Content Refresh: pruning, expansion, repurposing strategies

A comprehensive installation and audit reference for the content refresh discipline. Refresh is substantive content modification driven by decay signals or freshness needs. It is distinct from…

The Operational Standard for Updating Existing Content to Defend Rankings, Recover Decayed Pages, and Sustain AI Overview Citation

A comprehensive installation and audit reference for the content refresh discipline. Refresh is substantive content modification driven by decay signals or freshness needs. It is distinct from rewrite (major restructuring) and retirement (remove or 301 redirect). This document specifies the decay diagnostics that detect when refresh is required, the decision framework that chooses between refresh, rewrite, retire, and consolidate, the production workflow that executes refresh without breaking schema or violating helpful content guidelines, the ROI measurement frame that proves refresh value, and the audit rubric that scores refresh quality. Dual purpose: installation manual and audit document.

Cross stack implementation note: the code samples in this framework are written in plain HTML for clarity. For React, Vue, Svelte, Next.js, Nuxt, SvelteKit, Astro, Hugo, 11ty, Remix, WordPress, Shopify, and Webflow equivalents of every pattern below, see framework-cross-stack-implementation.md. For pure client rendered SPAs (no SSR/SSG) see framework-react.md. For Tailwind specific concerns see framework-tailwind.md.

1. Document Purpose and How to Use This Document

1.1 What This Document Is

This is the canonical operational reference for the content refresh discipline. Existing content does not stay performant indefinitely. Animalz 2018 documented that every published article moves through five lifecycle phases (spike, trough, growth, plateau, decline), with decline setting in for most evergreen content between months 12 and 24 after publication. Median publisher organic traffic fell 10 percent year over year in the first half of 2025 per Similarweb (news publishers down 7 percent, non news content sites down 14 percent). Ahrefs February 2026 measured a 58 percent CTR reduction on top ranking pages where Google AI Overviews appears, up from 34.5 percent in April 2025. Decay is the operating state, not the exception.

Refresh discipline is the response. HubSpot's historical optimization program (2 to 3 post updates per week at scale) increased monthly organic search views on refreshed posts by 106 percent and tripled monthly leads. Backlinko reported 260.7 percent organic traffic uplift on a single refreshed page within 14 days. Animalz documented an average 50 percent or greater pageview increase from systematic refresh. The 2026 addition: Ahrefs February 2026 found content updated in the last three months averages 6 AI Overview citations versus 3.6 for outdated pages. Refresh has shifted from a ranking maintenance practice to a citation defense practice.

Sections 4 through 12 cover decay diagnostics through citation defense. Sections 13 and 14 are audit rubric and maintenance schedule.

1.2 Three Operating Modes

Mode A, Install Mode. Establish refresh infrastructure. Sections 2 through 14 in order.

Mode B, Audit Mode. Evaluate existing refresh practice. Skip to Section 13.

Mode C, Hybrid Mode. Audit first, then install for failing items.

1.3 How Claude Code CLI Should Consume This Document

1. Section 2 client variables.
2. Section 4 decay diagnostics to identify candidate pages.
3. Section 5 decision framework to route to refresh, rewrite, retire, or consolidate.
4. Section 6 triggers and cadence to schedule work.
5. Section 7 production workflow per refresh.
6. Section 8 dateModified discipline and Section 9 schema preservation validation.
7. Section 10 ROI measurement.
8. Section 13 audit at engagement start and quarterly thereafter.
9. Section 14 reports.

1.4 Conflict Resolution Rules

1.5 Required Tools

GSC (Performance report, 28 day rolling comparison, AI Overview impression visibility since 2024); GA4 (engagement signals); crawler (Screaming Frog or equivalent) for sitemap, internal link audit, schema validation at scale; schema validator (Google Rich Results Test, Schema.org Validator); curl for substrate verification; spreadsheet for refresh queue, audit log, ROI tracking; IndexNow client (Bash script, Section 7.11) for push notification to Bing, Yandex, Naver, Seznam.

1.6 Relationship to Neighboring Frameworks

This framework is the operational standard for refresh itself. Broader content audit methodology lives in framework-contentaudit.md. Helpful content guidance for refresh substance and anti-date-manipulation rules lives in framework-hcs.md. YMYL refresh cadence requirements live in framework-ymyl.md. The AI Overview citation defense connection lives in framework-aioverviews.md. Information Gain framing lives in framework-infogain.md. This document is what those references point to when refresh specific operational guidance is required.

2. Client Variables Intake

# CONTENT REFRESH FRAMEWORK CLIENT VARIABLES

# --- Business and Site Identity (REQUIRED) ---
business_name: ""
primary_domain: ""
business_industry: ""
ymyl_classification: ""              # full_ymyl, partial_ymyl, lite_ymyl, non_ymyl
cms_or_stack: ""

# --- Content Portfolio Scale (REQUIRED) ---
total_indexable_pages: 0
total_content_pages_excluding_product: 0
oldest_content_publication_year: 0
pages_published_more_than_24_months_ago: 0
pages_published_more_than_12_months_ago: 0

# --- Current Refresh Practice (REQUIRED) ---
documented_refresh_cadence: false
trigger_based_refresh_model_in_use: false
refresh_changelog_practice: false
date_only_updates_currently_happen: false   # Honest answer. See Section 8.
last_audit_of_refresh_quality: ""            # Date or "never"

# --- Decay Baseline (REQUIRED) ---
pages_with_28d_traffic_decline_above_30pct: 0
pages_with_avg_position_drift_below_position_10: 0
pages_with_zero_clicks_28d_but_have_impressions: 0
pages_in_orphan_state: 0

# --- Routing Queue Counts (REQUIRED) ---
queue_refresh_count: 0
queue_rewrite_count: 0
queue_retire_count: 0
queue_consolidate_count: 0

