Content Audit and Refresh: Using AI on Existing Work
A page needs a refresh. You already know that — the Brass-SEO Declining Pages report flagged it, or you ran it through the refresh, merge, or delete decision tree and landed on a verdict. The harder question sits one step later. How do you actually get an AI tool to help rewrite three-year-old content without the result reading like a patch bolted onto the original page?
That question is what this post covers, not the one before it. Brass-SEO is built by Copper Sun Content and Creative, LLC, the same company behind Copper Sun AI, a separate product for marketing teams. Copper Sun AI published a technique for this exact step: Content Audit and Refresh Using AI. Copper Sun AI runs $1,500 a month for five seats, built for teams running full campaigns, not a solo owner reading a $25-a-month SEO tool's blog. The technique in that post does not need the product it came from. It works with any AI chat tool, including whichever one you already have open.
This post narrows to the mechanical part: what to paste in, what kinds of decay the comparison actually finds, and where a person still has to make the call.
Quick Navigation
- The Decision You Already Made
- The Technique Behind This Post
- Three Kinds of Decay AI Comparison Finds
- What to Feed the AI
- What AI Does Not Decide
- A Walkthrough Example
- Where This Fits With Brass-SEO
- Frequently Asked Questions
The Decision You Already Made
Brass-SEO's content-maintenance framework already answers whether a page needs a refresh, a merge, or a deletion, and this post assumes that verdict is done. If you have not run that step yet, start there rather than here.
The Refresh, Merge, or Delete decision tree sorts a page into one of four fates using its Google Search Console impressions. The full Content Maintenance pillar page covers the wider system, and the audit-your-content-keep-fix-delete framework and the 30-minute content refresh workflow cover the diagnostic side of finding which pages slipped and why. Brass-SEO's content maintenance AI prompts collection includes a Refresh vs. Rewrite Assessor that decides how big a fix a page needs before you start.
Once one of those tools says refresh — not merge, not delete, not a full rewrite — the technique below is what actually pairs an AI tool with the page and gets you a draft.
The Technique Behind This Post
The technique pairs the old page with the current top search results for its target query, then asks an AI tool to sort the gap between the two into three categories.
Copper Sun AI published the approach as a single workflow. Load the existing post and the current top-ranking pages for the same query into an AI tool, then let the model compare them side by side. Source: Content Audit and Refresh Using AI, Copper Sun AI blog. The output of that first pass is a gap analysis, not a rewrite. The model tells you where the page fell behind. A person decides what to do about each gap before any new text gets drafted.
That order matters. Skip straight to "rewrite this page" and the AI has no baseline for what actually changed since publish — it will happily invent improvements to a page that was already fine. The comparison step is what keeps the output tied to something real: your page, against what currently outranks it, on the query you actually care about.
Three Kinds of Decay AI Comparison Finds
A side-by-side comparison of an old page against current top results surfaces three distinct kinds of decay: facts that changed, coverage a competitor added, and search intent that shifted since publish.
Factual staleness. Stats, prices, screenshots, and dates that were accurate when you published and are not anymore. This is the easiest decay to fix and the easiest for an AI tool to catch, because it just needs to flag anything time-bound in your text and check whether the current top results say something different.
Coverage gaps. A section a competing page added since your last update — a subtopic, a data point, an angle — that now sits in the results without you. Not every gap is worth filling; more on that below.
Structural shift. The search intent behind the query changed. A page written as a definition now competes against comparison tables or step-by-step guides, because that is what searchers expect for that query today, even though nothing about your page's facts is wrong. This is the hardest decay to spot manually and the one an old-page-versus-current-results comparison is best suited to surface, since it shows up as a format mismatch, not a factual error.
What to Feed the AI
The comparison needs three inputs: the full text of the existing page, the top three to five current results for its main query, and the query itself, named directly rather than summarized.
Pull the page's actual content, not an outline. An AI tool needs the real sentences to catch what is missing; a bullet summary already discards the detail the comparison depends on. Search the target query yourself and copy the visible content from the top three to five organic results, skipping ads and your own page if it still ranks. Name the query directly in your instructions. A vague topic label lets the comparison drift toward generic advice instead of the specific gap between your page and what currently outranks it.
Ask the AI to sort what it finds into the three decay categories above, and to rank which are the highest-value fixes rather than listing everything it notices. A model that returns twenty flagged items with no ranking has not finished the job — narrowing the list to what matters is still a person's call, which is the next section.
