How to Refresh Old Blog Posts With AI Using Search Console Data

7 min read·3 October 2026

The quickest way to refresh old blog posts is to let Search Console choose which ones, then use AI for the slow parts of the rewrite while you supply the facts and the judgement. Compare the last three months with the three before, pick the posts losing clicks or sitting just off page one, and fix the specific thing each one is missing. Don't rewrite everything, and don't just change the date.

That's the whole method. The rest of this post is the detail: how to read the data, what to ask an AI tool to do, what it can't do, and how to tell whether the refresh worked.

How to spot content decay in Search Console

"Content decay" just means a post that used to bring in visitors from Google is slowly bringing in fewer. It rarely falls off a cliff. It slides.

To find it, open the Performance report, go to the Pages tab and compare two date ranges. Google's help page on comparing groups in the Performance report explains that when you compare two values, the table shows a Difference column for each row. Sort by the clicks difference and the slipping posts rise to the top.

Two settings make the comparison fairer:

• Compare like with like. Three months against the previous three, or this quarter against the same quarter last year if your trade is seasonal.
• Use weekly or monthly granularity. Google recommends this because it neutralises the "day-of-the-week" effect, so you're not comparing a busy Monday with a quiet Saturday.

Then click into each slipping page and switch to the Queries tab. You want to know which searches it lost, not only that it lost them.

Rule out the false alarms first

Not every drop means the post needs rewriting. Google's own guide to debugging drops in Search traffic lists several causes that have nothing to do with the words on the page:

• Seasonality. Check the post's main queries in Google Trends. If demand fell everywhere, your post didn't get worse; fewer people searched.
• Technical issues. A stray noindex tag or a broken page shows up in the Page indexing report, not in your prose.
• Small wobbles. For a small drop in position, such as 2 to 4, Google says it recommends "avoiding making radical changes if your page is already performing well".

What's left after that is your refresh list: posts where demand held up but your page lost ground.

Four kinds of page worth refreshing, and the fix for each

Different symptoms need different fixes. Treating them all as "rewrite it" wastes your time.

What Search Console shows

What it usually means

What to fix

Clicks and position falling, impressions steady

Competitors wrote a better answer

Fill the gaps: missing steps, outdated facts, weak examples

Ranking 4–20 for a query with decent impressions

Google half-trusts the page for that query

Add a section that answers that query directly

Page one, lots of impressions, few clicks

The result doesn't earn the click

Rewrite the title and meta description

Two of your posts swapping places for one query

Cannibalisation

Merge them, then redirect the weaker one

A quick word on terms. "Impressions" are how often your page appeared in results. "CTR" (click-through rate) is, in Google's words, the click count divided by the impression count. "Cannibalisation" is when two of your own pages compete for the same search, so neither does as well as one would.

The second row is the one most small blogs overlook. We've covered how to find those pages in our guide to striking distance keywords. If a query doesn't really fit any existing post, it may deserve a new one; our post on finding topic ideas in Search Console covers how to choose between a new post and an update.

How to update blog content with AI without making it worse

AI is useful here, as long as you give it the right material and a narrow job. "Make this post better" gets you a longer, blander version of the same post. A precise brief gets you something you can use.

Give the tool four things:

1. The current post, in full.
2. The queries the page ranks for, exported from Search Console with position and impressions.
3. The diagnosis from the table above, for example "position 9 for 'how long does a boiler service take', no section answers it".
4. What's changed in your trade since you wrote it: new prices, new regulations, a product you've dropped.

Then ask for specific outputs rather than a finished rewrite. Hypothetically, say you're a heating engineer in Cork with a 2023 post on boiler servicing that's slipped. A useful request would be: "List the questions in these queries that the post doesn't answer. Draft a 150-word section for each. Suggest three title options under 60 characters. Flag any sentence that states a price or a date."

That last instruction matters most. It turns the AI into a checker for the parts most likely to have gone stale.

What AI can't do here

Be clear about the limits before you start:

• It doesn't know what's true today in your business. It will happily keep a 2023 price or invent a plausible new one. Every number needs checking against a source you trust.
• It can't add first-hand experience. Google's guidance on helpful, people-first content asks whether content shows "first-hand expertise and a depth of knowledge". That comes from you: the job that went wrong, the question customers always ask.
• It can't see why a competitor outranks you. Read the top three results yourself. You'll spot the missing piece faster than a prompt will.

Refreshing posts with AI is not a problem in itself. Google's spam policies target scaled content abuse, which it describes as "many pages are generated for the primary purpose of manipulating search rankings and not helping users". Carefully updating a post that already helps people is the opposite of that. We go into more detail in what Google's guidelines actually say about AI content.

What to change, and what to leave alone

A good refresh is surgical. Keep what's working and fix what isn't.

Keep the URL. The page's history in Google is attached to it. Changing the slug to add "2026" throws that away for nothing.

Keep the sections that rank. If a page gets clicks for three queries and has slipped for a fourth, don't restructure the parts serving the first three. Add to the page; don't gut it.

Update the date only when the content has really changed. Google's helpful content page lists, as a warning sign, "changing the date of pages to make them seem fresh when the content has not substantially changed". If you did make substantial changes, show a visible "Last updated" date. Google's byline date guidance also suggests setting dateModified in your structured data and making sure the visible date and the structured date match. Most WordPress SEO plugins handle the structured data for you; in Strapi you may need to map the field yourself.

Fix the title if CTR is the problem. For low-CTR pages, the title and meta description are often the whole refresh. Change them, leave the body alone, and you'll know exactly what caused any difference.

Check your internal links. Old posts often point to pages you've since deleted or replaced. Also link from the refreshed post to anything newer you've written on the topic.

After you publish: recrawl, wait, measure

Once the update is live, ask Google to look at it again. In Search Console, paste the URL into the URL Inspection tool and request indexing. Google notes that crawling can take anywhere from a few days to a few weeks, and that requesting it several times won't speed anything up.

Then leave it alone. Google's traffic-drop guide says you'll likely want to wait a few weeks before checking whether your changes helped, and adds that there's no guarantee they will.

To measure fairly:

• Write down the date you published the refresh.
• Compare the 28 days after it with the 28 days before, filtered to that one page, using weekly granularity.
• Look at the specific queries you targeted, not just total clicks. If the striking-distance query moved from position 11 to 6, that's the refresh working, even if total clicks barely moved yet.

If nothing has changed after six to eight weeks, your diagnosis was probably wrong. Go back to the top three results and look again.

A content refresh strategy you can actually keep up

Most refresh plans fail for the same reason most blogs go quiet: they're too big. A realistic content refresh strategy for a small business looks more like this:

• Once a month, run the date comparison and pick the two or three posts with the biggest click losses that aren't explained by seasonality.
• Once a quarter, check for striking-distance queries and low-CTR pages.
• Keep a simple log: URL, date refreshed, what you changed, which query you were targeting, and the before-and-after numbers.

The log is the part people skip, and it's what makes this work. After six months you'll know which kind of fix works on your site, whether that's new sections, better titles or merged posts, and you can do more of that.

The core idea is simple: Search Console tells you which post to fix and what's wrong with it, AI speeds up the rewrite, and you supply the facts and experience that make it worth ranking. Your next step: open the Performance report, compare the last three months with the three before, and write down the three pages with the biggest drop in clicks.

Researched and drafted with Blog AI, checked against its sources, and edited before publishing. How the checks work