# --- AIO Citation, Schema, E-E-A-T, Tracking (REQUIRED) ---
priority_queries_tracked_for_aio: 0
queries_currently_cited_in_aio: 0
queries_lost_aio_citation_in_last_90d: 0
schema_validates_clean_on_priority_pages: false
core_graph_present: false
faqpage_schema_mirrors_visible_qanda: false
authors_credentialed_with_visible_bylines: false
reviewers_credentialed_for_ymyl_pages: false
gsc_property_verified: false
ga4_property_verified: false
indexnow_key_provisioned: false
refresh_log_spreadsheet_location: ""

Refresh work routes to baseline frameworks when prerequisites fail. If schema_validates_clean_on_priority_pages is false work routes to framework-schema.md. If authors_credentialed_with_visible_bylines is false on a YMYL site work routes to framework-eeat.md. Refresh against a broken substrate produces no measurable lift.

3. What Content Refresh Is

3.1 Definition

Content refresh is substantive content modification driven by decay signals or scheduled review. Mandatory components: documented trigger (decay signal, schedule, or external event), content change addressing the trigger, dateModified update reflecting the change, changelog entry documenting what changed, post refresh validation (schema, substrate, internal links).

Refresh is distinct from three adjacent operations:

• Rewrite is major structural change: new heading hierarchy, new sections, substantial removal, sometimes new URL. Appropriate when the page is structurally mismatched to current intent or the topic has shifted enough that incremental edits cannot close the gap. URL preservation is typical; only change URL if the previous URL signals intent the rewrite contradicts.
• Retirement is removal: 410 (Gone) for content with no successor and no link equity worth preserving; 301 to a topical successor where one exists.
• Consolidation is merging multiple pages into one comprehensive page with 301 redirects from absorbed URLs. Appropriate when cannibalization is detected or separate thin pages would serve users better as a single resource.

3.2 Why Refresh Is a Distinct Discipline

Pre 2018 SEO treated refresh as an occasional maintenance task: republish a few pages a year, change the date, hope for ranking lift. That practice produces measurable harm in 2026 because Google's December 2025 Helpful Content guidance lists date changes without content changes as a Trust violation and because the content quality bar has risen.

Modern refresh is a production discipline. Animalz recommends review every 3 to 6 months and refresh of any page in decline for three consecutive months. HubSpot runs 2 to 3 refresh cycles per week at scale. Backlinko's 260.7 percent traffic lift came from refresh executed as a structured process with image updates, new case studies, and contextual backstory additions, not from a date change.

The discipline has four components: detection (when to refresh), routing (refresh, rewrite, retire, consolidate), execution (the production workflow), and measurement (did the refresh work).

3.3 The 2026 State: Why Refresh Discipline Compounds

Three numbers define the surface in 2026: AI Overviews appear on approximately 48 percent of all Google searches and over 70 percent on informational and how to queries (Surfer SEO December 2025, 173,902 URLs); 70 percent of AI Overview content changes when the same query is run twice and 45.5 percent of cited sources are replaced on regeneration (Ahrefs and Surfer 2025 studies); content updated in the last three months averages 6 AI Overview citations versus 3.6 for outdated pages (Ahrefs February 2026, 863,000 keywords).

Static unmaintained pages drift out of the AI Overview candidate pool over months. Refresh keeps a page in the pool. This is the citation defense case for refresh discipline; the classic ranking case (preventing rank decay) compounds it. A page that is refreshed substantively earns both. A page that is not loses both over a 12 to 24 month window.

3.4 Four Pillars Context

Refresh discipline operates across all four pillars: SEO (classic ranking decay defense), AEO (Answer Engine Optimization, featured snippet retention), AIO (AI Overview citation defense, framework-aioverviews.md), and GEO (Generative Engine Optimization, broader AI citation across ChatGPT, Claude, Perplexity, Bing Copilot, Meta AI, framework-aicitations.md). SEO weights query intent and link equity; AEO weights answer block freshness; AIO weights Information Gain and dateModified honesty; GEO weights authority recency. A single refresh that addresses all four is the production target.

4. Content Decay Diagnostics

Decay is detectable before it surfaces as a ranking drop. The diagnostic layer is what separates trigger based refresh from calendar based refresh. Seven decay categories follow, each with detection method.

4.1 Factual Decay

Statistics, dates, prices, regulations, product specs, named partners, or organizational facts have changed since publish. Factually outdated YMYL content is systematically demoted. AI Overview synthesis preferentially extracts from pages with current data.

Detection: manual review of numeric claims and dated references on priority pages; GSC query mining for current year queries the page no longer ranks for; spot checks against more current sources; a "facts that may change" inventory built at publish time (tax rates, prices, vendor counts, regulatory rules) with scheduled revisits.

4.2 Query Decay

Search intent for the target query has shifted. Example: "best CRM 2024" decays into "best CRM 2026" with different shortlist expectations. Query decay causes CTR drop without rank drop.

Detection: GSC Performance filtered to the page, CTR trend on top queries over rolling 28 day windows (steady rank with falling CTR is the signature); manual SERP comparison for topic, format, or framing shifts; coverage shift to competitor pages with a different angle.

4.3 Competitive Decay

Competitors published better content on the same topic. Classic ranking is relative; a page can lose rank purely because competitors improved.

Detection: manual comparison against the current top 10; Information Gain triage on the new top results; backlink delta where competitors earned significant new links; format mismatch where the SERP shifted format.

4.4 Algorithm Decay

A Google core update reweighted ranking signals in a way that disfavors the page. Core updates apply broadly; pages depending on downweighted signals lose visibility regardless of prior performance.

