Refreshing Old Articles: Update, Rewrite, or Merge

Refreshing Old Articles: Update, Rewrite, or Merge — A Decision Table

Bottom line first: Refreshing is the fastest-recovering content investment for the same hours worked — an old article that once ranked usually climbs back faster than a new one builds from zero. Judge by four signals only, and there are only three ways to handle it: minor update if the data and dates are stale (2–4 hours), rewrite if the structure holds up but the writing is dated (1–2 days), merge if it competes for the same intent as another weak article. After refreshing, check results at 14, 30, and 60 days. Leave old articles alone if they hold steady at position 1 with stable traffic.

1. This article covers one thing only: how to decide on refreshing a single old article

Draw the boundaries before you start. For how to protect rankings when you change domain or site structure, we have a dedicated article on migration and 301 mapping; for how to handle multiple pages competing for the same keyword,keyword cannibalization gives the full decision and merge process; for the maintenance tasks to run every month,six monthly maintenance tasks sets the rhythm. Once you’ve dealt with all three, one type of problem remains: a batch of individual old articles that look fine on their own but are slowly losing rankings and traffic. This article covers just that one thing — how to decide on refreshing a single piece of old content: which signals call for a refresh, how far to go, and how to verify the results.

The value of refreshing comes from a simple fact: old articles have history. They have an indexed URL, backlinks, and accumulated click data, and search engines have already made a quality judgment about them. A new article has to go through the whole crawl-evaluate-gain-traffic process from scratch; a refresh only needs the search engine to re-evaluate once. For the same 4 hours of work, a refresh typically pays back in one-third to one-half the time of a new article.

But refreshing is also the most easily abused tactic. Many teams refresh by “fixing whichever article looks off,” keep no record, set no observation window, and three months later can’t say which change made the difference. Refreshing only compounds if you manage it as a small project with clear criteria and a verification step.

2. Four signals: which old articles belong in the refresh pool

“This one feels old” is not a reason to start. Of the four signals below, the first three come from data, and the fourth from the search results page itself.Start if two signals hit at once, or if the third (outdated information) hits on its own If only the fourth signal applies, redo intent confirmation before you touch anything.

Signal	Where to check the data	How to tell	Description
Rankings dropped off the first page	Search Console query data	Main keyword fell from positions 1–10 to 11–30 and stayed there for 60+ days	Short-term position swings don’t count. Look at the 60-day average.
Traffic keeps declining	Organic clicks, month over month	Cumulative drop of more than 30% over the past 6 months	Rule out seasonality: compare with the same period last year, not just last month.
Content contains outdated data	Manual spot check	Data, years, prices, or standard versions cited in the body are more than 24 months old	This one stands on its own. Lowest risk.
SERP intent drift	Search the main keyword manually	The page types in the top 10 are clearly different from yours (e.g., you wrote an explainer, but the top results are all comparison pages)	It means what buyers want has changed. A small update won’t save it.

Of the four signals, the third is the easiest to overlook and the easiest to fix. Say you wrote an Export Certification Checklist in 2023 and it still cites the old standard versions. Swap in the current standard numbers, years, and effective dates, and rankings often respond within two weeks. The fourth is the hardest: intent drift means the problem isn’t how you wrote it, it’s that buyers no longer want this article. In that case, a refresh will most likely fail. See the give-up line in section seven.

How to build a refresh pool: every quarter, export impressions and clicks for all pages from Search Console and run them against the thresholds in column two. You’ll usually flag 5%–10% of your pages. On a 60-article site, that’s typically 3–6 articles per quarter, a workload one person can clear in a week.

3. Decision table: update, rewrite, or merge?

Once an article enters the refresh pool, run an intent and structure check, then pick one of three paths. The core question is simple:Is the question this article answers still the same question?

Path	When it applies	Effort	Observation period	Main risk
Small update	Intent hasn’t drifted, H2 structure still works, only data, years, links, and screenshots are outdated	2–4 hours per article	21–30 days	Almost none. Lowest cost.
Rewrite	Intent hasn’t drifted, but the structure and writing are stale and coverage is thin. Needs 30%+ new information.	1–2 days per article	30–60 days	Big changes trigger re-evaluation. Rankings will swing in the short term.
Merge	Competes with another weak article for the same intent; neither ranks on its own	0.5–1 day per group	60 days	The weak article’s backlinks are lost. Requires the full 301 process.

Three sentences cover this table:Same question, outdated answer: small update. Same question, outdated approach: rewrite. Two articles chasing one question: merge. For the exact merge sequence, how to pick the main page, how to 301, how to re-point internal links, follow the six steps in the keyword cannibalization article. No need to repeat them here.

The cost of picking the wrong path is asymmetric. If you do a minor update on a page that should have been merged, you waste 4 hours but lose nothing. If you rewrite a page that only needed a minor update, you spend 2 days for no gain and disrupt rankings that were stable. So when you’re not sure,start with a minor update, and only escalate to a rewrite if it shows no effect after 30 days— this order follows the same logic as “rewrite first, then merge” in cannibalization handling: do the cheap thing first.

