Refreshing old articles: the highest-return work on your site

Most sites sit on an asset they ignore: old articles with history, links, and impressions that sit one good edit away from real traffic. Refreshing them is usually a better investment than publishing something new into the void. This guide covers which articles to refresh, what a proper refresh actually involves, and when merging beats refreshing. It is part of how to write articles that rank; the step-by-step version is our content refresh checklist.

Which articles to refresh (read Search Console)

Open Performance for the last 3 months and hunt these patterns:

• High impressions, low position (roughly 8–25): Google already shows the page; it just is not good enough yet. The single best refresh candidate.
• High impressions, low CTR: the title/intro promise is losing the click — refresh the framing before the body.
• Slow decline: a page sliding month over month usually faces refreshed competitors or intent drift.
• Query mismatch: impressions for queries your article barely answers reveal the sub-topic Google wants from it — expand to serve them.

Skip pages with no impressions and no links until you have diagnosed them with the diagnosis guide — refreshing a page whose problem is indexing or orphaning is wasted effort.

What a real refresh includes

1. Re-check intent: read today's SERP for the main query. Has the dominant format or angle moved? Match the current one.
2. Fill coverage gaps: run the comparison against the current top three (the method in thin content) and add the genuinely missing sub-answers.
3. Update facts: dates, prices, screenshots described, version numbers, dead recommendations — anything a reader could catch being stale.
4. Rewrite the introduction using the Answer–Problem–Map pattern from writing introductions.
5. Repair links: refresh outbound links that died, and add internal links to newer cluster siblings the article predates.
6. Update the "last updated" date — only because you substantively changed the article, not as a cosmetic trick. (Google has warned that merely bumping dates without real changes does not help and can look manipulative.)

When to merge instead of refresh

If the refresh candidate overlaps heavily with a sibling — same intent, overlapping coverage — refresh energy is better spent merging them into one stronger page with a redirect, exactly as described in keyword cannibalization. Rule of thumb: refresh when the page's intent is distinct and valid; merge when two pages chase one intent.

What to expect afterwards

Movement typically shows in impressions first (1–3 weeks), then position, then clicks. Judge a refresh after 4–6 weeks, not 4 days, and change one cluster at a time so you can attribute results. Keep a simple log — page, date, what changed — because in six months you will not remember, and the log becomes your site's most valuable SEO document.

A sustainable cadence

For a small site: one refresh per week outperforms one new article per week once your catalogue passes ~30 posts. Alternate — new spokes into your clusters one week, a refresh the next — and the whole portfolio compounds instead of only its newest members.

Sources (5)

Google Search Central on site freshness and date signals (developers.google.com/search/docs) and its guidance that dates should reflect substantive updates; Search Console Performance reporting documentation.

1. Google Search Central — Creating helpful, reliable, people-first content
2. Google Search Central — SEO Starter Guide
3. Search Console Help — URL Inspection tool
4. Google Search Central — Google Search Essentials
5. Google Search Central — How to specify a canonical URL

The refresh log (steal this format)

Keep one row per refresh in a spreadsheet: URL · date · baseline position/clicks for main query · what changed (three bullets max) · review date · verdict. The verdict column, filled in 4–6 weeks later, is where the value compounds: after a year it shows which change types move your site (for most small sites: intent realignment and coverage additions outperform cosmetic edits by a wide margin), and it stops you re-trying interventions your own data has already voted against. Publishers routinely credit "SEO instinct" for what is actually this log, read patiently.

Refresh traps

• Refreshing everything at once: you lose attribution and risk a sitewide wobble you cannot explain. Batch by cluster.
• Rewriting winners: an article at position 2 earning clicks is not a refresh candidate; protect it, and spend the hour on position 12.
• Changing URL, title, and body simultaneously: if it works (or fails), you will never know which lever mattered.
• Refreshing into a different article: if the rewrite changes the page's target query, you have not refreshed — you have published a new page at an old URL and orphaned the old one's intent equity. Decide which you are doing.

Communicating changes to readers

For articles readers may have bookmarked or cited, a brief "updated" note at the top or foot — what changed and when — is both courteous and a trust signal: it tells returning readers the page is maintained and tells new ones the advice is current. Keep it factual and short. The date mechanics are covered in the checklist; the editorial principle is that maintenance is a feature to show, not a trick to hide.

The ethics of updating

A short, serious note: refreshing must never become quiet revisionism. If an old article made a claim that events have falsified, correct it visibly where the stakes warrant — a dated note beats a silent edit when readers may have acted on the original. The same transparency that makes "last updated" dates honest (see the checklist) makes a site trustworthy over years, and trust, unlike a ranking, cannot be re-earned quickly once spent.

Your first refresh session

Block ninety minutes this week. Pull the Performance report, pick the single highest-impression page sitting between positions 8 and 25, and run the full checklist on it — nothing else, no new article, no site tweaks. One completed, logged, measurable refresh teaches the whole discipline better than a month of planning, and its result (visible in about a month) is the evidence that funds the habit.

The portfolio view

Zoom out once a quarter and look at the catalogue as a portfolio: what share of your articles earn impressions, what share earn clicks, and what share earn neither? A healthy small site sees the middle band — page-two articles — as its growth engine and feeds it refreshes systematically. The tail of never-impressions is a diagnosis queue, not a refresh queue. Sorting your catalogue into those three buckets, honestly, is the single most clarifying hour in content management.

Part of: Write Articles That Rank — the complete guide for this topic.

Continue reading in this series

Topical authority, explained practicallyContent clusters, explained with a live exampleMatching search intent before you write (a five-minute routine)