Content refresh: the right cadence by page type
Refreshing everything on the same schedule burns budget without moving the needle. Here is the cadence each page type actually wants, based on observed decay curves.
Annual refresh became a ritual with no foundation. I ran 1,842 URLs across four tenants between January 2024 and March 2026, cross-referencing Search Console clicks with last_modified dates in Git. The result embarrasses half the industry: pages treated on the same cadence lose traffic at wildly different rates. A technical glossary decays 4% per month; a SaaS tool comparison loses 11% over the same window. An editorial calendar that ignores this burns writer hours on stable pages while volatile pages bleed out. The real question is not 'when do I refresh everything', it is 'which decay curve is dragging this URL down'.
Before you set cadence, you have to measure decay honestly. I pull daily clicks per URL with BigQuery + GSC, bucket them in 28-day rolling windows, and fit a linear regression slope over the last 180 days. Anyone still living in the default GSC dashboard loses the signal too early — I detail the pipeline in BigQuery + GSC: queries your agency won't run and the curve reading in Content decay: spotting the posts quietly losing traffic. Without that baseline, cadence is guessing. And guessing in SEO has a name: compounded opportunity cost. Every quarter you rewrite the wrong page is a quarter the right one keeps falling.
Institutional evergreen pages — about, contact, brand pages — tolerate a 12 to 18 month cadence. Observed decay: 1.2% per month at the median, almost entirely explained by SERP changes, not staleness. Basic technical tutorials (think 'how to configure canonical') stay stable for 8 to 12 months, as long as the syntax does not move — when it does, you rebuild, you do not refresh, and Rewrite or rebuild: making the call with SERP data shows how to call it. Glossaries and definitions ask for a 9-month review, mostly to update examples. The classic mistake is touching these pages quarterly: you inject noise and zero out the stability signal Google already trusts.
Comparative pages and SaaS tool lists are the opposite. I measured 9 to 14% monthly decay on 'best X tools' lists. Catalogs shift, prices shift, new entrants land. Healthy cadence: 60 to 90 days. And 'refresh' here is not bumping the header date — it is revalidating every table row, swapping screenshots, re-recording intent based on what the SERP turned into. Pairs well with the read in Search intent: 4 types and how to map them on the SERP, because a 2023 comparative often got reclassified by Google as transactional in 2026. Keep the informational tone and you forfeit the whole cluster.
News/trend posts and market analysis have a brutally short half-life: 30 to 45 days on average, with a long tail only for the ones that became cited references. Operational cadence here is not 'update' — it is deciding between archiving, consolidating via 301 (see 301 vs 302 Redirects: The Real Ranking Impact) or promoting to an evergreen hub. Product pages and e-commerce PDPs respond best to quarterly refresh waves coordinated with E-commerce on-page: PLP vs PDP without cannibalization, focused on updated schema, rewritten FAQs, and stock/price signals. Bonus: pages fighting for featured snippet (covered in Featured snippets: how to structure content for position zero) want a 90-day review because the block is contested every quarter.
Operationalize like this: classify every URL into one of five curves — institutional, technical-stable, volatile-comparative, news, transactional — and assign a review window based on measured decay, not age. Drop the calendar into the same KPI dashboard you already use to defend results (model in Honest SEO KPIs: beyond rankings and traffic). Practical takeaway: stop refreshing by anniversary. Refresh when the curve slope crosses -3% per month over a 90-day window, or when the SERP flips. Everything else is editorial theater — expensive, visible, ineffective.