Content operations
SEO Content Refresh: Fix the Right Pages, In the Right Order
Most sites do not need more articles. They need the fifteen pages that already earned links, clicks and rankings to be brought back to date — without losing what made them rank. That is the whole job of an SEO content refresh: diagnose decline, rewrite surgically, republish carefully, measure by page.
Scope-based quote · no pricing on this page · monthly or project
- title: 10 Tips for Widgets- updated: March 2021 h1: 10 Tips for Widgets- stat: 42% of buyers section: picking a widget- image: widget-old.jpg- links: 0 internal+ title: How to Choose a Widget (2025)+ updated: reviewed this quarter+ h1: How to Choose a Widget+ stat: current source, dated+ section: comparison table+ image: diagram, alt text+ links: 6 internal, 2 sourcesDiagnosis
Decline is quiet before it is obvious
A page rarely falls in one step. It slides: the data ages, competitors publish something better, the title was written for a query nobody uses any more, and internal links move to newer posts. Search Console shows the result months later, and by then the page has already lost the compounding value of its links.
Four signals that a page is decaying
Click decay on a stable query set is the clearest sign: impressions hold while clicks fall, which usually means the snippet or the intent match is failing. A slow position slide from 4 to 9 across a quarter is the second. The third is factual staleness — a dated statistic, a discontinued product, a law or price that changed. The fourth is cannibalisation: three of your own pages competing for the same query and none of them winning.
What does not count as decay
Seasonality looks like decline every year and needs a year-over-year comparison, not a month-over-month one. A single algorithm update that reshuffles the whole category is not page-level decay either. Sorting those three apart before touching anything prevents the worst outcome: rewriting a page that was healthy.
Where to look, in order
Start with pages that had traffic a year ago and lost at least a quarter of it, but still hold impressions. Then the pages ranking in positions 5 to 15 with a click-through rate well below the position average. Then anything containing a date older than eighteen months, a statistic without a source, or a product that no longer exists. The full inventory comes from an SEO Content Writing Services-style crawl, but the shortlist is usually visible in an afternoon.
The dashed line is where a refresh still compounds. Past the flat section, the page has lost its link equity advantage and the same work buys less.
Scope of change
The refresh checklist, in six layers
A refresh is not a rewrite from zero. Keep the URL, the intent and the earned links; replace what is wrong, thin or missing. Work top-down: the layers below assume the page has already passed the diagnosis in the previous section.
Every number dated and sourced. Statistics from 2019 are replaced with current figures or removed. Original data beats borrowed data, so add one small dataset only you can publish: a survey, an internal benchmark, a sample of client results.
Screenshots that show last year's interface, case studies with retired products, and quotes from people who have moved on all get replaced. New examples should be specific enough that a reader can act on them.
Add the section that is now the real question. Merge duplicated headings, split the paragraph doing three jobs, add a comparison table, and move the answer closer to the top for readers arriving from a snippet.
Link out to the pages published since the original: the newer guide, the pricing page, the case study. Add the missing entities — product names, standards, tools — as plain text in the body, and fix anchors that still say "click here".
Rewrite for the query as it is searched now, keep the primary keyword, and check the snippet against the competition. Update Article or HowTo schema, adjust the visible date only when the content genuinely changed, and leave the URL alone.
Replace screenshots of interfaces that no longer exist. Compress, add descriptive alt text, and give diagrams a caption so the page still makes sense when an image is not loaded. Media weight is usually the fastest Core Web Vitals win on an old post.
Decision
Refresh or start again
Refreshing keeps history. Publishing a new URL starts from zero unless you redirect. The choice comes down to three questions, and the answer is usually refresh.
Keep and update when
- The URL has backlinks or ranks for a set of queries
- The search intent is unchanged, only the facts moved
- The page is thin but the topic is still your core offer
- Two of your pages compete: merge the weaker one in and redirect
- Traffic dipped but impressions held
Start again when
- The intent has changed and the URL now answers the wrong question
- The topic is off-strategy and dilutes the cluster
- The page carries a penalty-risk history worth leaving behind
- Nothing links to it and nothing ranks for it
- The content is accurate but belongs in a different format entirely
When the decision is a merge, the weaker URL is redirected to the survivor with a 301, its best sections are folded in, and its internal links are updated in the same release. When the decision is a new page, the old URL still needs a redirect if it ever earned a link — otherwise you are throwing away the only asset that page had.
Sequence
The order that avoids "we updated it and lost rankings"
Ranking loss after a refresh is almost always caused by doing six things at once and having no way to tell which one broke it. The sequence below keeps the URL stable, changes content before metadata, and gives each release time to settle.
Record the baseline
Export clicks, impressions, average position, top queries and conversions for the page, thirty days before touching it. Without this, nobody can prove what happened afterwards.
Photograph the old version
Save the rendered HTML, the title, the H1, the schema block and a screenshot. If something breaks, the restore point is exact rather than approximate.
Change the body first
Data, examples, structure, links. Keep the title, H1 and URL untouched in this release. Content improvements rarely cause a drop on their own, so this isolates the variable.
Wait one to three weeks
Watch impressions and average position. If they hold or improve, the body work is safe and the metadata change can follow. If they fall, revert the body before touching anything else.
Then rewrite the title and meta
One metadata change at a time, logged with the date. Expect a short CTR fluctuation while the new snippet is evaluated.