Detection: ranking drop coincident with a confirmed core update window (December 2025 core update ran December 11 to 29, 2025; March 2025 core update; June 2025 spam update); cross reference with framework-coreupdates.md; pattern matching where pages with similar profile drop together.

4.5 Authority Decay

Linking domains drop, author public activity drops, brand mentions slow, publishing organization loses prominence. Slow but cumulative.

Detection: referring domain tracking over rolling 90 day windows; author activity check on /authors/[slug]/ pages; brand mention velocity (target 5 to 10 third party unlinked mentions per quarter for a small business per framework-aioverviews.md Section 9.6); Wikidata or Knowledge Graph maintenance status.

4.6 Format Decay

Page format no longer matches the current SERP feature mix or AI Overview extraction pattern. Example: a prose only article on a comparison query when the SERP now shows a comparison table feature.

Detection: manual SERP inspection for features (featured snippet, PAA, AI Overview, Local Pack, video carousel, comparison table); page format inventory; mismatch flag where SERP feature does not match page format. Reduces feature eligibility and AI Overview citation probability (synthesis engine preferentially extracts from format matched pages, see framework-aioverviews.md Section 5.6).

4.7 Performance Decay

Load speed slipped, images broken, third party scripts crashing, layout shift introduced, mobile usability broken. Core Web Vitals is a ranking signal; sites with poor scores experienced 20 to 30 percent more severe traffic losses in the December 2025 core update per Raptive's December 2025 analysis.

Detection: PageSpeed Insights or Lighthouse audit per priority page; Core Web Vitals report in GSC; GA4 engagement metric trends; manual mobile load test in incognito.

4.8 Multi Decay Pages

Some pages exhibit multiple decay categories simultaneously. A 2022 published "best CRM" article may have factual decay (tools changed), query decay (intent shifted), competitive decay (better comparison pages exist), and format decay (prose only when SERP wants tables). Multi decay pages have higher refresh ROI because the refresh addresses several leverage points at once. Prioritize multi decay pages in the refresh queue.

4.9 The Decay Scorecard

For each priority page, compute a decay score. Each category scored 0 (no decay), 1 (early decay), 2 (significant decay). Total possible: 14.

5. The Refresh, Rewrite, Retire, Consolidate Decision Framework

Not every decayed page should be refreshed. The decision framework routes each page to the right operation. Four operations, specific routing rules.

5.1 The Decision Matrix

5.2 The 28 Day Stability Rule

A single day or week traffic drop is not a refresh trigger. The minimum window for routing decisions is 28 days. If a page has dropped 30 percent or more for 28 consecutive days versus the prior 28 day period, it enters the routing queue. Shorter windows produce false positives.

Exceptions: confirmed algorithm update windows (routing can begin within 7 to 14 days because the cause is known); clear factual errors that need immediate correction (a content fix, not a discipline refresh).

5.3 The Information Gain Test for Refresh Routing

When routing to refresh versus rewrite, ask whether the page has a defensible Information Gain contribution that survives the refresh. If yes, refresh preserves and surfaces it. If no, refresh has nothing to build on and rewrite is the correct route. Pages with original research, first hand experience, original case studies, or unique data sets refresh well. Pages that aggregated other sources without original contribution often need rewrite. See framework-infogain.md.

5.4 The Cannibalization Check Before Refresh

Before routing a page to refresh, check for cannibalization. If two or more pages target the same query cluster, refresh of one may not lift the cluster because internal competition splits ranking and citation. Detection: GSC Performance filter by query, view Pages tab; queries with multiple URLs above 100 impressions are cannibalization candidates. Internal topical overlap grouping by topic identifies pairs or clusters. If confirmed, route to consolidation (Section 11.6) before refreshing.

5.5 The Retirement Threshold

Retirement criteria, all four required: zero or near zero traffic for 90 consecutive days; no referring domains of value; topic no longer relevant to site primary focus areas; no internal link equity flowing through the page. A page meeting all four is a retirement candidate. A page meeting three is a refresh candidate (the missing criterion is the redemption path). Section 11.7 covers the retirement workflow.

6. Refresh Triggers and Cadence

6.1 The Trigger Based Model

Every refresh is documented with a trigger. Five trigger categories: scheduled review (calendar driven by content type, Section 6.2); decay signal (a Section 4 diagnostic has fired); external event (industry, regulatory, product, named partner change); algorithm update (confirmed Google core update window triggers review of pages in scope); manual flag (author, editor, or stakeholder identified an issue).

Trigger documentation goes in the refresh log entry (Section 7.2). "Republished because six months passed" is not a trigger. "Tax rates updated for fiscal year 2026" is.

6.2 Cadence by Content Type

6.3 The Quarterly Anchor Cadence

For most sites the quarterly anchor is the operational backbone. A quarterly anchor schedule (Q1, Q2, Q3, Q4) ensures every priority page receives at least one review per year and time sensitive content receives four. Industry research cited by theStacc (2026) indicates quarterly refresh cadence delivers 42 percent better results than annual refresh on aggregate organic traffic. The quarterly anchor is not a substitute for trigger based refresh; it is the floor under it.

6.4 When to Refresh More Aggressively

Three conditions break the quarterly anchor: active rank slide (page dropped average position more than 5 over 28 days, refresh immediately); lost AI Overview citation (tracked priority query no longer cites the page, apply framework-aioverviews.md Section 9.7 stability rule of two week re sample, then refresh if drop sustained); external event (industry data, regulatory change, product release, named partner change has invalidated content, refresh within days).

6.5 When to Refresh Less Aggressively

Two conditions slow refresh: page in active citation rotation (cited 60 percent or more on priority queries; refresh on longer interval and only when Information Gain can be added cleanly); canonical reference page stable two or more years with SERP intent unchanged (maintain factual currency, resist structural change).

