Check this page with an assistantOpens a chat asking it to summarise this article and name the evidence behind each claim.
Claude opens with the prompt on your clipboard: Anthropic does not support prefilled prompts on the web, and we would rather copy it than ship a button that drops it.
Why Do Blog Posts Lose Traffic Over Time?
Content decay has four ordinary causes: newer competing pages outrank yours, the facts and years in the post go stale, the intent behind the query drifts away from what the page answers, and your own newer posts cannibalize the old one. Nothing mystical is happening. Each cause has a distinct fix, which is why diagnosis comes before edits.
The four causes leave different fingerprints. A competitor publishing a deeper guide erodes your position slowly across months. Stale facts hit harder at year boundaries, when a post titled for 2024 starts losing clicks to anything titled for 2026. Intent drift shows up as steady rankings but falling clicks, the searchers still see you and no longer want what the snippet promises. And cannibalization appears the day you publish a neighboring post: Google now has two of your URLs answering one query and splits signals between them, often ranking both worse than either would alone.
Treating these as one problem produces the classic wasted refresh: rewriting a post whose real issue was a cannibal sibling, or adding 800 words to a page that only needed its title updated. The data to tell them apart is already sitting in Search Console.
Finding Decay in Search Console Before It Gets Expensive
The method is one comparison: in Search Console's Performance report, compare the last 3 months against the same 3 months one year prior. Same-season comparison cancels out seasonality, so any decline you see is real decay, not a summer lull. From that view, two filters surface the highest-return work.
First filter: pages whose clicks fell while impressions held or rose. Google is still showing the page; searchers have stopped choosing it. That is a title and snippet problem, a CTR contest the page is losing, and it is the cheapest fix in SEO because it requires rewriting one line, not one article.
Second filter: switch to the Queries tab and list every query where the page sits at positions 5 through 15. This is striking distance. A move from position 12 to position 5 can multiply clicks several times over, while a move from 45 to 38 changes nothing, so these queries offer the biggest gain per hour of editing. Give each striking-distance query its own dedicated section in the post and point internal links at the page using that query's phrasing.
While you are in the Queries tab, check whether two of your pages rank for the same query. If they trade places week to week, that is cannibalization, and the fix is a merge, not two parallel refreshes. This whole analysis takes an hour or two by hand each quarter; the Growth Map runs this exact comparison automatically and returns the same output, a ranked queue of pages to refresh, merge, or leave alone.
What to Change in a Refresh, and What to Leave Alone
Change five things: stale years, prices, and screenshots; subtopics competitors now cover that you do not; the title, when CTR is the diagnosis; a sharper direct answer at the top of the post; and internal links out to newer related pages. Leave three things alone: the URL, headings that already rank, and any section earning a featured snippet.
The URL rule is absolute. A ranking URL holds age, backlinks, and query history, and changing it, even with a clean 301, puts all of that through a lossy transfer for zero benefit. Slug looks dated? Leave it. The same conservatism applies to headings that Search Console shows ranking for their own queries: rewording a heading that ranks is volunteering to re-run an auction you already won. Sections earning featured snippets are the extreme case, Google has told you it likes that exact phrasing, so edit around it, not through it.
Everything else is fair game. Scan competitors currently ranking above you and list the subtopics they cover that your post skips; those gaps are usually why they outrank you. Update every year reference, price, and screenshot, since visible staleness is the fastest way to lose a click to a fresher-looking result. Then rewrite the intro, and do it last. Most decayed posts bury their answer under three paragraphs of throat-clearing, and once the rest of the refresh is done you finally know exactly what the direct answer should say. Freshness pays twice now: AI assistants weight current sources in retrieval too, which is how businesses get recommended by ChatGPT without ever buying a placement.