4. The six-item minor update checklist

A minor update is not “changing a couple of sentences.” Only when all six items below are done does it count as a proper minor update, totaling 2–4 hours of work.

No.	Item	What to do	Time
1	Data and years	Replace every year-bearing number in the body with the latest verifiable version, and note the year of the data source	40–60 minutes
2	Broken links	Open every external and internal link one by one, replace 404s with new sources on the same topic, and repoint dead internal links to existing pages	30 minutes
3	Add new information	Add 2–3 information points the original draft didn’t have (new standards, new cases, new policies) — this is adding, not rewriting	40–60 minutes
4	Verify conclusions	Reread the conclusion section and confirm it still holds under the latest data; rewrite or delete any paragraph that doesn’t	20 minutes
5	Screenshots and images	Update outdated screenshots, product images, and interface images to the current version	20–30 minutes
6	Mark and submit	Mark the page with the update date and submit a recrawl request in Search Console	10 min

The third item is the one most often skipped, yet it’s exactly the key signal that tells search engines “this article has changed.” If you only swap numbers without adding information, the page differs so little from the old version that it may be treated as an insignificant edit.The 2–3 new information points should go in the first half of the body or in their own paragraph — don’t pile them at the end.

The third item comes with another rule: added information must serve the original intent. For an old article about selection methods, what you add should be a parameter comparison for the new year, not a company news item squeezed in — the latter dilutes topical density and is negative optimization.

5. The order of operations for a rewrite, and how to handle the update date

The most common mistake in a rewrite is “tearing it all down”: changing the URL, the title, the primary keyword — that’s throwing away everything this article has built. The right approach is to keep the skeleton assets and replace only the flesh. The order of operations has five steps:

Steps	Action	Key constraint
1 Keep the URL and primary keyword	Keep the URL unchanged; if the original primary keyword is synonymous with the new one, keep the original and put the new one in H2s and body text	Changing the URL makes it a new article, and rankings start from zero
2 Rearrange the H2 structure	Re-list the H2s according to current intent, keeping overlap with old H2s under 30%	H2s are the main signal for re-evaluation
3 Add more than 30% new information	New cases, new data, new comparison dimensions — and original compilation, not reposted content	If new content is under 30%, treat it as a minor update
4 Update internal links and images	Update at least 3 outbound and 3 inbound link anchor texts, and redo the images	Inbound links bring authority back from the new article
5 Submit for recrawl and log it	Submit to Search Console and log the date and scope of changes in the refresh log	Without a log, you can’t attribute results after 60 days

Date handling is a detail you can’t skip in a rewrite. Pages usually carry two dates: the publish date and the updated date. Here’s how to handle each:

Date field	What to do	Reason
Publish date	Keep the original value, don’t change it	Preserves the “evergreen content” seniority signal
Updated date shown on the page	Set it to the actual refresh date, e.g. “Updated September 2026”	Both users and search engines can see the content is current
dateModified in structured data	Update it to match the refresh date	Keeps it consistent with the page and avoids conflicting signals
Changing the publish date to today	Don’t, unless the whole piece is published as new content on a new URL	A date jump hurts snippets and trust

Changing the publish date to today is a common mistake. It strips the article of its “evergreen” seniority signal, does nothing for rankings, and can create date confusion in snippets. There’s only one case for changing both: the content is effectively a different article. Then publish it as new content on a new URL and 301 the old one.

Once the rewrite is done, archive the full old version before you overwrite it. If rankings drop instead of rising during the 60-day window and show no recovery after 30 days, you need to compare against the old version to see what you lost. Without an archive, you’re just guessing.

6. Post-refresh validation: check in at 14, 30, and 60 days

Finishing the refresh isn’t done. Passing validation is done. Each checkpoint has its own bar:

What to Watch	What to check	Bar	Action when it fails
Day 14	Crawling and indexing	Updated content recrawled at least once, no new indexing errors	If not crawled, submit the URL manually and check the sitemap
Day 30	Rankings	30-day average for the main keyword back within ±5 positions of its best position in the 90 days before the refresh	If not, compare against the archive to find the missing information
Day 60	Organic clicks	Monthly clicks at least 80% of the same period before the refresh (minor update) or flat (rewrite)	If still below 80%, plan a second refresh or consider merging

Ranking swings of 5–15 positions between days 7 and 20 are normal, especially after a rewrite: search engines are re-evaluating a “familiar page that changed.” Don’t write this swing into your conclusions or trigger any corrective action. What should worry you is no direction by day 30: it means the changes missed the mark, usually because the new information is off-topic or the intent has actually drifted.

Log every refresh in a single table with six fields: URL, action type (update / rewrite / merge), refresh date, summary of changes, data from the three checkpoints, and verdict (on track / rework / abandon). Review this table at the end of each quarter. After a year, you’ll see which action type pays off best on your site. Most sites find that small updates account for more than 60% of all work — which is why the decision table puts small updates on the first path.

7. How far to go with a refresh, and when to walk away

A refresh has two brakes. The first is “three cases where you should not touch it,” and the second is “three cases where you should walk away.” The first means leave it alone for now; the second means it will never be worth touching.