6.6 The Anti Pattern: Calendar Only Refresh

Calendar only refresh ("every page updated every 6 months regardless of state") produces three failure modes: consumes capacity refreshing healthy pages; delays refresh of pages that decayed inside the calendar window; pushes operators toward date manipulation. Trigger based refresh anchored quarterly avoids all three.

7. The Refresh Production Workflow

Step by step process for executing a single refresh. Each step has a definition of done.

7.1 Pre Refresh Snapshot

Before touching the page, capture the baseline: GSC Performance 28 days prior (clicks, impressions, CTR, average position); GA4 landing page report (sessions, average engagement time, scroll depth, conversion rate); SERP screenshot for the page's top 3 priority queries; AI Overview citation status from manual sampling per framework-aioverviews.md Section 10.3; current backlink count and top referring domains; internal link count. The baseline is the comparison frame for ROI measurement (Section 10). Capture before any edit; do not edit later.

7.2 Trigger Documentation

Write the refresh trigger into the refresh log entry:

page_url: "https://example.com/quarterly-estimated-taxes-2026/"
refresh_id: "2026-05-14-001"
refresh_owner: "Amanda Emerdinger"
trigger_category: "external_event"
trigger_description: "IRS published 2026 estimated tax safe harbor amounts."
decay_categories_active: ["factual", "query"]
prior_28d_baseline: { clicks: 4218, impressions: 92456, avg_position: 4.7, ctr: 4.56 }
aio_citation_baseline: "Cited on 'when are quarterly estimated taxes due' weekly samples"

The log entry is the audit trail. It proves the refresh was triggered by something specific, not by a calendar tick.

7.3 Content Audit Against Current State

Read the page as if encountering it in search results today. For every numeric claim, citation, named source, dated reference, product mention, and example, ask: is this still accurate. Mark each confirmed, needs update, or needs removal. Then ask: what sections are outdated and need substantive update; what new sections should be added (emerged sub topics, edge cases, new data); what sections should be removed (outdated, no longer relevant, low value padding); what Information Gain can this refresh add that the original lacked. Section 7.5 expands the Information Gain check.

7.4 Schema and Substrate Pre Check

Before editing content, run the substrate validation the page must pass after refresh: H1 in first byte, lede present, schema present, FAQPage block present where FAQ exists. Use curl -A "GPTBot" -s URL with grep checks for <h1>, application/ld+json, and FAQPage. Record the pre refresh state of each. Section 9 specifies how to preserve through edit.

7.5 Information Gain Check

The most leveraged refresh investments add Information Gain. The ten categories from framework-infogain.md: original data, first hand experience, novel synthesis, contrarian finding, expert quote, specific case, comparative testing, primary source citation, methodological transparency, corrective claim. Ask whether the refresh adds at least one new Information Gain instance the page did not have before.

Refresh that only updates dates and tweaks wording fails. Refresh that adds a new data point, edge case, comparison, first hand observation, or expert quote passes. 2026 evidence: pages with substantive Information Gain refresh earn AI Overview citation at higher rates than pages with cosmetic refresh per Ahrefs February 2026 (content updated in last 3 months averages 6 AI citations versus 3.6).

7.6 Schema Audit and Update

Schema must match visible content after refresh. Section 9 covers preservation in detail. At minimum: Article headline matches visible H1 exactly; Article author Person @id resolves; Article reviewedBy Person @id resolves for YMYL; FAQPage Question name matches visible summary text exactly; FAQPage Answer text matches visible answer text; HowTo step name and text match visible content; HowTo step positions are sequential without gaps; dateModified updated to today; datePublished is not changed.

7.7 Internal Link Audit

Refresh is an opportunity to fix internal link debt. Inbound: does this page receive at least 3 internal links from topically related pages (per framework-aioverviews.md Section 5.8 and Princeton GEO SIGKDD 2024). Outbound on site: does the page link to related pages with descriptive anchor text. Outbound to authoritative sources: does the page cite primary sources, .gov, .edu, peer reviewed journals, standards bodies. Inline outbound citations boost AI Overview citation probability by approximately 30 percent per Princeton GEO study (SIGKDD 2024).

7.8 Author and Reviewer Credit Verification

Every refresh of YMYL content carries reviewer credit. If the refresh adds new YMYL claims, the reviewer Person node and the visible "Reviewed by" line must reflect the current reviewer. Visible byline and schema author must align. See framework-eeat.md Section 4 and framework-ymyl.md.

7.9 dateModified and Changelog Entry

dateModified updates to today only when the refresh meets the substantive standard from Section 8. Below the byline, add or update the visible changelog:

<details class="article-changelog">
  <summary>Last updated 2026-05-14, see what changed</summary>
  <ul>
    <li><time datetime="2026-05-14">May 14, 2026</time>: Updated 2026 safe harbor amounts to current IRS publication. Added new section on Q1 deadline coinciding with weekend.</li>
    <li><time datetime="2026-02-03">February 3, 2026</time>: Added comparison table of payment processors and fees.</li>
    <li><time datetime="2025-12-10">December 10, 2025</time>: Initial publication.</li>
  </ul>
</details>

The changelog is visible. It is also the audit trail that proves the dateModified change was substantive (Section 8). Sites without a visible changelog should add one before continuing refresh discipline; the changelog acts as both a user trust signal and a discipline enforcement mechanism (writing the entry forces the operator to articulate what changed).

7.10 Post Refresh Validation

After publish, run substrate check (curl -A "GPTBot" -s URL | grep -c "application/ld+json"), Schema.org Validator and Google Rich Results Test (zero errors required), a crawler pass to verify outbound links resolve, and an indexability check (curl -I URL | grep -i "x-robots-tag"). All must pass. If any check fails, the refresh is incomplete; fix before recording the refresh as done.

