you what to do about each one.

What content decay actually is (and what it isn't)

Content decay is the gradual erosion of organic performance on a page that previously performed well. The key word is previously. A page that never ranked isn't decaying, it's just underperforming, which is a different problem with different fixes. Decay implies you had something and are losing it.

It shows up in three distinct patterns, and conflating them leads to wrong decisions:

• Ranking decay, average position worsens while impressions hold roughly steady. The query demand is intact; your page is losing the SERP competition.
• Demand decay, impressions fall while position holds. The topic itself is cooling (seasonal, a fad fading, a deprecated technology). Refreshing the page won't bring back searches that aren't happening.
• CTR decay, impressions and position are stable but clicks drop. Usually a SERP feature change: an AI Overview, a new People Also Ask block, or a competitor with a more compelling title now sits above the fold.

You cannot tell these apart from a clicks line alone. You need the full quartet of clicks, impressions, position, and CTR, compared across time.

The monitoring approach: GSC trend deltas

The detection engine is a period-over-period comparison at the page level. Pull two equal, recent windows and compute the delta for each metric. A 28-day window compared against the prior 28 days smooths out weekly noise; a 3-month-vs-prior-3-month view catches slower bleeds that a 28-day window misses.

Run a query grouped by page, dimension page, for both windows, then join on URL and compute:

• clicks_delta_pct = (clicks_now - clicks_prev) / clicks_prev
• impressions_delta_pct, same formula on impressions
• position_delta = position_now - position_prev (positive means worse, since lower position is better)
• ctr_delta = ctr_now - ctr_prev

One critical guardrail: compare like windows against like windows. Year-over-year is the gold standard for anything seasonal, because last month vs. this month will scream "decay" every January for a business that's simply post-holiday. If you only have one comparison to make, make it the same period last year.

Then apply a materiality floor before you alert on anything. A page that dropped from 4 clicks to 1 is a 75% decline that means nothing. Require an absolute threshold, say, the page must have had at least 20, 30 clicks in the prior window, so you're triaging pages that actually move the business. A useful decay flag is the combination: clicks_delta_pct < -0.25 AND prior_clicks >= 25 AND position_delta > 0.5. That isolates real, ranking-driven losses on pages that matter.

Reading the signals: diagnose before you touch

Once a page trips the flag, the metric pattern tells you the cause. This is the diagnostic table I run in my head:

• Clicks down, impressions down, position worse → classic ranking decay. You're being outcompeted. This is the prime refresh candidate.
• Clicks down, impressions flat, position flat, CTR down → a SERP-layout change. Don't rewrite the body; rework the title tag and meta description, add or restructure for the snippet/PAA, and consider schema. The content is fine; the packaging lost the click.
• Clicks down, impressions down, position flat or better → demand decay. The topic is shrinking. Refreshing won't manufacture searches. Decide between consolidation or pruning, not a rewrite.
• Clicks down but a sibling page is climbing on the same queries → keyword cannibalization. Two of your URLs are splitting and diluting authority. The fix is consolidation, not refresh.

Always cross-reference the query view for the decaying page (get_search_by_page_query or equivalent). If the page lost a specific high-value head term, that's a targeted content gap. If it lost a long tail of dozens of minor queries evenly, that's broad relevance erosion or a sitewide signal, and refreshing one page won't fix a sitewide problem.

Refresh, consolidate, or prune, the decision

Detection is the easy half. The discipline is matching the page to the right intervention.

1. Refresh when the page still targets a live, valuable query set and lost ground to fresher or deeper competitors. The query has demand, you used to win it, and a rival simply did the work more recently. Update facts, statistics, and dates; expand thin sections that competitors now cover; refresh examples and screenshots; strengthen internal links into the page from your strongest related content; and re-examine whether the page still matches current search intent. Then resubmit via URL inspection.
2. Consolidate when two or more pages compete for the same intent, or when several thin pages would be stronger as one authoritative resource. Pick the URL with the most existing equity (links, age, historical ranking) as the canonical target, merge the unique value from the others into it, and 301 the losers to the winner. Consolidation is also the right call for cannibalization that a refresh can't solve.
3. Prune when demand is genuinely gone, the page has no inbound links worth preserving, and the topic no longer fits your strategy. Pruning means removing low-quality, dead-weight URLs so they stop diluting your site's overall quality signals. Either 410/remove it, 301 it to the nearest relevant page, or noindex it if it still serves users but shouldn't compete in search.

A blunt but honest heuristic: if the topic still matters and you'd write the page today, refresh. If you'd write a page but not this one, consolidate. If you wouldn't write it at all, prune.

Operationalizing it

This only works if it runs on a schedule, not when someone remembers to look. Set a monthly cadence: pull the 28-day and prior-28-day page-level deltas, filter to the materiality floor, and produce a ranked list of decaying pages sorted by absolute clicks lost, not percentage. Absolute clicks lost prioritizes the pages that actually cost you traffic.

For each flagged page, attach the query-level view and the diagnostic pattern so the person acting on it isn't starting from zero. Track every intervention with a date so you can measure recovery: tag the URL, note the action and the date, and re-check the same deltas 4, 8 weeks later. Reindexing and re-ranking take time, so resist judging a refresh after two weeks.

Common mistakes

• Alerting on percentages without an absolute floor. Tiny pages produce huge percentage swings and bury the signals that matter.
• Comparing mismatched windows. Month-over-month on seasonal content invents decay that doesn't exist. Use year-over-year for anything cyclical.
• Treating every decline as a refresh job. Demand decay and cannibalization don't respond to rewrites. You'll burn hours editing a page whose problem is structural.
• Ignoring CTR-only decay. A title/snippet fix is the cheapest, fastest win in the whole system and is invisible if you only watch clicks.
• Updating the visible date without changing the content. Search engines evaluate the actual content delta, not the timestamp. A cosmetic date bump is not a refresh.
• Refreshing and walking away. Without a tagged date and a follow-up check, you never learn which interventions worked, so you never get better at the next decision.

The system is deliberately simple: equal-window deltas, a materiality floor, a four-metric diagnosis, and a three-way decision. Run it monthly and decay stops being the thing you discover six months too late and becomes a routine, low-drama maintenance task.