Cases where you should not touch it	Characteristics	When to revisit
Ranked #1 with stable traffic	90-day average position for the main keyword ≤3, and click-through change year over year under 10%	Revisit the refresh pool when it drops below position 5
Published less than 6 months ago	Has not been through a full evaluation cycle, so the data tells you nothing	Add to the quarterly screening once it hits 6 months
The site is in the middle of major changes	A migration, redesign, or bulk adjustment is underway, or ended within the last 30 days	Handle it the quarter after the major changes wrap up

An old #1 article is the most sensitive asset on your site. A refresh triggers re-evaluation — pure risk, zero expected return.

Cases where you should walk away	Basis for determination	Replacement move
The intent is gone	Monthly impressions for the main keyword have fallen to single digits, or the format of the top results has changed completely	Keep the page but stop investing in it, and shift your effort to the new intent
The product line has been discontinued	The product in the article is no longer sold, and there is no replacement model	301 it to a similar product page; do not refresh
It conflicts with your brand positioning	The content direction is unrelated to your current core business	Do not 404 it. Do one round of trimming and rewriting, then freeze it

Abandoning is not deleting. Deleting the URL throws away all the backlinks and long-tail keywords that page has built up. The right move is downgraded maintenance: keep it, stop spending hours on it, and check every six months whether a new intent has appeared. The principle of conserving resources for refreshes is —spend all your effort outside the quarterly refresh pool on new content. If refreshes eat up more than 20% of your content hours, your topic pipeline has a problem. Go build up your topic bank instead of dredging up old work. If you want to see how one person can set the pace for content, see Capacity planning for a one-person team for how to count hours.

FAQ

How many articles should you refresh in a year?

Work backward from a cap of 15%–20% of your content hours: a team publishing 4 new articles a month should spend no more refresh hours than it takes to produce 1 new article a month, which works out to roughly 6–12 refreshes a year. Go past 20% and your pipeline of new topics is too thin — build up topics first instead of refreshing more. Set the exact number by your quarterly screening results, not by the calendar. If only 3 articles qualify, do 3 — don’t pad the list.

How big a change means you should review it as a new article?

If more than 60% of the body has changed, the H2 structure has been fully reordered, or the primary keyword has been swapped, review it against the new-article standard: does the title still match the content, is the intent still clear, do all internal links need to be repointed? A rewrite falls into this tier. Minor updates (swapping data, changing links, adding 2–3 pieces of information) do not trigger a review, but you still need to check the conclusion section. The dividing line is the share of new information: under 30% is an update, over 30% gets the new-article standard.

What if rankings drop after a refresh?

First tell fluctuation apart from a real decline: a 5–15 position swing within 7–20 days needs no action. If on day 30 you are still more than 5 positions below your best pre-refresh ranking, do three things: compare against the archived version to find lost information blocks, check that internal links have been repointed correctly, and confirm the primary keyword is not being cannibalized by other pages on your site. If there is still no recovery 60 days after these three checks, roll back to the old archived version or start a merge project. Do not stack a second round of changes within 30 days, or you will not be able to attribute the results.

Should I change the update date on the page?

Set the update date to the actual refresh date and keep the original publish date. Changing only the publish date without marking an update means users still see old information, which damages trust; changing the publish date to today throws away the seniority signal of “evergreen content,” and the date in your snippet will keep jumping around. There is only one case where you change both dates: when the content is effectively a different article, in which case publish it as a new article with a new URL and 301 the old one.

What should I do with old outbound links in an old article?

Open and check each one, then handle it in one of three ways: if the target page is still live but the content is outdated, swap it for a more authoritative new source; if the target page is a 404 or a permanent redirect, find an equivalent page to replace it, or delete it if you cannot; if the target page contradicts your position (for example, an old standard replaced by a new one), you must replace it, because citing outdated evidence is a serious hit to your content’s credibility. Run this check on every refresh; it takes about 30 minutes per article.

How should I flag industry data cited in the body that has gone stale?

Replace it with the latest version and note the source year, for example “according to a 2025 industry report.” If you cannot find new data and only old data is available, keep the old data but note the year of the statistic in the sentence, and tell readers in the conclusion section when the data was current. What you must never do is change only the year and not the numbers: once the numbers and the year do not match and a reader or an AI cross-checks them, the credibility of the whole article comes into question.

Final thoughts

A refresh decision needs a table because its biggest cost isn’t the hours — it’s refreshing the wrong page. You touch what should be left alone, and you rewrite what should have been retired, over and over. Four signals screen the candidates first, three paths then set the action, and three checkpoints verify the result at the end. Every step in between has a pass line. Only then is refreshing asset management rather than a side task.

The starting advice is simple: this quarter, pull a full-site data export from Search Console once, use the thresholds in section two to build your refresh pool, and start with the two pages whose signals are strongest — usually one small update for outdated information and one rewrite for stale phrasing. Log them, watch for 60 days, and once you’ve run one full cycle, scale up.

One line worth remembering: refreshing isn’t about making an old article look better — it’s about making the search engine answer once more: “is this article still worth ranking near the top?”