How to Recover From a Google Core Update (Without Making It Worse)

You’ve done the diagnosis. You waited for the rollout to finish, compared the right weeks, ruled out tracking problems and seasonality, and the picture is clear: the core update reassessed your site and your organic traffic is down.

If you haven’t done that work yet, stop and read how to tell whether a core update actually hit you first. Recovery plans built on a misdiagnosis are how sites make a bad month into a bad year.

This post is about what comes next. It’s not a list of tricks. Core updates don’t respond to tricks. It’s a realistic plan for working out which pages lost, why the pages that beat you are winning, what to change, and how to know whether the changes worked.

Set Your Expectations First

Two facts shape everything below.

Recovery takes months, not days. Google’s own guidance on core updates says that pages affected by one may not see their rankings change until the next broad core update, and those arrive every few months. Between updates Google ships smaller unannounced changes that can move things earlier, but you shouldn’t plan around that. If you make good changes in week two, the first real read on whether they worked is often the next core update.

There is nothing to “fix” in the technical sense. A core update is not a penalty and doesn’t flag a specific violation. Your pages were re-evaluated against a quality bar that moved, and other pages now clear it better than yours. Recovery means genuinely closing that gap, not finding the switch that got flipped.

That sounds discouraging. It’s actually useful, because it tells you to spend your effort on content and site quality, and it tells you not to expect a signal for a while. Sites that recover are usually the ones that made real improvements and then had the patience to leave them alone.

Step 1: The Things Not to Do

Most self-inflicted damage after a core update happens in the first two weeks, when the pressure to do something is highest. Before you change anything, take these off the table.

Don’t make “quick fix” changes. Google explicitly warns against removing page elements because you heard they were bad for SEO. Stripping out author boxes, dates, internal links or FAQs on a hunch does nothing for quality and can make pages worse.
Don’t delete content in bulk on day one. Pruning has its place (see Step 4), but a panic purge of every page that lost traffic removes the pages that might have recovered and any internal links that pointed through them.
Don’t migrate, redesign or change URLs. A site move during recovery mixes two signals together. If a redesign was already planned, finish recovery first or accept you’ll never know which change did what.
Don’t disavow links or chase “toxic backlinks”. Core updates are about content quality. Link cleanups are a response to a different problem and almost never move the needle here.
Don’t rewrite everything with AI to “refresh” it. Republishing hundreds of pages with lightly reworded text is the opposite of what the update rewarded. If anything, it produces more of the thin, generic content that lost.

If you find yourself about to do one of these because a forum thread said it worked for someone, close the tab.

Step 2: Find Out Exactly What Lost

You can’t fix “the site”. You can fix specific pages that lost specific queries. Build that list before you touch anything.

In Search Console’s Performance report, use the date comparison. Set the current period to the two or three weeks after the update finished rolling out, and the previous period to the same length immediately before it started. Then:

Pages tab, sorted by click difference. This is your priority list. Export it. Note whether losses cluster in one section, template or content type. A drop concentrated in your comparison articles, or your programmatic city pages, or your old news archive, tells you where Google’s reassessment landed.
Queries tab, sorted by click difference. For each page on your list, which queries did it lose? Pay attention to the position change, not just the clicks. A page that fell from position 3 to 6 has a different problem from one that fell from 6 to 40.
Check what held or gained. This is the step people skip. Pages that improved during the same update are your best evidence of what Google now prefers on your site. If your detailed first-hand guides gained while your roundup posts lost, you’ve learned something a generic checklist can’t tell you.

Group the losers into a handful of categories: by section, by content type, by intent (informational, commercial, navigational). Recovery work is done per category, not per page, because the fix for “our thin product roundups lost” is one decision applied to fifty pages.

Step 3: Look at Who Beat You

For each category, pick three or four representative queries and look at the live results. Not your ranking tracker, the actual page of results. You’re trying to answer one question: what do the pages now ranking above me have that mine doesn’t?

Common answers:

The intent shifted. Google now treats the query as wanting something different: a tool instead of an article, a quick answer instead of a deep guide, fresh news instead of an evergreen explainer. If the entire top ten changed format, no amount of polishing your existing format will help.
They show first-hand experience. Photos they took, tests they ran, numbers from their own data, opinions with reasoning. Yours summarises what everyone else already said.
They’re more specific. Your page covers a topic. Theirs answers the exact question, with the answer near the top.
They’re maintained. Current prices, current screenshots, current product names. Yours still references a version from two years ago.
They’re a different kind of site. Forums, official documentation, retailers, video. If that’s what now fills the results, your best move may be to stop competing for that query and redirect effort to the ones you can win.
An AI Overview took the click. Impressions steady, clicks down, position roughly flat. This isn’t a core update effect and improving the page won’t bring the click back. Treat it separately.

Write down what you find. Vague conclusions like “they have better content” lead to vague rewrites. “They include a comparison table with current prices and we don’t” leads to a fix.

Step 4: Triage Into Three Buckets

With the losing pages grouped and the winners understood, decide what happens to each category. Every page lands in one of three buckets.

Improve. The query still matches what you do, you have something genuine to add, and the page is worth the work. This is where most of your effort goes.

Consolidate. You have several pages circling the same topic, none of them strong. Merge them into one comprehensive page and redirect the others to it. Overlapping thin pages were a common casualty of recent updates, and consolidation is the one form of “removal” that reliably helps.

