Content refreshes

Content refreshes are the highest-ROI activity available to mature SEO programmes. Existing articles have accumulated backlinks, internal links, indexation history, and brand recognition. Refreshing them improves ranking on the same investment — versus writing new content that has to earn all those signals from scratch.

“For sites past 18 months old, 60-70% of content production hours should go to refreshing what already ranks 4-15 — the recovery wins are bigger and faster than equivalent new-content investment, because the existing pages already own most of the signals that take new content years to build.”

What a refresh actually changes

The test for whether a refresh counts: a knowledgeable reader comparing before and after should agree the new version is materially better. If they wouldn’t, the refresh didn’t happen — what happened was activity that produces no ranking signal change.

How to pick refresh candidates

Score every existing article on four axes:

• Current rank (Search Console position). Sweet spot: 4-15. Articles at #1-3 don’t need refresh; articles past page 3 may need new content not refresh.
• Demand for the topic (search volume, impression trend). High demand + slipping rank = high-priority refresh.
• Existing signals (backlinks, internal links, brand mentions). High existing signals = preserve and refresh; low signals = consider writing new.
• Topical fit and content quality drift (when last refreshed, has the topic moved on). Stale on a fast-moving topic = high priority.

Highest priority refresh candidates: articles ranking 4-15 with high topic demand, decent existing signals, and content that’s 6+ months old in a topic that’s evolved. Lowest priority: articles ranking #1, articles past page 3 with no signals, or articles on stable topics where nothing material has changed.

The refresh workflow

• 1. Audit current state. Pull the article’s Search Console queries, current ranking position per query, current word count, current sections.
• 2. Analyse the SERP. Look at the top 5-10 ranking pages for your target query. Note: word count, structure, sections, recent updates, what they include that you don’t.
• 3. Identify the gap. Comparing your article to the SERP, what’s missing? What’s outdated? What sections do top-rankers have that you don’t?
• 4. Write the refresh brief. Specify what stays, what gets cut, what gets added, what gets restructured, what data needs updating, what sources to cite.
• 5. Execute the refresh. Substantive changes per the brief — not cosmetic tweaks. Add sections, remove obsolete ones, update data with current sources, restructure for intent fit.
• 6. Update technical signals. Refresh the title tag if intent shifted, update meta description, fix any broken internal links, add new internal links from related articles, update schema, set lastmod to today.
• 7. Re-promote. Internal links from recent articles, social distribution, email mention, optionally request indexing in Search Console.
• 8. Monitor for 4-12 weeks. Track ranking on the target queries; expect material movement within 6 weeks for pages ranking 4-15.

The 2026 addition — refreshing for AI engines

AI engines (ChatGPT, Claude, Perplexity, Gemini) extract content for citation. Articles structured for extraction get cited; articles structured for narrative reading don’t. When refreshing, restructure for AI-engine readability:

• Each H2 should answer a specific sub-question. AI engines extract by section — sections that don’t answer a discrete question rarely get cited.
• Add comparison tables for any topic with multiple options. Tables are highly extractable and frequently cited.
• Add clear definitions early. The first paragraph after a major H2 should define the concept; AI engines often cite this paragraph as the canonical answer.
• Add a FAQ section with 5-7 questions covering the most common follow-ups. FAQ schema makes the section extractable; AI engines pull from FAQ blocks reliably.
• Use specific numbers and dates. “In March 2026 the threshold was 200ms” gets cited; “recently the threshold improved” doesn’t.

How often to refresh

Common refresh mistakes

• Date stamp updates with no content change. Theatre. Google detects content-diff and ignores cosmetic lastmod updates.
• Refreshing articles that should be killed. An article ranking page 5 for a query where the SERP is full of comparison content can’t be refreshed into a comparison; needs new content with the right intent fit.
• Refreshing in isolation. A refreshed article without renewed internal linking from other articles loses half the lift.
• Adding length without adding substance. Padding an article from 1,500 to 3,000 words by repeating existing points doesn’t move rankings.
• Refreshing the wrong article. Picking refresh candidates by gut instead of by data wastes the refresh budget on articles that wouldn’t lift much.
• Refreshing without checking the SERP. Refreshing what your team thinks is missing instead of what the top-ranking competitors actually include.
• Updating publish date dishonestly. “Updated 2026” on an article with the same content from 2022. Google sees the content-diff and discounts the signal; users notice and trust drops.
• No measurement after. Refreshing without tracking which articles improved closes the feedback loop and makes future refresh decisions worse.

The bottom line

Content refreshes are the highest-ROI investment available to mature SEO programmes. Score candidates by current rank (sweet spot 4-15), demand, existing signals, and topical drift. Make substantive changes (new sections, removed obsolete content, current data, restructured intent fit, AI-engine extractability) — not cosmetic edits. Hand writers a real refresh brief, not the URL alone. Update lastmod, re-promote, monitor for 4-12 weeks. Cadence by topic type. The compound effect: existing articles keep ranking, new articles get to focus on net-new topics, and the site’s overall content quality stays current without exponential production cost.