7.11 IndexNow Push

After refresh, push the URL to IndexNow capable search engines. IndexNow is a free open source protocol launched in 2021 and supported by Microsoft Bing, Yandex, Naver, Seznam, and Yep. Push notification produces up to 95 percent faster index inclusion per Bing's published data. Bash one liner: curl -X POST "https://api.indexnow.org/indexnow" -H "Content-Type: application/json" -d "{\"host\":\"example.com\",\"key\":\"${KEY}\",\"keyLocation\":\"https://example.com/${KEY}.txt\",\"urlList\":[\"${URL}\"]}". The key file must exist at https://example.com/[your_indexnow_key].txt containing the key as plain text. Provision once; reuse across all push events. Google does not consume IndexNow but the Bing push has measurable lift on Bing crawl velocity and Copilot citation recency.

7.12 Post Refresh Monitoring

Set a 14 day and 28 day check on the refresh log entry:

• Day 7. Re crawl with GSC URL Inspection, confirm Google has refetched. Run priority queries; spot check AI Overview status.
• Day 14. GSC Performance 14 day delta on the page; manual SERP sample on priority queries; note any rank movement.
• Day 28. Full 28 day delta versus pre refresh baseline. Apply Section 10 ROI measurement.

The 28 day delta is the formal outcome of the refresh and goes into the refresh log entry as the closing measurement.

8. dateModified Discipline

dateModified is the single most abused signal in content refresh. The discipline rules are simple and non negotiable.

8.1 The Substantive Update Rule

dateModified updates only when content has substantively changed. Google's December 2025 Helpful Content guidance lists "changing the date of pages to make them seem fresh when the content has not substantially changed" as a quality violation. The HCS framework reinforces this (see framework-hcs.md Section 9.6).

What constitutes substantive update, any one of: new section added covering a sub topic, edge case, or recent development (typically a new H2 or H3 with at least 100 words of original content); existing section materially expanded or rewritten (at least 30 percent of the section's word count is new with new factual content); data table or comparison materially updated; statistics or figures updated to current values where the change is non trivial; examples or case studies materially updated; conclusions or recommendations changed based on new evidence; new citations added that substantially shift the page's evidentiary base.

What does not constitute substantive update: typo fixes; minor wording changes for tone; outbound link target updates without content change; plugin or theme updates that touch markup but not content; schema additions without visible content change; sidebar or footer changes; image alt text edits; image swaps without content change.

If the edit set falls entirely in the "does not" category, dateModified does not change. Add a changelog entry describing the edit as a minor fix if useful for internal record, but do not update dateModified or schema dateModified.

8.2 The Visible dateModified Rule

dateModified appears visibly on the page, not only in schema. Format using semantic time elements with ISO 8601 datetime attribute:

<time datetime="2026-04-15">Published April 15, 2026</time>
<time datetime="2026-05-14">Updated May 14, 2026</time>

Both serve a purpose: schema for the synthesis engine and ranking systems, visible for users who use it as a trust signal.

8.3 Schema dateModified Rule

Article schema dateModified updates in lockstep with the visible date. Mismatches (schema says 2026-05-14, visible says 2025-08-01) are quality red flags. The schema validation step in Section 7.10 catches these.

8.4 datePublished Is Immutable

datePublished records when the content was first published; it does not change. Even for major rewrites that effectively replace the article, datePublished stays unless the URL changes (a new URL is a new article and gets a new datePublished). Manipulating datePublished is a quality violation.

8.5 The Changelog Pattern

Section 7.9 specified the changelog HTML pattern. The changelog has three functions: it documents substantiveness for quality auditing, it acts as a user trust signal, and it is a discipline enforcement mechanism. An operator writing a changelog entry must articulate what changed; if they cannot, the dateModified update should not happen.

8.6 Bulk dateModified Anti Patterns

Three patterns observed on sites with weak refresh discipline: database driven mass update (a CMS bulk action sets dateModified on hundreds of pages at once because the operator wants to "boost freshness"; detectable from server logs with identical timestamps); theme update side effect (a theme or plugin update touches every page, the CMS sets modified time on save, every page gets a new dateModified without content change; mitigate by disabling automatic modified time updates on cosmetic edits); and AI content rewording (a script rewords prose without changing substance; fails the substantive test and risks HCS detection).

8.7 When YMYL Is Audited

YMYL pages are audited at higher frequency. dateModified older than 12 months is treated by the synthesis engine as potentially out of date and risks demotion. The fix is real refresh: schedule substantive review at the semi annual cadence, document the trigger, update content, and update dateModified honestly. See framework-ymyl.md for category specific cadence.

9. Schema Preservation During Edits

Content edits frequently break schema. Five common breakage patterns and the preservation protocol.

9.1 Common Breakage Patterns

• FAQ question reworded. The <summary> text changes but FAQPage schema name is not updated. Mismatch removes FAQ rich result eligibility and reduces AI Overview citation probability per Frase.io 2025 (FAQPage schema matching visible content earns approximately 3.2 times the citation rate of mismatched schema).
• New H2 added with FAQ content. FAQPage schema not extended; the new question is invisible.
• Author byline changed. Visible byline updated but Article author Person @id still points to old author or is broken.
• HowTo step added without renumbering. Visible numbering updates but schema position numbers do not, causing duplicate or non sequential positions.
• reviewedBy added without reviewer Person node. Article reviewedBy references an undefined @id.

