How to refresh a published article

A content refresh is a substantial change to an article you already published, so the page tells the reader something more accurate or more complete than it did. To update old blog posts well, you change the passage first. The date, the schema, and the sitemap record that change. They do not substitute for it.

What is a content refresh?

Teams use three phrases for one file. A refresh keeps the URL and the job, and replaces the parts that are no longer right. A rewrite replaces the job, because the results are now a template and the page is still an essay. Republishing a blog post by moving the date and leaving the sentences is a third habit, and it leaves the reader with the old explanation. The work this guide describes is the refresh. If the job itself changed, write that decision in an SEO content brief before anyone drafts. The brief is how you notice you are no longer editing.

Ahrefs’ guides on content decay, fresh content, and republishing agree on one operational rule: do not change the date and stop. They then route the work through their own reports, including a declining-traffic filter and an AI helper that suggests topic gaps. Those reports are useful when you can explain them. The gap they tend to fill is “add what competitors mention.” Adding every adjacent subtopic is how a standup procedure becomes a second pillar. A content update SEO pass should make the page more true for its job, which often means cutting.

This page assumes the URL was already labeled update. That label comes from an SEO content audit. If two of your URLs share the query, stop and merge. Refreshing both preserves the overlap.

Content decay is a reason to look

Content decay is a decline in clicks or visibility on a page that used to receive them. It is a symptom. The chart does not say which sentence failed. A dip can come from a stale fact, a shift in what the results show, a second article of yours that split the query, a results page that now answers without a visit, or a topic that was seasonal. Treating every decline as an order to rewrite will have you editing pages that were fine.

Read the page against the query before you read a tool’s topic score. If the procedure is still what a reader needs, and one vendor name is wrong, the refresh is a paragraph. If the results have become product roundups and your page is a procedure your customers use, you may keep the procedure and accept that the head query moved. Chasing the roundup on a how-to URL abandons the readers who still need the steps. The topic cluster page is where a genuinely new job gets its own URL, after you have checked that you do not already have one.

Declining clicks are worth a row in the audit. They are not a quota. “Refresh the ten biggest losers every quarter” produces date changes on pages whose only problem was a noisy chart.

What counts as a substantial update

Google’s helpful-content guidance asks two questions that apply directly here. Are you changing the date of pages to make them seem fresh when the content has not substantially changed? Are you adding or removing a lot of content mainly because you believe that will make the site seem fresh? Both are warning signs. Neither is a ranking method.

The practical line for “substantial” is in Google’s sitemap documentation. It says lastmod should reflect the last significant update of the page. An update to the main content, the structured data, or the links is generally significant. An update to the copyright date is not. Google also says it uses lastmod when that value is consistently and verifiably accurate, for example by comparing it with the page itself. A lastmod that moves on every trivial save teaches the field to be ignored.

Byline dates in search are Google’s estimate of when a page was published or significantly updated. The byline-date documentation says Google does not depend on a single date signal. You help that estimate by showing a visible date, labeled Published or Last updated, and by putting the same day in datePublished or dateModified. Keep the timezone consistent if you include one. Do not use a future date, or the date of an event the article describes, as the date of the page.

So the visible line on a refreshed article can read “Published 12 March 2024. Updated 6 October 2026.” datePublished stays 12 March 2024. dateModified becomes 6 October 2026. The sitemap lastmod for that URL becomes the same day. A copyright year in the footer is not any of those fields.

How to update one article

When you update an article for SEO, do the reader-facing work before the metadata. The sequence below is for a URL the audit already labeled update. Skip a step only when it does not apply, and write “none” in your notes so the skip is a choice.

1. Confirm the query and the format

Open the query in a private window. Write one line on what the results mostly are now. If they are a different job from the page, go back to the audit. You may be looking at a new URL, or at a page you should keep as it is. A refresh that quietly changes a procedure into a product roundup has changed the job without saying so.

2. Correct the claims a reader could be hurt by

List sentences that name a product, a price, a step, a policy, or a number. Check each one against a source you can link, or against the product as it works today. Remove a claim you cannot support. Do not replace an unsourced percentage with a newer unsourced percentage. Answer engines repeat precise sentences. A refreshed article that invents a fresher statistic has made the citation worse. Google’s helpful-content notes also say that giving credit to other sources does not replace original effort. Linking the policy is the attribution. The original effort is the judgment you add: what the policy means for this page’s reader.

3. Cut what another URL now owns

If a section repeats a sibling article, delete the duplicate explanation and link the sibling. Anchor text should describe that destination in the sentence you already needed. “Why status meetings run long” is an anchor when the next sentence is about meetings, and the destination is that essay. “Click here” is not. Crawlable links are ordinary HTML links, so the reader and the crawler can follow them. A refresh is a normal time to add those links, and Google treats a real link change as a significant update. It is also a normal time to remove a link that now points at a URL you merged away.

