Content Update and Rewrite SOP: Fixing Old Articles Without Losing Equity

Summary: Most "content refresh" projects fail at the first decision, not at the writing: they change the URL, split the page, or rewrite the whole article when a section-level edit would have been enough. This SOP fixes the order of operations — URL first, diagnosis second, rewrite third, measurement last — and gives the thresholds we use to decide between updating, redirecting, consolidating and retiring.

Key Takeaways

The URL decision precedes the writing decision. A rewrite that keeps the URL inherits every inbound link; a rewrite that moves the URL inherits nothing unless the redirect and internal links are handled in the same release.
Decay is measurable before it is visible. In a 2026 study of 18,400 refreshes across 240 portfolios, 67.4% of pages aged 18 to 24 months were already decaying, with a median traffic loss of 41.2%.
Refresh beats rebuild for most pages. The same dataset reports a median traffic lift of 47.3% after a refresh, against 84.2% for a full rebuild — but a rebuild costs several times more, so it belongs to the 14% to 18% of pages that cannot be patched.
"Updated the date" is not an update. Changing a timestamp while leaving claims, examples and screenshots untouched is a cosmetic edit; it neither improves retrieval nor persuades a reader.
Measure the rewrite on funnel outcomes, not pageviews. A refresh that recovers traffic but sends the same low-intent sessions into the same dead-end page has recovered nothing that matters.

I. Why a rewrite is first a URL decision

Every content refresh begins with a page that already has history: inbound links, internal links from sibling articles, saved rankings, and often a few hundred sessions a month that still convert. The rewrite either keeps that history or throws it away, and that choice is made before anyone opens a text editor.

The distinction matters because search engines and answer engines both treat a URL as an identity. When the content changes but the URL stays, the accumulated signals stay with the page. When the URL changes, the old identity must be transferred deliberately through a permanent redirect, and every internal link pointing at the old address must be updated in the same release. Half-done migrations are how teams lose 30% of a section's traffic and then blame the writing.

There is a second reason to decide the URL first: the rewrite scope follows from it. Keeping the URL usually means a section-level edit — replace three outdated claims, add one comparison table, rewrite the conclusion. Moving the URL usually means the page is being repositioned, which is a rebuild with a new title, new outline and new internal link targets. Treating a repositioning as an edit produces an article with a new address and an old argument.

Our rule is deliberately unglamorous: if the search intent behind the primary keyword has not changed, keep the URL. If the intent has changed enough that the current H2 outline answers a different question, build the new page and redirect the old one. If two pages answer the same question, consolidate and canonicalise rather than rewrite either of them.

II. Four dispositions and the evidence that selects one

Not every decaying page needs the same treatment. We sort candidates into four dispositions, each with an observable trigger, an expected effort and a specific failure mode.

Disposition	Trigger we use	Typical effort	What goes wrong if you pick it wrongly
Section-level update	Clicks down 15% versus the same quarter last year; ranking drift of two to five positions; facts older than 24 months	1 to 2 days	The page keeps an outdated outline and slowly decays again
Full rewrite, same URL	Intent unchanged but depth, examples and structure are behind the current top five	3 to 5 days	You spend a week to recover a page that a section edit would have fixed
New URL plus redirect	Intent has shifted; the page now targets a different question than the one it ranks for	5 to 8 days	Without a redirect and internal-link sweep, you split authority across two URLs
Consolidate or retire	Two or more pages answer the same question, or the page has no links and no conversions	0.5 to 2 days	Leaving near-duplicates in the index dilutes both pages and wastes crawl budget

The thresholds are ours, not universal constants, and we treat them as such: a 15% year-over-year click decline on a page with meaningful volume is enough to open a ticket, while a page with 40 sessions a month needs a commercial reason before it gets any editorial time. The 24-month fact rule is also operational rather than scientific — it is the point at which pricing pages, product names and regulatory references in our own library start to be wrong often enough to matter.

What the disposition table prevents is the most expensive habit in content operations: rewriting everything because rewriting feels productive. In the 18,400-refresh dataset, the median lift was 47.3%, which sounds like an argument for rewriting more. It is actually an argument for choosing better, because a median hides the pages that gained nothing.

III. The six-step rewrite SOP

Step one is to freeze the baseline. Before touching a word, export 16 months of Search Console data for the URL: clicks, impressions, average position, top queries, and the click-through rate by query. Sixteen months gives you a year-over-year comparison that survives seasonality, which is the single most common reason teams misdiagnose a normal seasonal dip as decay. Snapshot the current page too — a full-page capture, not a screenshot of the fold — because you will need it to prove what changed.

Step two is diagnosis, and it has exactly three branches. Content staleness: the page cites facts, prices or versions that are no longer true. Competitive displacement: the top five results now answer the query with something the page lacks, usually a comparison table, a decision framework or first-hand test data. Intent drift: the query itself now means something different, which you can see in the new top results rather than in your own analytics. Each branch has a different fix, and mixing them is why rewrites fail — a staleness fix will not recover a page that lost to a better comparison table.