Reading the Signals: Cause and Fix
| Signal in Search Console | Likely cause | The fix |
|---|---|---|
| Impressions steady, clicks down | Title lost the CTR contest on the results page | Rewrite the title and meta description; keep the URL |
| Impressions down, position down | Stronger competing pages published since | Add the missing subtopics and current facts |
| Position 5 to 15, stable for months | Almost there; page lacks focus on that query | Dedicated section for the query, internal links pointing in |
| Two of your URLs rank for the same query | Cannibalization splitting signals | Merge into the stronger URL, 301 the weaker |
| Traffic cliff on an algorithm-update date | Quality reassessment caught the page | Raise evidence and specificity across the whole post |
| Zero impressions for 12+ months | Wrong topic, or the page never indexed | Prune it or serve a 410; do not refresh |
The last row matters as much as the first five. A refresh queue works only if pages can also exit it: some URLs earn an update, some earn a merge, and some earn deletion. Spending refresh hours on a page the table says to prune is the most common way teams make this practice feel unproductive.
How Often Should a Library Be Re-Checked?
Quarterly for most sites, monthly once the library passes roughly 200 posts or the niche moves fast, think software pricing, hiring, anything regulated. Decay is gradual, so checking more often than monthly mostly produces noise, and checking less often than quarterly lets cheap fixes become expensive ones.
The harder question is how much of the content budget the re-check should feed. A workable operating ratio for a mature library, commonly used as a planning default, is roughly half of content hours on refreshes and merges rather than net-new posts. That is a planning heuristic, not a study result, and young sites should ignore it entirely: with 15 posts there is little to decay and everything to build. But once a library holds 30 or more posts, the pool of already-ranking URLs usually contains more recoverable traffic per hour than the next new topic does, and a calendar that is 100 percent net-new is quietly leaving that traffic to competitors.
The Break-Even Math on Refreshing Against Writing New
As typical working figures: a competent refresh takes 2 to 3 hours, while a comparable new post takes 8 to 10 including research. One writing day therefore funds either one new post or three refreshes spread across three already-ranking URLs. The refresh also pays back sooner, because the URL it improves is not starting from zero.
A refreshed URL already holds the three assets a new URL spends months accumulating: age, internal links, and a query history that tells Google exactly which auctions the page belongs in. That is why refreshed pages commonly recover positions in weeks while a new URL crawls through its cold start. None of these numbers are measurements from a study, and a genuinely broken post can eat far more than 3 hours, which is exactly when the prune option earns its keep. But as planning figures they explain why teams that add a refresh lane to their calendar usually see faster movement than teams that only publish.
The mechanical parts of this loop automate well. In SearchHandled's pipeline, the Refresh agent runs the decay detection from this article on a schedule and drafts the update for review, so the quarterly pass happens whether or not anyone remembers it. If you are doing it by hand, the free tools cover the diagnosis half, and the small business SEO checklist slots refresh work into the rest of a four-hour monthly routine.
Questions People Ask About Updating Old Posts
- How often should you update old blog posts?
Quarterly is the right cadence for most sites: pull a Search Console comparison of the last three months against the same three months a year prior and refresh the pages showing decline. Libraries over roughly 200 posts, or sites in fast-moving niches like software pricing or regulations, benefit from a monthly pass because decay accumulates faster than a quarterly review catches.
- Does updating a post's publish date help SEO?
Only when the content genuinely changed. Search engines compare the page against their stored copy, and a date bump with no substantive edits is a known trick they discount. Update the date when you have added sections, corrected facts, or rewritten stale material, and leave it alone otherwise. A modified date that reflects real changes reinforces trust; one that does not erodes it.
- Should you delete old blog posts that get no traffic?
Often, yes. A post with zero impressions across 12 or more months is not competing for anything, and if its topic sits outside your service area it dilutes the site's focus without returning anything. Merge near-duplicates into one strong URL with a 301 redirect, and prune true dead weight with a 410. Deleting a bad post helps a site more often than polishing it does.
- Should you update old blog posts or write new ones?
Update first when a post already earns impressions, holds positions 5 to 15, or has decayed from a higher position, because the URL keeps its age, links, and query history. Write new only when no page on your site addresses the topic at all. A workable planning split for a mature library is roughly half of content hours on refreshes and merges, half on genuinely new topics.
- How long does it take to see results from a content refresh?
Re-crawling typically happens within days of the update, especially if you request indexing in Search Console. Position movement commonly shows up within two to eight weeks, though this varies with crawl frequency, competition, and how substantial the changes were. Because the URL already holds age and query history, a refresh usually moves faster than a new page starting from zero.
