Skip to main content
Back to Blog
13 min readBrass-SEO Team

Content Maintenance AI Prompts

A published post doesn't stay finished. Every page on a site is either holding its ranking or quietly losing it, and the loss rarely shows up until months of traffic are already gone. Brass-SEO flags that decline from connected GSC and GA4 data before it costs real traffic. These eight prompts do the same diagnostic work from raw exports, for anyone maintaining an aging site by hand. Works with ChatGPT, Claude, Gemini, and any AI chat tool.

Quick Navigation


Content Decay Detector

A single monthly snapshot hides slow decline. This prompt reads a full trend export and flags pages losing traffic gradually enough that no one has noticed yet.

Read my Google Search Console trend data and flag pages in a slow, silent decline before it's visible in a single monthly snapshot:

Site: [site URL]
GSC Data: [paste 12-month trend export by page — clicks/impressions per month]
Pages To Check: [specific URLs, or say "all top 50 by traffic"]

For each page, calculate:
1. TREND DIRECTION — is the page's click trend up, flat, or down over the full period
2. DECAY RATE — if declining, what's the month-over-month percentage drop
3. DECAY STAGE — early (just started, under 10% cumulative loss), mid (10-30% cumulative loss), late (30%+ cumulative loss, someone probably already noticed)
4. LIKELY CAUSE — content aging (stats/dates out of date), algorithm update timing, new competitor, or cannibalization from a newer page on the same site

Rank all declining pages by combined traffic value and decay rate, so the biggest quiet losses surface first, not just the fastest droppers.

Finish with the 5 pages where catching this now, instead of in 6 months, saves the most traffic.

---
Prompt by Brass-SEO (brass-seo.com)
---

How to Use

  1. Export a 12-month trend by page from GSC Performance, not a single-period snapshot
  2. List specific URLs to check, or ask for all top 50 by traffic
  3. Paste in

Tips

  • A single month-over-month comparison hides slow decay — always use the full trend
  • Decay stage matters more than raw traffic loss — a big page in early decay is more urgent than a small page in late decay

View Markdown Version | Download YAML Format


Refresh / Merge / Prune / Delete Verdict

A page's fate should be one answer, not a menu. This prompt runs a single page through a decision tree in order and stops at the first rule that fits.

Run this one page through a decision tree and give me a single verdict — refresh, merge, prune, or delete:

Page URL: [URL]
Current Performance: [GSC clicks/impressions trend, GA4 engagement]
Page Age: [publish date or approximate age]
Content Topic: [what the page covers]
Similar Pages On Site: [any other URLs covering an overlapping topic, if known]

Walk the decision tree in this order:
1. Is another page on the site targeting the same primary query? If yes, lean MERGE.
2. Is the page still getting meaningful traffic but the information is outdated (stats, screenshots, pricing, dates)? If yes, lean REFRESH.
3. Is the page getting near-zero traffic and covering a topic with no ongoing relevance? If yes, lean DELETE (with redirect).
4. Is the page getting near-zero traffic but the topic still matters to the business? If yes, lean PRUNE FROM NAV but keep indexed, or REFRESH once.

Give ONE verdict, not a menu, with the specific reasoning that ruled out the other three.

---
Prompt by Brass-SEO (brass-seo.com)
---

How to Use

  1. Pull the page's GSC and GA4 performance
  2. Note the page's approximate publish date
  3. List any other pages on the site covering an overlapping topic

Tips

  • A verdict that hedges between two options isn't a verdict — push for one answer per page
  • Run this on your lowest-traffic 20 pages first, where cleanup decisions matter most

View Markdown Version | Download YAML Format


6-Month Re-Optimization Calendar Builder

Great pages don't stay great without attention. This prompt turns a top-pages list into a recurring calendar so winners get revisited on a schedule instead of by accident.

Build a recurring re-optimization calendar for my site's top pages so winners don't quietly decline:

Site: [site URL]
Top Pages: [list your top 10-20 pages by traffic or revenue]
Publish/Last-Updated Dates: [for each page, if known]
Team Capacity: [how many refreshes per month realistically]

Build a 6-month calendar that:
1. Schedules each top page for a check-in at least once every 6 months
2. Front-loads pages that are already showing early decay signals (if provided) or are furthest past their last update
3. Spaces refreshes evenly against stated team capacity — don't front-load month 1 and leave months 4-6 empty
4. Distinguishes a QUICK CHECK (verify stats/dates still accurate, 15 minutes) from a FULL REFRESH (rewrite outdated sections, 1-2 hours)

Output as a month-by-month table: month, page, check type, and the one thing most likely to need updating on that page.