9.2 The Pre Publish Schema Validation Hook

Before publish, validate schema against visible content. Pattern using curl plus jq:

URL="$1"
HTML=$(curl -A "GPTBot" -s "${URL}")
SUMMARIES=$(echo "${HTML}" | grep -o '<summary>[^<]*</summary>' | sed 's/<\/\?summary>//g')
SCHEMA_NAMES=$(echo "${HTML}" | grep -A 1000 'application/ld+json' | jq -r '.. | select(.["@type"]? == "Question") | .name' 2>/dev/null)
diff <(echo "${SUMMARIES}") <(echo "${SCHEMA_NAMES}") && echo "MATCH" || echo "MISMATCH, fix before publish"

For sites with custom CMS schemas the extraction needs adjustment. The goal is automated detection of visible versus schema drift before content goes live.

9.3 The FAQPage Preservation Protocol

When editing FAQ content: edit the visible <details> and <summary> pairs; open the FAQPage JSON-LD source; update or add Question name to match new visible summary text exactly (including punctuation); update or add acceptedAnswer text to match new visible answer text exactly; remove Question entries for FAQs removed from visible content; run Section 9.2 validation; run Google Rich Results Test for zero errors.

9.4 The Article Schema Preservation Protocol

When editing article content: if H1 changed update Article headline; if lede changed update description if it referenced the lede; if author changed update author @id and ensure the new author Person node exists; if YMYL and reviewer changed update reviewedBy @id and ensure the reviewer Person node exists; update dateModified to today only if Section 8 substantive standard met; datePublished is immutable; run Schema.org Validator and Rich Results Test.

9.5 The HowTo Schema Preservation Protocol

When editing procedural content: inventory current visible steps with their order; for each new, removed, or reordered step update HowToStep position numbers in the schema; step name matches visible H3 or list item heading; step text matches the visible description; if totalTime changes materially, update; run validators.

9.6 The Author Schema Preservation Protocol

When updating an author: Author Person node exists with stable @id; /authors/[slug]/ page exists with ProfilePage schema (see framework-eeat.md Section 4.2); Article author references the Person @id; if author changed, verify the new author has a Person node and /authors/[slug]/ page before publishing under the new byline.

9.7 The Multi Page Schema Coherence Check

Site wide schema coherence is broken by piecemeal edits over time. Quarterly, run a schema coherence sweep with a crawler in schema validation mode, or with curl in a loop against a priority pages list. Pages without server rendered schema or with broken validation route into a remediation queue. The quarterly sweep is the maintenance counterpart to per refresh validation in Section 9.2.

10. Refresh ROI Measurement

Refresh that cannot be measured cannot be defended in budget conversations. The ROI measurement frame uses a 28 day pre versus 28 day post comparison.

10.1 The Baseline Frame

Before the refresh, snapshot: GSC Performance 28 days pre refresh (clicks, impressions, CTR, average position); GA4 landing page report 28 days pre refresh (sessions, average engagement time, conversion count and rate); AI Overview citation status across priority queries (manual weekly sample); featured snippet ownership on snippet eligible queries. Capture once at refresh time; do not edit later.

10.2 The Post Refresh Measurement Window

The 28 day post refresh window starts the day after publish. Capture the same metrics and compare: click delta percent, impression delta percent, CTR delta absolute and relative, average position delta absolute (improvement of 2.3 positions for example), conversion count delta absolute and percent, AI Overview citation rate delta, featured snippet ownership delta.

Industry benchmarks for context: HubSpot's historical optimization program produced 106 percent monthly organic view increase on refreshed posts. Backlinko reported 260.7 percent organic traffic lift on a 14 day case. Animalz reported 50 percent or greater pageview lift on average across refreshed content portfolios. Backlinko 2025 cited up to 146 percent lift on updates including structural and Information Gain improvements. Individual page outcomes vary widely; the portfolio average is the meaningful number.

10.3 The Cumulative Attribution View

Single page ROI is noisy. Roll up across the refresh portfolio: total refresh count last 90 days, aggregate click change across refreshed pages (28 day pre versus 28 day post), aggregate conversion change, refresh hours invested, lift per hour. The cumulative view answers the budget question: is refresh discipline producing more value than it consumes.

10.4 The 90 Day Confirmation Window

The 28 day post measurement is the headline number. The 90 day measurement is the durable number. Pages refreshed at high ROI on 28 day often regress part of the gain at 90 days (initial freshness boost fades), but durable refresh retains 60 to 80 percent of the 28 day gain at 90 days. Compute both and track which refresh types produce durable lift versus transient lift.

10.5 When Refresh Fails

If 28 day post refresh shows zero or negative click change, the refresh failed. Failure causes:

• Refresh was cosmetic, not substantive (Section 8 substantive test failed); the synthesis engine did not detect material change
• Refresh did not address the actual decay category (page had query decay; refresh addressed factual decay only)
• Refresh introduced schema breakage that was not caught in validation
• Refresh introduced format mismatch with current SERP
• External cause: competitor refreshed at the same time, algorithm update intervened, seasonality

The triage: rerun decay diagnostics (Section 4), identify what was missed, and execute a second pass refresh addressing the missing decay category. Two consecutive failed refreshes on the same page route the page to rewrite or retirement per Section 5 routing.

10.6 Monthly Refresh Roundup Template

# Refresh Roundup, {{MONTH YEAR}}

## Activity
Refreshes published, rewrites completed, retirements executed, consolidations completed.

## Performance, 28 day post versus 28 day pre baseline
Aggregate clicks, impressions, conversions lift. Pages with positive / neutral / negative lift.

## AI Overview Citation Movement
Priority queries cited pre versus post baseline; new gains; losses.

## Highlights
Top performing refresh; refresh that failed with root cause triage.

## Capacity
Refresh hours invested; lift per refresh hour.

The monthly roundup is the operational rhythm document. It feeds the quarterly audit (Section 14) and the annual portfolio review.