4. Put the answer where it can be quoted

Move the direct answer to the start of the section. Then explain it, then give the example. A person scanning, and an answer engine extracting, both meet the claim before the history. The answer engine optimization guide is the longer version of passage shape. On a refresh, you are repairing a passage that already exists, which means deleting the throat-clearing that used to sit above the steps. Information gain, on an update, is the correction and the example you can stand behind. It is not a block of related terms pasted under the old introduction.

5. Change the title only when the promise changed

If the page still teaches the same procedure, keep the title that already describes it. If the promise changed, rewrite the title and the H1 so they match the new first paragraph. Google’s title-link guidance is about titles that describe the page. A year swapped into an otherwise identical title is a date costume. A keyword stacked onto the old title is the same costume. Write the promise in ordinary language, near the length of a result, and leave the slug alone unless the old slug is actively misleading. Historical optimization, below, is why the slug usually stays.

6. Set the dates, the schema, and lastmod

After the article on the live URL says the new thing, set the visible updated date, dateModified, and lastmod together. dateModified belongs in the article schema. lastmod belongs in the XML sitemap. If the page has a visible FAQ, and you changed an answer, change any FAQ schema text to the same words. FAQ rich results stopped appearing in Google Search on 7 May 2026, so the markup describes the visible FAQ. It does not produce an accordion. Mismatched schema is a trust problem. If you want the URL fetched again, use your indexing setup once this version is the one you mean. A recrawl of the old paragraph does not repair it.

A content refresh example

The audit labeled /async-standups as update. The job is still “help a six-person product team run an async standup.” The results are still procedures. The page should remain a procedure. Here is the edit, and here is what stays put.

• A stale product name. The steps tell people to post in “StandupBot.” The team moved to a thread in the tool they already use. Replace the name with the current action, and drop the sentence that praised the old vendor. There is no new statistic in that fix.
• An unsourced claim. The introduction says most teams waste forty minutes in live standups. Nobody measured that for this company. Delete it. The page is stronger without a number you cannot point to.
• A section that belongs elsewhere. “Why meetings run long” repeats the essay at /why-status-meetings-run-long. Replace those paragraphs with two sentences and a link, anchor “why status meetings run long,” because a reader of the procedure may want the causes next.
• A buried answer. The steps start after three paragraphs of history. Move a six-step list under the first heading, then keep one short example of a written update that is specific enough to be useful. That is the passage an answer engine can quote without also quoting the history.
• What you do not add. A comparison of Scrum and Kanban, a retro template, and a list of “standup best practices” that repeats the steps. Those are other URLs. Adding them would make this refresh a second cluster.

When that edit is live, the page can say “Published 12 March 2024. Updated 6 October 2026.” dateModified and lastmod use 6 October 2026. The slug stays /async-standups. The title stays a promise about running the standup, because that promise did not change.

Historical optimization keeps the URL

Historical optimization is the name for improving a URL that already has history: queries in Search Console, links from your own site, sometimes links from other sites. The history is attached to the address. Publishing /async-standups-2026 and leaving the old procedure up splits that history and creates the merge you will have to do later. Keep the address.

Change the slug when the old slug promises the wrong document. A URL that says /free-standup-tool and now explains a procedure is misleading. Pick a slug that matches the promise, redirect the old one to it, and update internal links. Google’s notes on consolidating duplicate URLs are the mechanical side of that move. A canonical tag is for the same document still open at two addresses, such as a print view. For an ordinary refresh, you do not need a new address. You need a truer page on the address you have.

Content maintenance is how you refresh existing content on a schedule you can keep. Once a year is enough for a procedure that depends on a tool. Sooner is warranted when a policy you cited has changed, or when you know a passage is wrong. A calendar reminder that says “bump dates in Q4” is the date-only pattern with a recurring invite. Put the reminder on the claims: which sentences name a product, a price, or a rule that can expire.

Optimize old content in the same voice as the rest of the site. A refresh written like a different author, stuffed with the primary query, reads as a layer added for a crawler. Use the query where it is the honest name of the job. Use the entities the reader needs, each explained once. A content rewrite that exists to raise a word count fails the helpful-content question about writing to a supposed preferred length. Add a section when the reader needs it. Cut a section when they do not.

After you publish the refresh

Give the URL time, and look at the right report. In Search Console, the queries for this page should still be the job you kept. A new query that belongs to the pillar or the comparison means the refresh widened the page. Cut that section and link the URL that owns it. Impressions can move for reasons that have nothing to do with your paragraph, including a results page that answers on its own. Do not declare the refresh a failure, or a success, from one week of clicks.

Check three consistencies on the live URL. The visible updated date, dateModified, and lastmod name the same significant day. The FAQ schema, if you have one, matches the FAQ a person can read. Internal links you added resolve, and links you removed do not still appear in a menu or a related-posts block. Then stop editing. A second pass the next morning that only swaps adjectives is how lastmod becomes noise.

BloGoose reads the sitemap before it proposes another article. When a calendar title is the job of a URL you already have, the useful action is this refresh, not a second draft that will need a merge. The pipeline can write a new page when the audit says the job is missing. It should not write a twin of a page labeled update. Review the refreshed draft the way you review any draft: the claims, the scope, the opening answer, and the dates, in that order.

Questions about content refreshes