---
Prompt by Brass-SEO (brass-seo.com)
---

How to Use

  1. List your top 10-20 pages by traffic or revenue
  2. Note the last-updated date for each, if known
  3. State how many refreshes your team can realistically do per month

Tips

  • A calendar that's never followed is worthless — set capacity honestly, not aspirationally
  • Feed pages flagged by the Content Decay Detector to the front of the schedule

View Markdown Version | Download YAML Format


Content Pruning Candidate Finder

Cutting dead weight can lift a whole site, but only when the cut preserves whatever equity the page still holds. This prompt checks backlinks before recommending removal.

Find pages worth pruning without losing the SEO equity they've built:

Site: [site URL]
GSC Data: [12-month traffic export for all indexed pages]
Site Section Context: [what parts of the site are core vs. peripheral]

Flag pages as PRUNING CANDIDATES if they meet BOTH:
1. Near-zero organic traffic over the full period (define your own threshold, default to under 5 clicks/month)
2. No meaningful internal links pointing to them from higher-traffic pages (orphaned or near-orphaned)

For each candidate, do NOT just recommend deletion. Instead determine:
- Does it have any backlinks? (check before removing — equity worth preserving via redirect)
- Is there a live, related page it should 301 redirect to?
- Is it safe to fully remove with no redirect (truly dead topic, no backlinks, no equity)?

Output a table: URL, traffic, backlinks (yes/no if known), recommended action (redirect-to-X, safe-delete, or keep-and-refresh-instead), and reasoning.

---
Prompt by Brass-SEO (brass-seo.com)
---

How to Use

  1. Export 12 months of GSC traffic for all indexed pages
  2. Note which sections of the site are core vs. peripheral to the business
  3. Paste in

Tips

  • Always check backlinks before deleting — a zero-traffic page can still carry link equity worth redirecting
  • Pair with the Redirect Mapping Assistant once candidates are confirmed

View Markdown Version | Download YAML Format


Keyword Cannibalization Finder

Two pages competing for the same query split each other's authority instead of adding up. This prompt separates true cannibalization from acceptable overlap.

Find pages on my site competing against each other for the same search query:

Site: [site URL]
GSC Query Data: [paste query-level data showing which pages rank for which queries]

For each query where more than one page on the site receives impressions:
1. List the competing pages and their individual position/clicks/impressions for that query
2. Determine whether this is TRUE cannibalization (both pages target the same intent, splitting authority) or ACCEPTABLE overlap (pages serve genuinely different sub-intents or funnel stages)
3. For true cannibalization, recommend the fix: merge into one page, canonical tag pointing to the stronger page, or differentiate the weaker page's target query entirely

Rank the cannibalization cases by combined lost traffic potential — how much a single unified page might outrank the sum of the two split pages.

---
Prompt by Brass-SEO (brass-seo.com)
---

How to Use

  1. Export GSC query data with the pages ranking for each query
  2. Paste in
  3. Review each flagged case before acting — not every overlap is a problem

Tips

  • Two pages ranking for the same query isn't automatically a problem — check for genuinely different intent first
  • Re-run after any site restructure, since cannibalization often reappears as new content gets added

View Markdown Version | Download YAML Format


Redirect Mapping Assistant

A merge or prune decision is only half finished without a redirect target that actually fits. This prompt scores candidate destinations before you commit to one.

Map a page being merged or pruned to the correct consolidation target so I don't lose SEO equity:

Page Being Removed: [URL]
Reason For Removal: [merge / prune / outdated and no longer relevant]
Candidate Target Pages: [list 2-4 URLs that might be a good redirect destination]
Content Overlap: [briefly describe what the removed page covered vs. what each candidate covers]

For each candidate target, score it on:
1. TOPICAL MATCH — how closely the target actually covers the same query/intent
2. TRAFFIC HEALTH — is the target page itself healthy, or would redirecting into it drag it down
3. USER EXPERIENCE — would a visitor following the old URL land somewhere that still answers their original question

Recommend the single best redirect target, or say clearly if none of the candidates are close enough and a redirect to the homepage/category page is more honest than forcing a bad match.

