4 types of content decay and how to fix each one

Every page you people is susceptible to postulation decay. The mobility is whether you drawback it erstwhile it’s down 15% aliases 80%, and whether you fix the correct thing erstwhile you do.

Most teams drawback it late, past scope for the aforesaid instrumentality each time: refresh it. Update the date, adhd a fewer 100 words, republish. Sometimes that works. Often it does nothing. Occasionally, it makes the page worse.

That’s because falling clicks are only a symptom. A page tin suffer postulation for astatine slightest 4 reasons, and each requires a different fix.

The content decay playbook astir of america inherited treats each diminution arsenic the aforesaid problem pinch the aforesaid cure. In 2026, that playbook is missing a cause, and it’s 1 galore teams still overlook.

Here’s really to place the type of decay you’re dealing pinch utilizing information you already have, and what to do astir it.

Content decay isn’t 1 problem

Content decay is simply a sustained nonaccomplishment of integrated clicks and impressions complete time. One-week fluctuations don’t qualify. It isn’t a caller idea. For years, SEOs person explained it successful position of 3 guidelines causes: a competitor improved, hunt intent shifted, aliases request for the taxable declined.

That exemplary is still mostly right. It’s conscionable incomplete because it predates AI Overviews.

In 2026, fewer than 1 successful 3 Google searches sends a click to the unfastened web. About 68% extremity without a click, up from astir 60% 2 years ago. On queries wherever an AI Overview appears, the apical integrated consequence loses around 58% of its clicks. AI Overviews besides appear acold much often connected informational queries than connected commercialized ones, which is precisely the benignant of query astir blogs are built to win.

AI Overviews introduced a caller measurement for pages to suffer traffic. Rankings tin clasp steady, request tin enactment the same, and clicks tin still disappear.

That’s why contented decay is nary longer 1 problem. It’s four.

The 4 types of contented decay

Each type of contented decay leaves a different fingerprint successful your data. Here’s what to look for.

1. Ranking decay

Clicks down, impressions down, mean position worse. A competitor overtook you, the contented went stale, you mislaid links, aliases 2 of your ain pages are cannibalizing each other. This is the classical case, and the only 1 a contented refresh reliably fixes.

2. Zero-click seizure (the caller one)

Clicks down, but impressions are level aliases up, and position is unchangeable aliases better. You’re still ranking, often higher than before, and still losing clicks. 

That’s the fingerprint of an AI Overview, a featured snippet, aliases different SERP characteristic answering the query correct connected the results page. A regular refresh won’t bring the clicks backmost because you didn’t suffer rankings. You mislaid the click to the reply box.

3. Intent drift

Clicks down, position astir holding, but the SERP astir you changed shape. Google reinterpreted the query and now rewards a different format, for illustration video, a comparison table, aliases a merchandise page, and yours nary longer matches. This 1 needs a quality oculus connected the SERP. The numbers unsocial won’t emblem it.

4. Demand decay (the imposter)

Clicks down, impressions down, but your position held aliases moreover improved. You didn’t suffer anything. The taxable is simply searched less. This is the 1 that fools teams into refreshing a page that was ne'er coming back.

Dig deeper: Why much contented is nary longer a reliable measurement to turn SEO

How to show them apart

You don’t request an costly tool. You request Google Search Console and a spreadsheet.

Pull 2 things for each page: monthly integrated clicks for the past six months, which springiness you the trend, and a three-month year-over-year comparison of clicks, impressions, and mean position, which gives you the diagnosis. 

Three months smooths retired the noise, year-over-year comparisons cancel retired seasonality, and the full point fits wrong GSC’s 16-month memory. A six-month year-over-year comparison would require 18 months of history Google doesn’t keep.

Then publication the signature: the telltale measurement clicks, impressions, and mean position move together. Each type of decay has its ain signature. Match your page to the array below.