Step three is the edit itself, and we hold to one constraint: every changed paragraph must be traceable to a diagnosis item. If the diagnosis says three claims are outdated, the release replaces those three claims, adds the source and date for each, and stops. Rewriting paragraphs that were not diagnosed adds risk without adding evidence, and it destroys the ability to attribute the outcome to a specific change.

Step four is the integrity pass: internal links in and out, canonical tag, hreflang if the article exists in three languages, image alt text, table markup, and the publication and modified dates. This is where most silent damage happens. A refresh that drops a canonical, or forgets to update a translated sibling, can lose more than the rewrite gains.

Step five is observation with a pre-declared window. We use 90 days, with checkpoints at 14, 30 and 60 days, and we declare in advance which metric decides success: clicks for a demand-recovery rewrite, assisted conversions for a commercial page, and citation appearances for a page that lives in AI answers. Declaring the metric afterwards is how teams end up celebrating impressions while pipeline falls.

Step six is archiving the decision. The ticket closes with the diagnosis branch, the diff summary, the metric result and the next review date. Without that record, the same page gets "refreshed" again in nine months by someone who cannot see what was already tried.

IV. What Tengence Insight adds: judging the rewrite on funnel outcomes

Traffic recovery is the easiest metric to move and the least useful one to stop at. A page can recover every lost click and still contribute nothing, because the recovered sessions may be researchers rather than buyers, or because the page's call to action leads to a form that nobody completes.

This is where Tengence Insight does the work that pageview dashboards cannot. It is an analytics platform for funnel, retention, path and audience analysis, and we use it to answer three questions about every rewrite. First, did the refreshed page change the mix of intents arriving at it — are the new sessions arriving on queries that historically converted, or on informational queries that never did? Second, did the downstream path change — do refreshed-page visitors reach pricing, documentation or a demo request at a higher rate than before? Third, did the rewrite cannibalise a sibling page — did total section conversions rise, or did one page simply take traffic from another?

Those three questions turn a content ticket into a measurable experiment with a hypothesis, a control and an outcome. It also changes how the team prioritises: a page that recovers 40% of its traffic but lifts demo requests from 6 to 11 per month outranks a page that recovers 60% of its traffic and changes nothing downstream.

The uncomfortable implication is that some high-traffic pages should not be refreshed at all. If the funnel shows that the page's visitors almost never convert and the topic has no strategic role, the honest disposition is retire-and-redirect, and the budget belongs to a page where a rewrite can change a decision.

V. Boundaries: four ways a good SOP still fails

Boundary one: refreshing for the date. Adding a new "last updated" line without changing substance is the most common fake refresh. Readers notice, and answer engines increasingly compare page versions rather than trusting timestamps. If nothing substantive changed, do not touch the date.

Boundary two: rewriting pages with no commercial role. A 20% traffic recovery on a page that has never produced a lead is a hobby. Portfolio-level decisions come before page-level craft.

Boundary three: ignoring the translated siblings. In a trilingual library, an English rewrite without the corresponding zh-hans and zh-hant updates creates three pages that disagree on facts — the exact condition that makes an answer engine quote the wrong version. Rewrite the set, not the file.

Boundary four: measuring too early. Ranking and click recovery after a rewrite usually takes weeks, not days; the 2026 refresh data reports a median lift of 47.3% across a portfolio, not per page on day three. Declaring failure at day ten guarantees that good rewrites get reverted.

FAQ

Q: Should I change the publish date when I update an article?
A: Only when the substance changed. Update the modified date as a by-product of real edits; do not use the date as the edit. If the only change is the timestamp, you have made a cosmetic edit and readers can tell.

Q: How do I know whether to rewrite or to write a new article?
A: Compare the search intent, not the keyword. If the current outline still answers the question people ask today, rewrite in place. If the intent has moved to a different question, build the new page and permanently redirect the old one.

Q: What is a realistic recovery expectation?
A: The 2026 portfolio study reports a median 47.3% traffic lift after a refresh, with rebuilds averaging 84.2%. Treat both as portfolio medians — individual pages vary widely, and pages with no commercial role should not be in the queue at all.

Q: How long should I wait before judging the rewrite?
A: Declare a 90-day window with checkpoints at 14, 30 and 60 days, and pick the deciding metric before the release. Judging at day ten mostly measures indexing latency.

Q: Do rewrites need to be done in all three languages?
A: Yes if the article exists in three languages. A fact changed in one language and left stale in two others produces contradictory answers, which is worse than a single outdated page.

About Tengence

Tengence builds the visibility and measurement layer for AI-mediated discovery. Tengence GEO combines GEO and SEO to make brands retrievable and consistently described across answer engines and classic search. Tengence Search is an enterprise site-search and retrieval platform that turns internal query logs into demand evidence. Tengence Insight is the analytics platform behind the funnel, retention and attribution numbers that decide whether content work paid.

Data Sources

Visionary Marketing, "Content Refresh ROI Statistics 2026: An 18,400-Page Refresh Impact Study" (June 2026) — decay curve by page age, median lift by tactic.
Ahrefs, "Content Decay: Why It Happens and How to Fix It" — mechanism-level definition of decay and its causes.
Clear Design, "Content Update Plan: 7 Proven Steps" (April 2026) — update cadence by content type, citing a Semrush analysis of 20% to 40% annual traffic loss.