Learn how content decay shows up in the AI search era, how to catch it early with decay velocity, and a triage framework to fix it.
Content decay used to be a quiet problem. A page would slide from position 3 to position 9 over a year, traffic would thin out, and eventually someone would notice the graph trending the wrong way. In 2026, content decay ai search dynamics move a lot faster than that, and the damage isn't limited to rankings anymore. Your page can hold its position and still lose the traffic, because an AI Overview or a ChatGPT answer already gave the searcher what they needed.
That's the part most teams haven't adjusted for yet. Decay isn't just a ranking story now. It's a citation story too.
This guide walks through what content decay actually looks like in 2026, the four things that cause it, how to catch it early with a proper detection system, and a practical triage framework for deciding what to refresh, rewrite, or retire. We'll also get into how Serpely's Content Decay screen automates most of this, because honestly, doing it by hand across a few hundred URLs isn't a great use of anyone's Tuesday.
If you run content, SEO, or growth for a site with more than a handful of published pages, this applies to you whether you've noticed the slide yet or not. Decay is happening somewhere in your archive right now. The only real question is whether you catch it in month one, when a refresh takes an afternoon, or month eight, when you're rebuilding a page from scratch and hoping it recovers.
Content decay is the gradual loss of organic visibility on a page that used to perform well, and it rarely announces itself. Ahrefs describes it as the slow erosion that happens well after a page's initial ranking peak, and that's still true. What's changed is the shape of the loss.
Three years ago, decay mostly meant rank drift. You'd fall from page one to page two and traffic would follow. Today you can hold your ranking position and still watch sessions decline, because Google's AI Overview or a competing LLM answer already resolved the query before the searcher clicked anything.
Search Engine Land's content decay guide frames this well: decay is reversible when you catch it early, but the window to catch it has gotten shorter. A page that decays for six months in the AI search era has usually also stopped showing up in AI-generated answers, which compounds the traffic loss on top of whatever the rankings did.
The practical takeaway is that traffic and rank are now lagging indicators. If you're only watching those two metrics, you're finding out about decay after it's already cost you visibility in two channels instead of one.
Not all decay comes from the same place, and the fix depends entirely on the cause. Here's what's actually driving it in 2026.
This is the newest cause and arguably the most disruptive. When Google's AI Overview answers a query directly, organic clicks on that query collapse, even for pages still ranking well. Seer Interactive's tracking of 2.43 billion impressions across 53 brands found organic CTR on AI Overview queries fell from 1.76% to 0.61% between June 2024 and September 2025, a 65% collapse.
A separate randomized field study reported by Search Engine Journal found a 38% reduction in organic clicks on AI Overview-triggered queries, and Ahrefs has tracked reductions as high as 58% depending on the query type. The range is wide, but the direction isn't in question.
Your page didn't get worse. The search result got shorter. The practical response isn't to abandon the page, it's to make sure your content is structured so an AI system can lift a clean, attributable answer from it, which is a big part of what citation worthiness means in 2026.
Sometimes decay is just someone else doing their job. A competitor refreshes their guide, adds a data table, tightens their answer paragraphs, and Google (or an LLM) starts preferring their version. This is the most common cause and also the easiest to fix once you spot it, because the fix is usually “match or beat what they just did,” not a strategic overhaul.
You'll typically notice this cause by checking who's now ranking or getting cited above you. If it's a page that didn't exist or wasn't competitive six months ago, that's your answer. Pull up their content, see what they added that you don't have (a comparison table, fresher data, a section you skipped), and close the gap.
Core updates and quality-focused algorithm changes can suppress pages that no longer meet the bar, even if nothing on the page itself changed. This one's trickier to diagnose because the trigger is external. You'll usually see it as a sudden step-change in traffic tied to a known update date rather than a gradual slide.
The fix here is less about a single edit and more about a broader quality pass: check for thin sections, verify claims still hold up, and make sure the page genuinely earns its position rather than coasting on age or backlinks alone.
The classic cause, still very real. Outdated statistics, dead internal links, screenshots of a UI that no longer exists, pricing that changed two years ago. LLMs are particularly unforgiving here because they weight recency heavily when deciding what to cite, more so than traditional rankings ever did.
If you want the fuller list of ways pages quietly become invisible to AI search, our breakdown of common SEO mistakes killing AI visibility covers the patterns we see most often across client audits.
Most teams find decay by accident, usually during a quarterly traffic review when it's already too late to catch it cheaply. Detecting it early means watching four signals together, not traffic alone.
• Traffic velocity. Not just “is traffic down,” but the rate of change week over week. A page that's down 8% one week and 14% the next is decaying faster than the raw numbers suggest.
• Rank trajectory. Position over time, not a single snapshot. A page bouncing between 4 and 7 is healthy. A page that's moved from 4 to 6 to 9 to 12 over eight weeks is decaying, even if it's still technically “on page one” this week.
• Citation rate drops. This is the one most teams still aren't tracking. If your page used to get pulled into ChatGPT or Perplexity answers and stopped, that's a decay signal traditional rank trackers will completely miss. Tools built for this matter here. Tracking how often your brand actually gets cited across ChatGPT, Perplexity, Gemini, and AI Overviews needs to be a standing part of your detection workflow, not a one-off audit.