Clicks	Impressions	Avg. position	Diagnosis
Down	Down	Worse	Ranking decay
Down	Flat aliases up	Stable aliases better	Zero-click capture
Down	Down	Held aliases better	Demand decay
Down	Varies	Holding, but SERP reshaped	Intent drift (confirm connected the SERP)
Down site-wide, date-aligned	Down	Down	Algorithm update (different playbook)

Before you enactment connected immoderate of it, tally 1 cheque first. Did the driblet commencement correct aft you edited the page? If it did, you astir apt didn’t drawback decay. You caused it. Restore the erstwhile type and comparison earlier you heap a refresh connected apical of a self-inflicted wound.

One be aware circumstantial to this moment: Search Console belief information from 2025 is shaky for year-over-year analysis. First, successful September 2025, Google removed the &num=100 parameter, which stripped retired bot-inflated impressions and pulled counts down. 

Months later, Google disclosed thing bigger: a logging correction had been inflating impressions since May 2025. It corrected the rumor without repairing the humanities numbers. Clicks were ne'er affected.

So erstwhile a page looks for illustration request decay, pinch impressions down but position holding aliases improving, don’t spot the belief drop. Open the unrecorded SERP and look. If an AI Overview is sitting there, it’s zero-click seizure wearing a demand-decay costume.

I built a Google Sheets type of this diagnostic. Paste successful six months of clicks positive a year-over-year comparison, and it classifies the decay type and prioritizes pages by recoverable traffic. Use it arsenic a starting point, past validate the results earlier acting.

Want to return it further? GSC and a spreadsheet are the level connected purpose. Anyone tin tally that today, for free. Two upgrades make it moreover sharper:

• Add GA4: Conversions aliases gross per page fto you prioritize by what a page is really worth, not conscionable the clicks it lost. Engagement metrics help, too, because they often dip earlier clicks, giving you an earlier warning.
• Add a rank locator and SERP characteristic data: These fto you corroborate zero-click seizure crossed hundreds of pages astatine erstwhile alternatively of checking 1 consequence astatine a time.

Neither changes the diagnosis. They conscionable measurement it pinch finer instruments.

Dig deeper: How to support your contented caller successful the property of AI

---

The correct hole for each diagnosis

Once you cognize the type, the hole gets specific. Here’s what each 1 looks like.

Ranking decay: The existent refresh

In 2026, that intends information gain, not a caller date. Google’s guidance favors original, first-hand, people-first content, making cosmetic day swaps little useful than meaningful updates. 

Picture a buyer’s guideline that classed 3rd and slid to the bottommost of page 1 aft a competitor published a deeper type and earned links you didn’t. Adding 300 words won’t move it. Focus connected these moves:

• Add what the existent winners screen and you don’t: Original testing, a information constituent only you have, aliases the follow-up mobility each different consequence ignores.
• Check who those winners are first: If a Reddit thread now holds your aged spot, you’re up against a format Google chose to favor, truthful lucifer that intent alternatively of trying to out-write an article.
• Consolidate immoderate cannibalizing URLs into one: Rebuild the soul links that awesome authority.

Zero-click capture: Stop competing pinch the summary

The shape that throws group astir is simply a page that ranks amended and still loses clicks. I audited a high-intent money page that climbed from an mean position of 19 to 11, pinch impressions up 10% and clicks down by half. The reflex is simply a speedy refresh. 

That won’t activity because you didn’t suffer rankings aliases quality. An AI Overview is now answering the query connected the results page. Recovering worth takes a different play. Part of it is still existent activity connected the page.

• Upgrade the page into thing the reply container can’t replace: Original data, a calculator, a tool, aliases a genuine constituent of view. This is activity connected the contented itself, not a date-and-paragraph refresh.
• Make the page a quotable source: That aforesaid unsocial value, positive cleanable structure, gives the page a amended chance of being cited successful an AI Overview aliases utilized by reply engines, wherever ranking connected page 1 nary longer intends you’re the root being quoted.
• Know erstwhile the click is gone, and reallocate your effort: Some of these queries whitethorn ne'er nonstop a click again. Move your effort to a adjacent page person to conversion, for illustration a work aliases comparison page, that group still click.
• Follow the assemblage disconnected Google: If request has moved to Reddit aliases YouTube, meet it location by creating contented for those platforms. That’s a abstracted effort, not an update to this page.

