Content refresh SEO: when to update an old post instead of writing a new one

Most stores respond to a falling traffic graph by publishing more. That is usually the wrong move, and it spends the one resource an independent store never has spare. The posts you published two years ago are sitting at positions 6 to 15 with impressions attached. They have age, internal links and a crawl history. A blank document has none of that.

This is the decision rule we run every month on Supplements Wise, our own store, and it is the same rule whether you run it by hand in a spreadsheet or let software do the sorting. Worked examples below come from a skincare store, because the argument is clearest in a category where advice genuinely dates.

What is a content refresh in SEO?

A content refresh is an edit to a page that already ranks, keeping the same URL, so that it recovers positions it has lost or covers ground the current results expect. That is the whole definition, and the URL clause is the important half. The moment you publish at a new address you throw away every link, every ranking signal and every month of crawl history the old page accumulated, and you start from zero against competitors who did not.

Three things get confused with each other, so it is worth separating them now:

• A refresh keeps the URL and most of the page, and adds or corrects. Two hours of work.
• A rewrite keeps the URL and replaces the content. Half a day, and only when the original was aimed at the wrong thing.
• A new post is a new URL for an intent nothing you own currently serves.

Stores reach for the third option by default because it feels like progress. A published post is visible evidence of work; an edit to a two-year-old page is not, which is exactly why the edits do not get made.

Why does refreshing usually beat publishing something new?

Because Google has already decided your page is relevant, and moving from position 9 to 4 is a smaller ask than moving from nowhere to 9. On a query with 900 monthly impressions, that is roughly the difference between 20 clicks and 120. The refresh costs an afternoon; the new post costs a week and then has to outrank your own page to earn anything.

The second reason matters more on a store than on a blog. Publish a second post targeting a query one of your pages already ranks for and Google splits its confidence between them, so both slip. We have written up how that plays out on Shopify, where the loser is frequently the collection page that was quietly taking money. A refresh cannot cannibalise anything, because there is only ever one page.

How do you find the posts that are decaying?

In Search Console, open the Performance report, set the date filter to Compare, pick last three months against the previous period, then switch to the Pages tab and sort by the click difference. The pages at the top of that list lost the most, and they lost it while still being indexed and still being shown, which is the profile of a page worth saving.

Two refinements make the list usable:

• Ignore anything published in the last six weeks. New pages drift for weeks before they settle. Judging them produces noise and a lot of pointless editing.
• Ignore seasonal drops that will reverse. A skincare store’s SPF guide falling between September and January is weather, not decay. Compare the window against the same window last year first.

Then drill into each surviving page and read its query list. That is where the actual brief is. A page can hold its total clicks while the query that used to carry it collapses, and it can lose half of them because one query slipped from position 4 to 11. The page-level number tells you where to look; the query-level number tells you what to fix.

The queries at positions 5 to 20 in that drill-down are the same striking distance list we use for new topics, so the two workflows share one export. Our full Search Console walkthrough covers the filters and the prioritisation maths, and it applies unchanged here.

What are the four verdicts for an old post?

Every page on the decay list gets one of four verdicts, decided before any writing starts. That is what stops the afternoon disappearing into whichever page you opened first.

Refresh. The page targets the right intent and ranks somewhere on pages one or two, but it is missing subtopics that the current results cover, or its specifics have dated. This is the majority verdict and the cheapest work. A vitamin C serum guide that never mentions the pH and stability questions people now search for is a refresh.

Rewrite. The page targets the right query with the wrong premise or the wrong depth. The URL is worth keeping and the content is not. A 500-word post called “how to layer retinol and vitamin C” that concludes with “ask your dermatologist” and stops is a rewrite: the query has an answer, and the page declines to give it.

Merge. Two or more of your pages target one intent. Pick the one with the stronger position, fold the useful material from the others into it, then 301 the losers to the survivor. Merging is the highest-value verdict here and the one stores avoid, because deleting a page you paid for feels like a loss. It is not. Two pages at positions 14 and 19 are worth less than one at 7.

Leave alone. The page ranks in the top three and is stable, or the query it lost simply has no searches any more. Editing winners to chase adjacent variants is a reliable way to lose the ranking you had. If a stable top-three page is missing a variant worth having, add one H2 and stop there.

What actually changes in a refresh?