---
Prompt by Brass-SEO (brass-seo.com)
---

How to Use

  1. Confirm the removal decision first, using the Refresh/Merge/Prune/Delete Verdict prompt if unsure
  2. List 2-4 candidate pages that might serve as the redirect target
  3. Describe the content overlap between the removed page and each candidate

Tips

  • A forced redirect to a poorly matched page can hurt the target page more than losing the source page helps
  • Test the final redirect chain — redirects that point to another redirect lose equity

View Markdown Version | Download YAML Format


Refresh vs. Rewrite Assessor

Not every decline calls for a rewrite, and not every stale page can be saved by updating a stat. This prompt separates the two.

Decide whether this page needs a light data refresh or a full rewrite:

Page URL: [URL]
Page Content: [paste the current content, or a summary of it]
Current Performance: [GSC clicks/impressions trend]
What's Changed Since Publish: [any known product, pricing, or industry changes relevant to this topic]

Assess against these criteria:
1. STRUCTURAL SOUNDNESS — is the heading structure, argument, and format still the right shape for the topic, or has the format itself aged out
2. FACTUAL DRIFT — how much of the content is now inaccurate (stats, pricing, screenshots, dates, product features)
3. COMPETITIVE GAP — does the page still cover the topic as thoroughly as what currently ranks above it
4. TRAFFIC SIGNAL — is the decline (if any) severe enough to suggest the page needs more than a data update

Give a single verdict: QUICK REFRESH (update stats/dates/screenshots only, structure is fine) or FULL REWRITE (structure or depth no longer serves the topic), with the specific reasoning.

---
Prompt by Brass-SEO (brass-seo.com)
---

How to Use

  1. Paste the page's current content or a close summary
  2. Pull its GSC performance trend
  3. Note anything specific that's changed since publish

Tips

  • A page with a sound structure rarely needs a full rewrite even if every stat is stale — swap the facts, keep the shape
  • Check what currently outranks the page before deciding — competitive depth is often the real gap

View Markdown Version | Download YAML Format


Evergreen vs. Time-Sensitive Classifier

Most "evergreen" posts hide a stealth expiration date. This prompt flags which posts will age, what will go stale first, and how often to check.

Classify my published posts by how much upkeep each one actually needs:

Posts: [list post titles/URLs, or paste a full post list]
Post Content Summary: [brief note on what each covers, if not obvious from title]

Classify each post into:
1. EVERGREEN — no dates, prices, or version-specific facts; the advice holds regardless of when it's read
2. TIME-SENSITIVE — contains specific stats, prices, product versions, or "as of [year]" framing that will age
3. SEASONAL — tied to a recurring event or period (a year-in-review, a seasonal guide) that needs refreshing on a schedule, not just when it decays

For TIME-SENSITIVE and SEASONAL posts, note the specific element that will go stale first (a stat, a price, a screenshot) and a suggested recheck interval.

Output as a simple table: post, classification, stale-first element (if applicable), recheck interval.

---
Prompt by Brass-SEO (brass-seo.com)
---

How to Use

  1. List your published posts, titles or URLs
  2. Add a brief content summary for any where the topic isn't obvious from the title
  3. Paste in

Tips

  • Most "evergreen" posts actually contain a stealth time-sensitive fact — a price, a version number
  • Feed the recheck intervals straight into the 6-Month Re-Optimization Calendar Builder

View Markdown Version | Download YAML Format


Frequently Asked Questions

How is this different from just checking Google Analytics occasionally?

These prompts look at trend data over a full period, not a single snapshot. A page can lose 5% of its traffic every month for a year and still look fine in any one month-over-month comparison — the full trend is what exposes it.

Should I run all 8 prompts on every page?

No. Start with the Content Decay Detector across your top 50 pages, then run the Refresh/Merge/Prune/Delete Verdict only on the specific pages it flags. The other prompts are for the specific decision each verdict points to.

Don't delete it outright. Use the Redirect Mapping Assistant to find a live page worth consolidating it into, so the backlink equity carries over instead of disappearing with a 404.

How often should this whole maintenance cycle run?

Every 6 months for a full pass, with lighter monthly checks on pages already flagged as declining. Brass-SEO's Declining Pages report surfaces this on an ongoing basis from connected GSC and GA4 data.


More Prompts

See the full decision framework in Refresh, Merge, or Delete: A Decision Tree and the pillar page at The Aging Site: Content Maintenance. Looking for more prompts? Check out the full AI Prompts Guide or browse all prompts on GitHub.

Ready to try Brass-SEO?

Get AI-powered SEO insights from your Google Search Console and Analytics data.