Dig deeper: What replaces the eventual guideline successful AI search

Intent drift: Re-match the format

When Google reinterprets a query, the winning format changes, and a technically beardown page that nary longer fits gets near behind.

A “best [product]” hunt that utilized to aboveground buying guides mightiness now return merchandise pages. A how-to mightiness now lead pinch video.

If you published a communicative guideline wherever the SERP wants a comparison table, contented value is irrelevant. The format is wrong.

• Re-match the format the SERP now rewards, aliases divided the page into the versions it’s asking for.
• Keep the URL truthful you don’t suffer its equity.
• Recheck quarterly, not conscionable once. The SERP is much volatile than it has been successful years, pinch video carousels, forum blocks, and merchandise modules surfacing and disappearing connected the aforesaid query.

Demand decay: Usually, do thing to the content

An aged news article that spiked astir an arena and faded, aliases a guideline for a merchandise cipher buys anymore, isn’t coming back, and nary rewrite resurrects a hunt group stopped running. 

First corroborate the request is really gone, not conscionable relocated. A falling Google Trends statement doesn’t ever mean request disappeared. It whitethorn mean immoderate of that behaviour moved to AI assistants, societal search, forums, aliases video platforms.

• If the taxable still has an assemblage elsewhere, the play is to build a beingness connected that surface, not refresh the page.
• If it’s genuinely gone, consolidate it into a broader page that still has demand, redirect it, aliases prune it.

Prune without guilt. If a page is outdated, unhelpful, and has nary recoverable demand, keeping it whitethorn adhd sound alternatively of value. Google has agelong advised removing aliases improving unhelpful contented erstwhile it nary longer serves users.

Dig deeper: Refreshing content: How to update aged contented to thrust caller traffic

The mistakes that look for illustration strategy

Most wasted refresh effort traces backmost to a fistful of reflexes:

• Treating each driblet arsenic a contented problem. This is the halfway correction the full model exists to prevent.
• Changing the people day pinch thing down it. Readers and Google some spot done it.
• Padding connection count. Length was ne'er the ranking factor. Thoroughness was.
• Refreshing a request problem. No magnitude of editing recovers a taxable group stopped searching for.
• Refreshing excessively often to measurement anything. Give a alteration a 4th to beryllium itself earlier you touch it again.
• Refreshing a page you broke. If postulation fell correct aft an edit, reconstruct it. Don’t rewrite it.

Make it a habit, not a occurrence drill

The teams that enactment up of decay don’t hold for a page to suffer 80%. They tally a quarterly sweep. Tag each slipping page by type, benignant by recoverable postulation and business value, and enactment only wherever the test says action will work.

As zero-click results support climbing, that subject matters moreover more. The advantage nary longer belongs to whoever refreshes the astir pages. It belongs to whoever knows which pages are worthy refreshing and which ones to time off alone.

Key takeaways:

• Diagnose earlier you refresh. Falling clicks person astatine slightest 4 causes, and only 1 is fixed by rewriting. Identify the type first.
• Watch impressions and position, not conscionable clicks. Clicks show you a page is decaying. Impressions and position show you why. Rising impressions pinch falling clicks constituent to zero-click capture, not a contented problem.
• Know erstwhile not to act. Demand decay and “I collapsed it myself” pages don’t request a refresh. They request a different decision. Refreshing them wastes clip and budget.
• Confirm ambiguous cases connected the unrecorded SERP. Especially aft num=100, a demand-decay shape pinch improving position is often an AI Overview successful disguise.
• Systematize it. A quarterly, signature-based decay expanse thumps reactive, one-off rewrites each time.

Dig deeper: How to revise your aged contented for AI hunt optimization