Update the date only if it is honest
A visible "updated" date should reflect real change: new data, new sections, corrected claims. Backdating or bumping every month is a trust problem, not an SEO tactic.
Re-crawl and check the index
Request indexing, confirm the canonical is unchanged, verify internal links resolve, and check the sitemap still lists the URL once. Then leave it alone for a month.
Prioritisation
Scoring pages so the list stays short
An audit can produce two hundred refresh candidates. A team can do eight a month. The scoring model below keeps the choice honest: traffic at risk, how fast the facts go stale, and what the page is actually worth commercially.
| Page | Traffic at risk | Freshness risk | Commercial value | Action |
|---|---|---|---|---|
| buyer-guide | High · 4,100/mo | High · pricing section | High · demo page | Refresh now |
| compare-options | Medium · 1,800/mo | High · competitor list | High · trial signups | Refresh now |
| how-it-works | High · 3,200/mo | Low · stable concept | Medium | Next quarter |
| glossary-term | Medium · 900/mo | Low | Low | Link only |
| old-launch-post | Low · 120/mo | High · discontinued | None | Merge and redirect |
Traffic at risk is the twelve-month peak compared with the current month, not the absolute number: a 300-visit page that halved deserves attention before a flat 3,000-visit page. Freshness risk is judged by how quickly the facts decay — pricing, regulation, product availability, tooling. Commercial value is the page's distance from a conversion, which is why a lower-traffic comparison page can outrank a popular glossary term. When the three scores disagree, refresh in this order: high risk, high value, then medium risk with existing links.
Execution
How a refresh project runs
Shortlist
Decline scan, intent check and cannibalisation map. The output is a ranked list, not a spreadsheet of everything.
Rewrite
Editorial work on the body: sources, examples, sections, tables, links. Reviewed by whoever owns the subject.
Ship in stages
Body, then metadata, then schema, with a dated log per page and a restore point saved each time.
Measure and repeat
Page-level read at 30 and 90 days, then the next batch. Pages that do not respond are investigated, not rewritten again.
Evidence
What a refresh actually moves
Comparable engagements, reported by page rather than by site. These are examples of the pattern we look for, not a forecast of your numbers.
Twelve-month click growth on a refreshed buyer guide, same URL, after pricing data and a comparison table were updated. Ninety days.
Average position recovered on a product page after thin sections were replaced and internal links added. Seven weeks.
Share of refreshed pages that returned to their previous peak within one quarter, with no new backlinks built.
Measurement
Proving it worked, or admitting it did not
Measure per URL, not per site. A site-wide traffic line hides the fact that a refreshed page recovered while three others kept sliding. The working set is small: clicks and impressions on the page's own query list, average position within a band rather than a fraction, click-through rate against the position baseline, and the conversion event attached to that page — form, trial, add to cart, call.
Two comparison windows matter. Thirty days tells you whether the release broke something: a canonical change, a redirect loop, a lost internal link. Ninety days tells you whether the refresh worked commercially. Anything faster than that is noise on most sites, and treating noise as signal leads to a second unnecessary rewrite.
If a page declines after ninety days and nothing technical broke, the diagnosis was probably wrong: the intent moved, the competition improved, or the page was never the right asset. That is when a On-Page SEO Services brief beats another round of edits, because the problem is the argument the page makes, not the words around it.
Scope
What a quote covers
A refresh engagement is scoped by the number of pages, how much of the underlying research needs rebuilding, who reviews subject accuracy, and whether the implementation happens in your CMS or ours. A batch of eight informational posts is a different scope from a set of product pages that need new comparison data and pricing logic.
You get a written scope: the pages, the order, the layers of change applied to each, the review loop, and the reporting dates. Nothing is priced on this page because the page count is the smallest part of the variable — the review and implementation load usually decides the shape of the work.
Typical shape: a fixed diagnostic, then monthly batches with a page log you can keep.
Questions
FAQ
How long before a refreshed page recovers?
Two to six weeks for the first signal on most sites, and a full quarter before you can judge the commercial result. Pages with strong links and a stable query set recover fastest. Pages that lost rankings to better competitors take longer and may need new evidence, not just new wording.
Will updating the date alone help?
No. A new date on unchanged content is a cosmetic edit, and search engines evaluate the substance. If the change is minor, leave the date and log the edit internally. If it is substantial, updating the date is accurate rather than tactical.
Should I keep the old URL?
Yes, unless the intent has genuinely changed. Same URL keeps links, history and rankings. If a merge is required, redirect the weaker page to the strongest one with a 301 and update internal links in the same release.
How many pages should we refresh at once?
Batch by five to ten pages so each release is reviewable and measurable. Large simultaneous changes across a whole section make it impossible to attribute a movement, in either direction.
Does a refresh need new backlinks?
Not always. Existing links often carry the page back once the content matches the intent again. When the page never had links, the refresh should include an outreach or digital PR component, or the ceiling stays low.
What do you need from our team?
Access to analytics and Search Console, the CMS or a developer who can implement, subject review for accuracy, and one approver. The diagnostic runs without any of that, but implementation stalls without it.
Can you refresh pages we wrote in-house?
Yes, and it is the usual case. We keep the voice and the technical detail that only your team knows, and rebuild the structure, sourcing and metadata around it.
Get the shortlist before the rewrite
Send the site and read access to Search Console. You get a ranked refresh list with the reason each page is on it, before any content is touched.