10.7 Metrics to Avoid

Three metrics that mislead: single day post refresh rank (daily rank fluctuates, use 28 day average position); total page views without traffic source segmentation (refresh impact must be measured on organic search specifically); time on page in isolation (GA4 average engagement time is more meaningful, and pair with scroll depth and conversion rate).

11. Bulk Refresh Operations and Content Retirement

Two adjacencies to per page refresh: bulk operations at scale and retirement.

11.1 The Refresh Queue Prioritization Function

For sites with hundreds or thousands of pages the refresh queue prioritization is a function of three variables: traffic value, decay severity, and Information Gain capacity.

priority = (current_28d_clicks * conversion_rate) * decay_score * information_gain_capacity

Where:

• current_28d_clicks is GSC click count over the last 28 days
• conversion_rate is GA4 conversion rate for the page over a 90 day window
• decay_score is the Section 4.9 decay scorecard total (0 to 14)
• information_gain_capacity is operator judgment 1 to 3 (1 = no clear gain possible, 2 = moderate, 3 = significant new contribution available)

Sort the queue descending by priority. The top of the queue refreshes first.

11.2 The Spreadsheet Workflow

For small to medium portfolios (under 1,000 pages) the refresh queue lives in a spreadsheet:

The spreadsheet feeds the weekly refresh stand up. Each refresh closes by updating the row with refresh date, post 28 day delta, and status (Done, Failed, Routed to Rewrite, Retired).

11.3 The Tooling Path for Larger Portfolios

CMS bulk actions (WordPress, Hugo, Next.js content sources all support bulk meta updates, use with caution; bulk dateModified updates are a Section 8.6 anti pattern unless every page in the batch genuinely received substantive change); custom Bash scripts (curl plus jq is sufficient for schema validation sweeps and substrate checks across hundreds of URLs, the Section 9.2 script generalizes); spreadsheet driven content publishing (for programmatic content, update source data and republish; track refresh per row in source data not per page).

11.4 Quality Control at Scale

The risk of bulk refresh is uniformity. Quality control measures: per page substantive justification documented in refresh log; variation in changelog descriptions; batch sizes small enough to maintain quality (10 to 30 pages per batch solo, 50 to 100 for teams); sampling at every tenth refresh for second eyes; schema validation on every page in the batch.

11.5 Staged Rollout

For large refreshes that touch a category at once (all pricing pages, all tax content for new fiscal year), stage the rollout: Day 1 refresh 10 percent of the batch and monitor 48 hours; Day 3 next 25 percent; Day 7 remaining 65 percent. Staging catches systematic issues (broken schema template, global content error) before they propagate site wide.

11.6 The Consolidation Workflow

When the audit identifies two or more pages targeting the same query cluster: identify the canonical page (highest traffic, strongest backlink profile, or best Information Gain); merge unique content from the secondary pages into the canonical, preserving Information Gain from all sources; update the canonical page's schema, internal links, and outbound link strategy to reflect the consolidated scope; 301 redirect the secondary URLs to the canonical; update internal links across the site that pointed to the secondary URLs; add the consolidated page to the refresh log with trigger "consolidation"; submit the canonical to IndexNow; monitor 28 day post consolidation metrics (expected lift is significant because cannibalization is resolved and the consolidated page carries combined link equity).

11.7 The Retirement Workflow

When the audit identifies a page meeting all four Section 5.5 retirement criteria: verify no internal links flow to the page from priority pages (remove or redirect those that do); verify no significant backlinks (if any exist, route to 301 instead of 410). For 410 (Gone): configure the web server to return 410 on the URL, remove from sitemap.xml, do not redirect, let the URL disappear from the index over the next few weeks. For 301 (redirect): identify the closest topical successor (do not redirect to homepage or generic category, which Google treats as soft 404 per Intero Digital 2025 guidance), configure the redirect, update the sitemap, remove the URL from internal navigation and related posts widgets, submit GSC URL Removal for sensitive URLs needing faster de indexing. Log retirement in refresh log with trigger "retirement," date, and disposition.

Common 410 vs 301 decision:

410 produces faster index removal and is the cleaner signal where the page should be erased. 301 preserves link equity where it exists.

12. Refresh and AI Overview Citation

The connection between refresh discipline and AI Overview citation is operationally significant in 2026.

12.1 The Citation Freshness Signal

Pages updated in the last three months average 6 AI Overview citations versus 3.6 for outdated pages (Ahrefs February 2026, 863,000 keywords). The synthesis engine weights freshness in candidate selection. Static unmaintained pages drift out of the candidate pool over months; substantive refresh keeps them in.

12.2 Information Gain Refresh as Highest Leverage

Refresh that adds Information Gain (new sub topics, recent data, edge cases, first hand observations, contrarian findings) is the highest leverage refresh for AI Overview citation. Date only refresh fails the synthesis engine's substantiveness detection. See framework-aioverviews.md Section 9.4.

12.3 The Regeneration Volatility Frame

AI Overview content changes 70 percent of the time on identical query re runs; 45.5 percent of cited sources are replaced on regeneration. A page cited last week and uncited this week is normal volatility, not a refresh trigger. The refresh trigger is sustained citation loss over 28 days or longer. See framework-aioverviews.md Section 9.7 for the over rotation anti pattern.

12.4 Refresh Cadence for Citation Pool Membership

More aggressive than classic SEO refresh cadence because citation defense demands more frequent substantive change than classic ranking defense.

12.5 The Cross Pillar Refresh

A single refresh execution serves all four pillars (SEO, AEO, AIO, GEO) when it does the following:

• Updates factual content for SEO query intent matching
• Updates answer first paragraphs for AEO featured snippet eligibility
• Adds Information Gain for AIO citation candidacy
• Maintains entity declarations and author authority for GEO citation across multi engine AI

The refresh production workflow in Section 7 is designed to produce this multi pillar outcome by default rather than requiring separate refresh passes per pillar.