Retire. The page has no traffic, no links, no purpose, and you can’t honestly make it useful. Remove it or noindex it. Do this deliberately, in a batch, after the triage, and annotate the date. Don’t do it as a reflex in week one, and don’t expect it alone to recover anything. Pruning clears the ground so the improved pages stand out. It’s not the recovery itself.

If you’re not sure which bucket a category belongs in, default to Improve for anything that still gets impressions and Consolidate for anything that has siblings.

Step 5: Actually Improve the Pages

Google publishes a set of self-assessment questions for exactly this situation. They’re worth reading in full, but they boil down to a handful of things you can act on.

Add what only you can add. Original data, your own testing, real examples from your work, a clearly argued opinion. If a page could have been written by someone who had never done the thing, it reads that way.

Answer the query, then go deeper. Put the direct answer where a searcher scanning the page finds it in seconds. Then earn the rest of the page with detail they can’t get from the snippet.

Cut the filler. Definitions of things your reader already knows, long preambles, sections that exist to hit a word count. Shorter and denser beats longer and padded. Google has said repeatedly that there is no ideal word count.

Show who wrote it and why they’re qualified. A real author with a bio that says what they’ve done. For topics that touch health, money or safety, this matters more than anything else on this list.

Update what’s stale. Prices, screenshots, product names, statistics, dates. A page that says “as of 2023” about something that changed twice since then has told Google and the reader that nobody is looking after it.

Match the format the results now reward. If the winners use tables, use a table. If they use step-by-step numbered instructions, do that. Don’t copy them, but don’t insist on a format the results have moved away from.

Fix the page experience, but don’t mistake it for recovery. Intrusive ads, pop-ups on load, layouts that push the content below the fold. Google’s guidance mentions page experience as part of the overall assessment. Clean it up, but understand it’s a supporting factor. Nobody recovers from a core update by shaving 200ms off Largest Contentful Paint.

Step 6: Look at the Site, Not Just the Pages

Core updates evaluate quality at a site level as well as a page level. If a large share of your site is thin, the good pages get dragged down with it. After the page-level work, review the site as a whole.

Does the site have a clear purpose that a first-time visitor would understand in a few seconds?
Is there a real About page, a real way to contact you, real people behind the content?
Are there whole sections that exist only because they might rank: tag archives with one post each, auto-generated location pages, a “news” section nobody has updated in a year?
Does the ad density or affiliate link density make the site feel like it exists to monetise rather than help?

None of these are individually decisive. Together they form the impression Google’s systems are trying to approximate.

Step 7: Annotate Everything and Measure at the Right Moment

This is the step that separates people who learn from a core update from people who go through the same panic next time.

Annotate every change you make, with the date and what it covered: “Consolidated 14 roundup posts into 3”, “Rewrote the product guides section with new testing”, “Removed 80 empty tag pages”. Six months from now you will not remember what you did in which week, and without that record you cannot tell which changes worked.

Then measure at the next core update, not next week. Watch the Google Search Status Dashboard for the next core update announcement. When it finishes rolling out, wait a week, then run the same comparison you ran for the diagnosis: the weeks after the new update against the weeks before it, on the specific pages and queries from your Step 2 list. That is your first honest read.

Between updates, check in monthly rather than daily. Look for early movement on the improved pages, but don’t read too much into it either way.

If the next update doesn’t move things, go back to Step 3. Either the changes weren’t substantial enough, or the query landscape moved further than you thought, or the pages are competing against a kind of result you can’t become. All three are useful to know. Some pages don’t recover because the world changed, and the right response is to put the effort somewhere it can pay off.

Running This With AI Data Stream

Everything above is doable with Search Console and GA4 alone. What makes it painful is the bookkeeping: remembering the exact update dates, setting the right comparison windows, keeping a list of affected pages that survives to the next update, and remembering what you changed when.

AI Data Stream connects your Search Console and GA4 data and keeps a timeline alongside it. Confirmed Google algorithm updates are added as annotations automatically, so “the weeks after the last core update versus the weeks before” is a question you can ask in plain language rather than a date range you have to look up and type. Your own changes go on the same timeline, so when the next update lands you can ask which of the pages you rewrote in September recovered and which didn’t, and get an answer that already accounts for the dates.

Questions people run during a recovery:

“Which pages lost the most clicks after the last core update finished, and what were their top queries before the drop?”
“Compare clicks and average position for the pages tagged in my ‘guides rewrite’ annotation, before and after the most recent update.”
“Did impressions or clicks fall on these pages? Show me the split.”

You can also schedule a monthly report on your affected pages so the between-update check-in from Step 7 happens on its own, rather than relying on you remembering to look.

The Bottom Line

Recovering from a core update is slow, unglamorous work: find the pages that lost, understand what now wins, decide honestly which pages deserve the effort, make them genuinely better, and then leave them alone long enough for Google to reassess. Most of the damage sites do to themselves comes from skipping the first three steps and rushing the last one.

Keep a record of what you changed. The next core update will tell you whether it worked, and the one after that will tell you again. Sites that treat this as an ongoing loop rather than a one-off emergency are the ones that stop being surprised by core updates altogether.

Want the before-and-after comparison ready the moment the next update finishes? Try AI Data Stream free →