What AI Does Not Decide
AI can compare a page against its competition and draft updated sections, but it does not decide whether a coverage gap is worth filling or is just competitor mimicry. That judgment stays with a person.
As Copper Sun AI's post on the technique puts it: "AI handles the structural and comparative work in a content audit. It doesn't handle the judgment calls." In practice, that means filtering the flagged gaps against your own audience before you act on any of them. A competitor covering a subtopic your readers never search for is not a gap worth closing — it is noise the comparison surfaced anyway. Check each flagged item against your own GSC query data. If searchers are not asking that question in your data, adding the section serves the comparison, not the reader.
Once you have filtered the list to gaps worth fixing, the AI can draft the expansion for each one: updated facts, a new subsection, a rewritten paragraph that matches the current search intent. Read the draft the way you would edit a freelancer's copy. Cut anything that does not sound like your business. Verify every fact the AI added is actually current — do not take its word for a statistic or a date. Confirm no new claim needs a citation you do not have.
The last step is not optional. Update the publish date, add an internal link or two from related pages, and request reindexing in Google Search Console. The 30-minute content refresh workflow covers that step in detail — a rewritten page Google has not recrawled has not actually changed as far as search results are concerned.
A Walkthrough Example
A refresh candidate that is losing clicks while holding impressions is the clearest case for this technique, because the page still ranks and only needs to close a specific gap, not start over.
Say a page has held its impressions in Google Search Console for months but clicks have slipped, and the decision tree above already marked it refresh rather than merge or delete. Pull the page's current text and search its main query. Copy the visible content from the top three or four results, note which query you searched, and hand all three pieces to an AI tool with instructions to compare them and sort the differences into the three decay categories above.
Suppose the comparison flags three things: a statistic from two years ago that current top results have since updated, a comparison table the top-ranking competitor added that your page never had, and a subheading structure that reads as a general overview while every competing result now answers the query as a direct how-to. The stat is an easy fix. Add the table only if your own query data shows people are actually comparing options. The subheading structure is the bigger job — the kind of restructuring the Refresh vs. Rewrite Assessor prompt is built to catch before you start.
Where This Fits With Brass-SEO
Brass-SEO does not draft page rewrites. It identifies which pages have the refresh signal, impressions holding while clicks or position decline, so this technique has a specific target list instead of a guess.
Ask Brass-SEO which pages are losing traffic, or use the Brass-SEO Declining Pages button, and you get the candidates this technique is meant for. Brass-SEO reads your connected Google Search Console and Google Analytics 4 data and surfaces the pages worth this kind of attention; the comparison-and-draft work above is what you or an AI tool does once Brass-SEO has named the page. Start a three-day free trial to find your own refresh candidates, or read the full system at Content Maintenance.
Frequently Asked Questions
Do I need Copper Sun AI to use this technique?
No. The technique — pairing an old page with current top results and asking an AI tool to sort the gap into categories — works with any AI chat tool. Copper Sun AI is where the approach was first published, and it is a $1,500-a-month platform built for marketing teams running full campaigns, not the audience of this blog. A free-tier AI chat tool can run the same comparison.
How is this different from Brass-SEO's Refresh vs. Rewrite Assessor prompt?
The Refresh vs. Rewrite Assessor, part of the content maintenance AI prompts collection, decides how big a job a page needs before you touch it: a light data update or a full rewrite. This post assumes that decision is already made and covers what happens next — the actual comparison against current search results and the draft that follows.
How many competing pages should I paste in for the comparison?
Three to five current top-ranking results for the page's main query is enough. Fewer than that and the comparison misses patterns that show up only when you look across several results; more than that adds bulk without adding much signal, since the top few results usually cluster around the same coverage and format.
Will an AI-drafted rewrite hurt my SEO?
Google's Search Central guidance on AI-generated content states that Google rewards high-quality content however it is produced. The real risk is publishing an AI-drafted section before you have verified it. See Is Your AI Content Helping or Hurting SEO? for the fuller answer. Treat this technique's output as a first draft that needs fact-checking and a voice pass, not a finished page.
What if the comparison finds nothing worth fixing?
That is a useful result, not a failed one. If a page holding impressions but losing clicks shows no factual staleness, no coverage gap, and no structural mismatch against current top results, the problem may be somewhere the comparison cannot see — a title tag, a slow page, or a technical issue. Check the 30-minute content refresh workflow for the other places to look.
Brass-SEO connects to your Google Search Console and Google Analytics 4 data and flags the pages that need this kind of attention. Start your free trial and ask which pages are losing clicks.