Work in this order, because the order is roughly the order of impact:

1. The opening answer. Whatever the target query asks, answer it in the first two sentences of the page, in the words people search. Most decayed posts open with 150 words of preamble before saying anything. Fixing that one paragraph does more than the next four items combined, and it is also most of what makes a page quotable by AI answers.
2. The missing subtopics. Open the current top five results and list what they cover that you do not. Add the genuine gaps as new H2s. Do not add the padding, and do not add a section because a competitor has one; add it because a searcher would have a follow-up question you left unanswered.
3. The dated specifics. Prices, product names, screenshots, regulations, anything with a year in it. On a skincare store this is where most of the rot is: formulations change, the product you recommended in the second paragraph was discontinued last spring, and the guide now sends readers to a 404.
4. The internal links. Old posts were linked when they were published and then forgotten. Point the refreshed page at the collection it should sell, and point two or three newer articles back at it. Our guide to internal linking on a store has the full sweep, or the Internal Linking feature runs it as a scored graph.
5. The title and meta, but only if the problem is clicks. If the query sits at position 4 with a poor click-through rate, the ranking is fine and the promise is not. If it sits at position 14, no title rewrite will save it and the content is the problem.

What does not belong on that list: changing the URL, rearranging sections to look different, or padding the word count to match a competitor. Length is an output of coverage, not a target.

Should you change the publish date after a refresh?

Update it when you have genuinely changed the substance, and leave it alone when you have not. There is no ranking bonus for a fresher date and there is a real cost to lying about it: a reader who arrives at a post dated last week and finds advice about a product discontinued in 2024 will not trust the next page either. We show both dates, original and updated, which is the honest version and costs nothing.

Never change the URL. If the slug is genuinely wrong, for example it names a product you no longer stock, the change has to be worth losing a slice of the page’s history for, and it almost never is. When you do move a page, 301 it and expect a few weeks of wobble.

What will a content refresh not fix?

Refreshing cannot manufacture demand, and it cannot rescue a page aimed at an intent your store has no business serving. If a post has 40 impressions a month after two years, nobody is searching for it and no amount of editing changes that. Merge it into something with a pulse.

It also will not fix a page losing to results of a different type. If the query now returns a video carousel, a forum thread and three retailers with 40,000 reviews each, your 1,400-word guide is not competing against worse copies of itself. Spend the afternoon on a query where the page-one results look like pages you could write.

How does this fit into a monthly routine?

It is one pass a month, slotted into the weekly loop from the Shopify blog SEO system: pull the comparison report, sort by lost clicks, assign the verdicts, then work the refreshes before the new posts. Budget an hour for the sorting and two hours per refresh. Most months that means three or four refreshes and one new article, the opposite ratio to what most content calendars assume.

Doing this by hand works. We ran it in a spreadsheet for months before any of it was software, and the spreadsheet found the same pages. What it cost was the sorting hour plus the discipline to open the report at all, and the discipline is the part that slips first. Search Insights does the comparison and the query drill automatically, the Content Calendar carries the verdict on every entry so a refresh gets scheduled like a post rather than remembered like a chore, and Blog Writer drafts against the existing page rather than a blank one. The pricing page has the tiers, the free audit on QueryClimb runs the decay report on your own data, and our case study has the numbers from the store this was built on.

The judgement stays yours either way. Software can tell you a page lost 240 clicks. It cannot tell you whether the answer is two hours of editing or letting a dead page go.

Frequently asked questions

How often should you refresh old blog posts?

Review monthly and refresh what the review flags, which on a store with 50 to 150 posts is usually three or four pages. Refreshing on a fixed calendar, every post updated annually, wastes time on pages that rank fine and misses the one that collapsed in March.

Does updating an old post help SEO?

It helps when the page already ranks and the edit closes a real gap in coverage or corrects something stale. It does nothing when the page has no impressions to build on, and it can hurt when you edit a stable top-three page to chase a variant. The ranking you have is worth more than the one you are reaching for.

Should I delete old blog posts that get no traffic?

Merge them into a relevant page and redirect, rather than deleting outright. A post with no traffic may still hold internal links and the occasional external one, and a 301 keeps both. Delete only when nothing points at it and the topic has no home.

How long does it take for a refreshed post to move in rankings?

On our own store, one to three weeks for pages that already ranked, against four to eight weeks for new articles. Recrawling is the first step, so check the Page Indexing report if nothing has changed after a fortnight. A refresh Google has not crawled cannot rank differently.