• GEO Score trending down. Serpely's GEO Score blends ranking data, citation frequency, and content freshness signals into a single trend line per page. When it's sloping down for two or more weeks in a row, that's usually your earliest reliable warning, ahead of the traffic drop by a few weeks in most cases we've seen.
For a step-by-step walkthrough of pulling these four signals together into a repeatable check, see our guide to identifying decaying content.
Detection works best paired with a broader content gap analysis, not run in isolation. A content gap analysis usually looks at what you're missing entirely, topics competitors cover that you don't. Content decay detection looks at what you already have that's slipping. Run both and you get a fuller map of where to invest next instead of only chasing whatever's currently on fire.
Most content optimization tools handle one side of this or the other, gaps or decay, rarely both. If you're evaluating a content strategy analysis tool, ask specifically whether it surfaces decaying pages and content gaps in the same view. Stitching together two separate audits by hand every quarter isn't a great use of anyone's time either.
Decay velocity is the rate at which a page is losing visibility, measured week over week rather than as a single point-in-time comparison. It's the single most useful number for triage because it tells you how urgent a page is, not just whether it's declining.
The basic calculation is simple:
Decay Velocity = ((Traffic Week N-1 − Traffic Week N) / Traffic Week N-1) × 100
Run that weekly and you get a trend line, not a snapshot. A page decaying at 2-3% per week is worth watching but not urgent. A page decaying at 10%+ per week needs attention this week, because at that rate it can lose most of its value within a month or two.
The reason weekly cadence matters more now than it used to comes back to AI Overviews. Google can add or remove an AI Overview for a given query in days, not months, and when that happens the traffic impact is immediate. Monthly reporting cycles simply move too slowly to catch it while it's still cheap to fix.
Once you've got decay velocity for every page, the next problem is deciding what to fix first. You can't refresh everything at once, and refreshing your lowest-value pages first is a waste of the team's time.
A Refresh Priority Score solves this by blending three inputs into one rankable number:
• GEO Score trajectory – how fast is AI visibility declining, not just traditional rank
• Traffic decay rate – the decay velocity number from above
• Business value of the page – conversion rate, assisted revenue, or whatever your team uses to weight commercial importance
A page decaying fast but converting poorly might score lower than a page decaying slowly that happens to be your highest-converting landing page. That's the point. The score isn't just “what's falling the hardest,” it's “what's falling the hardest that actually matters to the business.”
Picture two pages. One is a glossary entry losing 12% of its traffic weekly but converts almost nobody. The other is a bottom-of-funnel comparison page losing 4% weekly but drives a third of your demo requests. Raw decay velocity says fix the glossary page first. A proper Refresh Priority Score says fix the comparison page first, because that's where the actual revenue risk sits. This is the gap that trips up teams who triage by traffic drop alone.
Serpely calculates this automatically and feeds the ranked output straight into a prioritization queue, so the question isn't “which of our 400 pages should we look at” but “here are the twelve that matter this week.” If you're building this manually, our framework for deciding which pages to fix first walks through how to weight each input without a tool doing the math for you.
Not every decaying page deserves the same treatment. Throwing a full rewrite at a page that just needs a stat update wastes time, and doing a light refresh on a page whose entire premise is outdated won't move the needle. Here's the framework we use:
Signal | Refresh | Rewrite | Retire |
|---|---|---|---|
Current ranking position | 4–20, still visible | Below 30, or gone from AI answers entirely | Rarely ranks, low search demand |
Core topic relevance | Still accurate and relevant | Search intent has shifted since publish | Topic no longer matters to the business |
Content quality | Solid structure, needs updated facts/data | Weak structure, thin coverage, needs a rebuild | Duplicate or redundant with a stronger page |
Business value | Meaningful traffic or conversions | Potential value if fixed properly | Low or no commercial relevance |
Typical fix | Update stats, add a table, tighten intros, refresh internal links | New outline, expanded coverage, restructured for answer-first format | Redirect or consolidate into a stronger page |
Refresh is the right call for pages that are still fundamentally sound but stale. It's the fastest and cheapest fix, and it's where most of your decaying pages should end up.
Rewrite is for pages where the bones are broken, not just the paint. If intent has genuinely shifted (people searching that term now want something different than they did two years ago), no amount of stat-updating will fix it.
Retire is underused. If a page gets little traffic, isn't cited anywhere, and overlaps heavily with a stronger page on your site, consolidating it (with a proper redirect) often helps the surviving page more than leaving both alive ever did. It's also a form of content seo hygiene that AI crawlers seem to reward, cleaner sites with less redundant content tend to get cited more consistently.
Once you've triaged a page into “refresh,” the next mistake is refreshing the wrong things. Not everything on a decaying page needs to change.