13. Audit Rubric

13.1 First 90 Days Subset (Per Page)

For a new client engagement, these five items move refresh quality and ROI fastest. Audit and remediate before deeper rubric work.

A refresh passing these five is in the disciplined refresh practice. A refresh failing any one is not, regardless of other work.

13.2 Per Page Refresh Audit Rubric (Full)

Score 15. World class refresh: 13 or higher with zero F1 to F5 fails.

13.3 Site Wide Refresh Practice Rubric

Score 10. World class refresh practice: 9 or higher with zero critical fails on S1, S2, S5, S6.

14. Maintenance Schedule and Report Templates

14.1 Maintenance Cadence

Weekly. Refresh roundup; day 7 and day 14 checkpoints on prior refreshes; spot sample top 10 priority queries for AI Overview citation; GSC 28 day delta review on priority pages.

Monthly. Section 10.6 monthly roundup report; refresh time sensitive content per Section 6.2; review competitor pages entering AI Overview candidate pool; routing decisions for pages with sustained 28 day decline.

Quarterly. Section 13 audit; site wide schema coherence sweep per Section 9.7; recalibrate refresh queue prioritization weights; portfolio audit per framework-contentaudit.md; refresh evergreen content batch.

Annually. Full framework review; refresh underlying frameworks (HCS, Information Gain, AIO citation, schema); strategic review of refresh investment versus alternative content investments; portfolio level retirement and consolidation decisions.

14.2 Implementation Report Template

# Content Refresh Framework Implementation Report

**Site**: {{BUSINESS_NAME}}
**Implementation Date**: {{TODAY}}

## Summary
Total content pages audited; pages routed to refresh, rewrite, retire, consolidate; refresh queue prioritization deployed; refresh log system in place.

## Decay Diagnostics Findings
Counts by decay category (factual, query, competitive, algorithm, authority, format, performance); multi decay pages.

## dateModified Discipline Audit
Pages with date manipulation history; pages requiring changelog backfill; anti-pattern instances flagged.

## Schema Preservation Status
Priority pages with clean schema validation; pages requiring schema remediation; schema coherence sweep results.

## Refresh Production Workflow Deployment
Workflow documented; refresh log system live; IndexNow client provisioned; schema validation hook deployed.

## First Refresh Batch Outcomes
Refreshes completed; aggregate 28 day post versus pre delta.

## Sign-Off

14.3 Audit Report Template

# Content Refresh Framework Audit Report

**Site**: {{BUSINESS_NAME}}
**Audit Date**: {{TODAY}}

## Executive Summary
One paragraph assessment of refresh discipline health.

**Site wide score**: X/10
**Average per refresh score**: X/15
**Refresh practice maturity**: Mature / Developing / Absent

## Decay Portfolio Status
Pages with decay score 6 or higher; pages with decay score 10 or higher (rewrite candidates); pages meeting retirement criteria.

## Refresh Discipline Findings
Trigger documentation rate; substantive update compliance rate; schema preservation rate; changelog adoption rate.

## ROI Snapshot (last 90 days)
Aggregate click lift, conversion lift, refresh hours invested, lift per refresh hour.

## Critical Failures
List with remediation.

## First 90 Days Subset Findings
Per refresh table of F1 to F5 results.

## Sign-Off

End of Framework Document

v1.0. Created 2026-05-14. By ThatDeveloperGuy.

Content refresh discipline defends rankings, recovers decayed pages, and sustains AI Overview citation across the volatility window. The classic practice (republish every six months, change the date) is detected by Google's December 2025 Helpful Content guidance as a quality violation. The modern practice is trigger based, schema preserving, substantively documented, IndexNow pushed, and measured against a 28 day pre versus 28 day post baseline. Sites that adopt this discipline compound traffic, citation, and conversion gains over multi year windows. Sites that do not lose share across the four pillars regardless of how much new content they publish.

Apply this framework after framework-contentaudit.md (initial inventory) and in parallel with framework-hcs.md, framework-infogain.md, framework-aioverviews.md, and framework-schema.md.

Companions

• framework-contentaudit.md, initial content portfolio audit and decay portfolio mapping
• framework-hcs.md, Helpful Content System and the substantive update standard
• framework-infogain.md, Information Gain as the substance of high leverage refresh
• framework-aioverviews.md, AI Overview citation defense and the regeneration volatility frame
• framework-aicitations.md, multi engine AI citation across ChatGPT, Claude, Perplexity, Bing Copilot, Meta AI
• framework-coreupdates.md, Google core update timeline and recovery framework
• framework-eeat.md, E-E-A-T pillars and author or reviewer credentialing
• framework-ymyl.md, YMYL refresh cadence requirements
• framework-sqrg.md, Search Quality Rater Guidelines
• framework-schema.md, schema graph patterns and preservation protocols
• framework-internallinking.md, internal link audit during refresh
• framework-keywordresearch.md, query intent shift detection
• framework-gscanalysis.md, GSC Performance analysis for decay detection
• framework-ga4.md, GA4 engagement metrics for ROI measurement
• framework-contentfirst.md, substrate doctrine and curl validation
• framework-cross-stack-implementation.md, per stack equivalents of the refresh patterns
• framework-topicalauthority.md, topical authority and pillar page maintenance (scheduled, this Phase 1 dispatch)
• framework-saas-seo.md, SaaS specific refresh cadence (scheduled, this Phase 1 dispatch)
• framework-attribution.md, refresh attribution in multi touch journeys (scheduled, this Phase 1 dispatch)
• SEO-Search-Appearance.md, multi engine surface map
• SERP-Optimization.md, feature targeting playbook
• 14 tier Engine Optimization Stack, Tier 2 Content Optimization Layer