Content Refresh and Decay Recovery

On any site with more than fifty published pages, updating what exists usually returns more than publishing something new. The page already has links, history and index presence. You're moving it from position 11 to position 4, not from nothing to something.

The hard part isn't the rewriting. It's picking the right pages and correctly diagnosing why each one fell.

Step 1: Find the candidates

Four queries against Search Console, in order of return.

Striking distance. Pages averaging position 5 to 20 with meaningful impressions. These have already been judged relevant and are one improvement away from real traffic. Highest return on the list.

Decayed. Compare the last 90 days against the same period a year ago. Sort by clicks lost. A page that fell from 900 clicks to 200 is telling you something specific.

High impressions, low CTR. The page ranks but nobody clicks. Often this is a title and description problem, not a content problem, and it's a 20-minute fix. Check the SERP first, because sometimes the answer is that an AI Overview now sits above the result and the clicks aren't coming back.

Zero-click pages. Indexed, no impressions, no clicks, no links. These are pruning candidates, not refresh candidates.

Sort every candidate by clicks lost times commercial value. Work down that list.

Step 2: Diagnose before rewriting

Rewriting a page that fell for a technical reason wastes a day. Work through these in order and stop at the first one that fits.

The check that matters most is the simplest one: run the query and read the results that beat you. Not the tool's summary of them. The actual pages. In ten minutes you'll usually know exactly what's missing.

Step 3: Decide the action

Four options, and picking correctly matters more than executing well.

Update. The page is fundamentally right and has aged. Refresh data, add what's missing, tighten. Most common outcome.

Rewrite. The angle is wrong or the intent shifted. Same URL, substantially new content. Keep the URL because it holds the history and the links.

Consolidate. Several thin pages compete for one topic. Merge into the strongest URL, 301 the rest, update internal links. This tends to produce the largest single jumps.

Prune. No traffic, no links, no purpose, no path to being useful. Delete and return 410, or redirect to the closest genuinely relevant page. Redirecting everything to the homepage is a soft 404 and gains nothing.

Pruning is undersold. A site carrying 300 pages nobody reads is asking a crawler to spend its budget on them, and it dilutes what the site appears to be about.

Step 4: Do the update properly

A refresh that changes the date and adds a paragraph does nothing. Google discounts it and readers notice.

Work through this list:

• Fix the intent match first. If the SERP now wants a comparison and you wrote a definition, that's the whole job.
• Answer the question in the first 100 words. Most decayed posts open with 200 words of throat-clearing.
• Add the sub-questions competitors answer and you don't. Pull them from People Also Ask and from reading the top results.
• Replace stale specifics. Prices, versions, screenshots, dates, statistics, tool names, anything with a year in it. This is the bulk of decay on practical content.
• Cut what's no longer true, and say so where the change matters to a reader.
• Add something genuinely new — original data, a real example, a table, a decision framework. The reason to rank again is that the page is better, not that it's fresher.
• Rewrite the title and description if CTR was the diagnosis. Test the promise, not the keyword placement.
• Re-do internal links in both directions. Newer posts should link to this one.
• Update dateModified in the schema, honestly, because you actually changed it.

Step 5: Consolidation mechanics

When merging pages, order matters or you'll lose the equity you're trying to gather.

1. Pick the winner: the URL with the most links and best existing position, not the best content.
2. Move the genuinely useful sections from each loser into the winner. Don't concatenate.
3. 301 every loser to the winner.
4. Update every internal link to point at the winner directly, not through the redirect.
5. Remove the losers from the sitemap; make sure the winner is in it.
6. Leave the redirects in place permanently. Removing them later loses everything they carry.

Step 6: Measure honestly

Set expectations before the work, not after. Refreshed pages typically move within two to six weeks. A page that hasn't moved in eight weeks was misdiagnosed, and the answer is usually that the intent was wrong rather than that the writing was.

Record for each page: date changed, what changed, position before, clicks before. Without that, you can't tell refreshing from noise.

The traffic-drop conversation

When someone reports a drop after an update, resist the urge to prescribe. Establish first:

• Is it traffic or rankings? Rankings flat with traffic down means the SERP changed, not the site.
• Is it the whole site or a section? Sectional drops are usually technical.
• Is it one query or all of them?
• Did anything ship that week? Deploys cause more "algorithm updates" than algorithms do.
• Are the pages still indexed?

Core updates are re-assessments of quality across the whole index, and recovery comes from the site being genuinely better, over months, not from a tweak. Anyone promising a fix in a fortnight is selling something. Say that plainly and then get to work on the parts that are real.

Don't do these

• Don't change the date and republish. It's a known pattern and it's discounted.
• Don't change URLs during a refresh. You'll trade a ranking problem for a redirect problem.
• Don't delete a page that has backlinks. Redirect it to the closest relevant page instead.
• Don't refresh everything at once. You'll never know what worked. Batch by topic, measure, continue.
• Don't add word count as a fix. Length isn't the variable. Answered questions are.
• Don't refresh a page whose query no longer has traffic. Check demand still exists before spending the day.

Where the data comes from

Search Console is the primary source for all of step 1 and it's free. Query it for the last 16 months, compare periods, and export. Everything in this skill can be done with it alone.

What it won't tell you is how the same pages are performing in AI answers, where a page can lose clicks while gaining citations or lose both without any ranking change. Keupera pairs Search Console data with AI Brand Radar for the citation side and content scoring for the rewrite itself, over MCP, which is useful when decay is happening on a surface Search Console doesn't fully report on:

claude mcp add --transport http keupera https://mcp.keupera.com/mcp -s user

Related skills

search-console-analysis, keyword-cannibalization, featured-snippet-passages, internal-linking-architecture, topical-authority-map, redirect-migration-plan