What to update:
• Statistics and data points (this is usually the highest-leverage fix; stale numbers are an easy signal for both users and LLMs to distrust a page)
• Structure, especially adding clear answer-first paragraphs under each H2 if the page was written before that mattered as much as it does now
• Schema markup, particularly FAQ and Article schema, since this is part of what makes a page easy for an AI system to parse and cite
• Internal links, both adding new ones to relevant newer content and fixing any that have gone dead
• Examples and screenshots that reference outdated versions of a product or process
What to leave alone:
• A URL that's already earned backlinks and rankings (don't touch the slug unless you have to)
• Core structure that's working fine, resist the urge to reorganize a page just because you're in there
• Voice and tone that matches the rest of your content, unless the whole site's style has shifted
A good on page seo checklist for refreshes should cover both of these lists explicitly, because teams default to touching everything when they open an old page, and that's often how a simple refresh turns into an unplanned rewrite.
Semrush's content pruning guide makes a similar point about restraint: the goal of a refresh is concentrated improvement, not a full rebuild disguised as an update.
Case in point, an SEO.ai analysis of CNET's publicly available traffic data found organic visits climbed from roughly 19 million to 24.5 million estimated monthly visits in the two months following a large content pruning and refresh pass, a 29% increase, without the site rewriting its entire archive from scratch. Targeted fixes, applied broadly, tend to outperform a handful of full rewrites done slowly.
The quarterly content audit isn't dead, but it's not sufficient on its own anymore. A page can decay meaningfully in the six or eight weeks between audits, especially if an AI Overview gets added to its target query mid-quarter. By the time the next audit catches it, you've lost visibility you can't fully recover.
Continuous monitoring means decay velocity and GEO Score get checked weekly, not quarterly, with alerts firing when a page crosses a threshold rather than waiting for someone to open a spreadsheet. This is also where reporting needs to change. If your monthly report to leadership only shows rank and traffic, it's missing the AI visibility layer entirely, and that's increasingly the layer where the real story is. Building a report that actually shows AI visibility alongside traditional rankings gives stakeholders a much more honest picture of where the site actually stands.
This is exactly what Serpely's Content Decay screen was built to automate. It tracks every page's decay velocity continuously, calculates the Refresh Priority Score by blending GEO Score trajectory, traffic decay rate, and the business value you assign to each page, and attributes each drop to a likely cause (AI Overview cannibalization, competitor movement, algorithmic suppression, or staleness) so you're not guessing why a page fell.
Flagged pages feed directly into the Content Prioritization queue, ranked and ready to work through, instead of living in a spreadsheet someone has to remember to update. For a site with a few dozen pages this is a nice-to-have. For a site with a few hundred or a few thousand, it's the difference between catching decay while it's a quick fix and finding it eight months later as a real revenue problem.
Even with a good detection system in place, execution is where a lot of this falls apart. A few patterns show up repeatedly.
Refreshing on a fixed schedule instead of by urgency. “We refresh the whole blog every January” sounds organized, but it means a page decaying fast in April sits broken for nine months. Priority should follow decay velocity, not a calendar.
Treating every page the same. A glossary entry and a bottom-of-funnel comparison page don't deserve equal attention, even if they're decaying at the same rate. This is exactly what a Refresh Priority Score is meant to prevent.
Updating content without checking why it decayed. Adding a fresh stat to a page that's actually suffering from AI Overview cannibalization won't fix the real problem. Diagnose the cause first, then pick the fix that actually addresses it.
Stopping at the fix. A refresh isn't done when you hit publish. Track the page for the next few weeks to confirm decay velocity actually flattened. If it didn't, you probably triaged it wrong, and it likely needs a rewrite instead of another refresh.
Content decay isn't going away, and in the AI search era it moves faster and hits more channels than it used to. The teams that stay ahead of it aren't the ones with the most content. They're the ones who catch the slide early and know exactly which pages to fix first.
Find your decaying pages automatically with a free trial of Serpely, or book a demo to see how the Content Decay screen and Refresh Priority Score work on your actual site.
Content decay is the gradual loss of organic traffic and rankings on a page that previously performed well, typically caused by outdated information, algorithm changes, competitor updates, or new SERP features like AI Overviews taking the click before a user reaches your page.
Watch four signals together: week-over-week traffic velocity, ranking trajectory over time (not a single snapshot), whether your page is still getting cited in AI answers like ChatGPT or Google's AI Overviews, and any downward trend in a composite score like GEO Score. A drop in any one signal alone can be noise. A drop across two or more is decay.
Quarterly audits catch the big, obvious cases but miss faster-moving decay caused by AI Overview changes, which can happen within days. Weekly or continuous monitoring of decay velocity is increasingly necessary, especially for pages driving meaningful traffic or revenue.
Refresh pages that are still fundamentally accurate but need updated facts, data, or structure. Rewrite pages where the ranking has collapsed below position 30 or search intent has clearly shifted. Retire pages with low traffic, no citations, and heavy overlap with a stronger page on your site, and redirect them instead.
Both, and increasingly the AI search side matters more. A page can hold its Google ranking and still lose most of its traffic if AI Overviews or LLM answers stop citing it. Citation rate is a decay signal traditional rank trackers won't show you.
Both, and increasingly the AI search side matters more. A page can hold its Google ranking and still lose most of its traffic if AI Overviews or LLM answers stop citing it. Citation rate is a decay signal traditional rank trackers won't show you.