Updating Buyer Guides With AI: When to Refresh, When to Rewrite and When to Leave It Alone
An affiliate site with a year of guides on it has a maintenance problem that grows with the site: picks go out of stock, products get replaced, a guide that ranked in March has slipped by September. AI tools make the fix cheap, and that creates a new way to go wrong, which is refreshing pages that were fine and losing the parts of a page that made it rank.
This post is a decision table for old guides by symptom, what an automated refresh can safely change and what it must not, how to keep the URL and the parts that earned, and the short edit that follows every refresh. It's the AI-specific companion to the updating old posts post, which covers finding the pages that need it.
The short answer
Match the symptom to the action: dead or stale picks get a refresh (same URL, new products), a page targeting the wrong keyword gets a rewrite, a page that's current and ranking gets left alone.
A refresh may change products, sections, the table and the FAQ; it must not change the URL, and it should carry over the title if it ranks and your judgement lines. Keep those lines in a note and paste them back; ten minutes per guide.
The decision table
| Symptom | Action | Why |
|---|---|---|
| One or two picks dead or out of stock | Refresh | The page is right; the products aren't |
| Top pick superseded by a newer model | Refresh | Same; and the verdict needs re-writing after |
| Rankings slid, impressions holding | Refresh, then re-check in a month | Google is still showing it; a current page often recovers |
| Rankings slid, impressions gone | Check the keyword first | If the query died, no refresh helps; if a competitor took it, refresh and deepen |
| Page targets the wrong keyword or intent | Rewrite (or redirect) | The page's identity is wrong, not its products |
| Intro and verdict outdated, picks fine | Edit by hand | Two paragraphs; a refresh is overkill |
| Picks current, rankings stable | Leave it alone | Every regeneration is a chance to lose something |
What a refresh can safely change
- The product list. Re-pulled from live data: dead picks out, current ones in, the same count.
- Product sections, pros and cons. Rewritten from the new listing data.
- The comparison table. Rebuilt for the new picks.
- The FAQ, where the questions referred to products that are gone.
- The meta description, if the picks it mentioned changed.
- The date shown on the page, if the page shows one and the content changed.
What it must not change
- The URL. The one thing that holds the page's history. A refresh that publishes to a new URL is a new page competing with the old one.
- The title, if the page ranks. Google has associated the title with the rankings; change it only if it's wrong.
- The keyword and intent. That's a rewrite, and it's a different decision.
- Your judgement lines. The opening in your words, the verdict, the one sentence per product. The editing post describes them; a regenerated page doesn't have them unless you put them back.
- Links into the page from elsewhere. Nothing to do here, as long as the URL stays.
Keeping what earned
The honest limitation of any regenerate-from-data tool is that the page it produces is new. The sections are new text from current listings, which is the point, and the sentences you added last time are gone, which isn't.
The practical answer is a note per guide: the opening paragraph you wrote, the verdict in your words, and the judgement line for each pick that's still on the page. Three or four short paragraphs. After a refresh, paste them back where they went, update the verdict if the top pick changed, and the page has its products from today and its voice from you.
If a tool keeps the title and URL and updates the live post in place, the URL side is handled. The judgement side is always yours.
The edit after the refresh
- Confirm the URL and title are unchanged (or changed on purpose).
- Read the new top pick's section. If the top pick changed, the verdict has to say why, in your words.
- Paste back the opening and the judgement lines for picks that survived; write new ones for picks that arrived.
- Check the five specs the specs post lists on the new products.
- Check internal links still point at live pages, and the hub still lists this guide.
- Note the date in your tracking sheet and leave the page alone for a quarter.
About ten minutes. The refresh itself takes a few minutes of machine time; the ten minutes are what stops a refreshed page being a worse page.
When to rewrite instead
A rewrite is for a page whose problem is what it's about. The keyword has no buying intent, the page is a guide where the results want a single review, two of your pages target the same query, or the category has changed so much that the old structure doesn't fit.
In those cases regenerate for the right keyword, publish at the old URL if the intent is close enough to keep the history, and 301 to a better page if it isn't. The decision is the one in the updating post's table; AI changes the cost of doing it, not whether to.
Doing it with AffSERP
AffSERP's Refresh option on a published guide re-pulls the products from Amazon, regenerates the guide from the current data, and updates the existing WordPress post in place, keeping its URL and publish date, so the identity of the page is preserved automatically.
What it doesn't preserve is the hand edits inside the old content, which is why the note-per-guide habit above exists: refresh, open the editor, paste the lines back, publish. The quarterly check that decides which guides to refresh is the monthly Search Console routine, and a guide that passes it is left alone.
FAQ
Should I update old affiliate posts with AI?
For the product layer, yes: re-pulling products, replacing dead picks and rewriting sections from current data is what a tool does well and a person does slowly. For the parts that earned the page its rankings (the URL, the title if it ranks, your verdict and opinions), no; those are kept or re-applied by hand after the refresh.
What's the difference between a refresh and a rewrite?
A refresh keeps the page's identity (URL, publish history, keyword, structure) and replaces the product content with current data. A rewrite changes what the page is about or how it's structured, which is a new page in all but URL. Refresh when the picks are wrong; rewrite when the keyword or intent was wrong.
Does regenerating a guide lose my edits?
In most tools, yes: a regenerated page is a new page from current data, so the opening, verdict and judgement lines you added are replaced. Keep those in a note per guide (three or four sentences), and paste them back after the refresh. It's five minutes, and it's why the refresh is a job rather than a button you press and forget.
How often should buyer guides be refreshed?
Check quarterly; refresh when the check finds something. The triggers are a dead or out-of-stock pick, a top pick that's been superseded, a comparison table with a replaced product, or a ranking that's slid while impressions hold. A guide with current picks and stable rankings isn't refreshed on a schedule; it's left alone.
Keep reading
