A stable-ranking page disappears from Google: site: returns nothing, Performance impressions crash to 0, URL Inspection says “URL is not on Google” with no obvious technical reason. Fastest triage: open Search Console → Security & Manual Actions → Manual Actions. If there is a notice, it is a manual action and you file a reconsideration request after fixing it. If it says “No issues detected”, it is almost certainly an algorithmic re-evaluation (core update), which no notification ever announces and which only recovers with site-wide quality work.
The two paths look identical from the outside but the fixes share nothing. Identify which bucket you are in before touching anything.
Symptoms
site:yourdomain.com/urlreturns nothing- Performance report shows impressions sliding to 0 over a few days
- URL Inspection shows “URL is not on Google” with no obvious technical reason
- HTTP 200, no
noindex, canonical is normal,robots.txtdoes not block the URL
If any of those last four are actually broken (a stray noindex, a bad robots.txt rule, a 404/5xx, a wrong canonical), it is a technical regression, not a policy event — jump straight to Step 6.
Which bucket are you in?
| Signal | Manual action | Algorithmic (core update) | Technical regression |
|---|---|---|---|
| Notice in Manual Actions report | Yes | No | No |
| Email from Search Console | Usually yes | No | No |
| Timing lines up with a known core/spam update | Sometimes | Yes | No |
URL Inspection shows noindex / blocked / 404 / wrong canonical | No | No | Yes |
| Scope | One section or whole site | Often site-wide, gradual | Specific URLs |
| Fix | Reconsideration request | Quality overhaul + wait for next update | Fix the regression, request indexing |
Common causes
1. Manual action — spam / thin content / unnatural links
A human reviewer or Google’s automated spam systems flagged the site. Common triggers:
- Scaled low-value content (bulk AI output published with no human review or fact-check)
- Buying backlinks / link schemes (a “Unnatural links to your site” or “Unnatural links from your site” notice)
- Cloaking (Googlebot sees one version, users see another)
- User-generated spam (comment / forum spam left unmoderated)
- Thin affiliate or doorway pages with no added value
- Site reputation abuse / “parasite SEO” (hosting third-party content to ride your domain’s authority — Google made this a named policy and enforces it manually)
How to confirm: Search Console → Security & Manual Actions → Manual Actions. Any notice here means manual. The report names the issue type, the affected scope, detection date, and sample URLs.
2. Core update demoted the page
Google ships roughly 3-4 broad core updates a year that re-assess “which pages deserve to rank.” If your overall content quality did not match the update’s direction, pages get demoted — often not fully deindexed, but impressions collapse so far it looks deindexed. The old standalone “Helpful Content Update” no longer exists as a separate launch: in March 2024 Google folded the helpful-content signal into the core ranking system, so it now runs continuously rather than as a periodic refresh. There is no separate “HCU recovery” event anymore — recovery rides the next core update.
Recent rollout windows to compare your drop date against (as of June 2026):
- December 2025 core update: Dec 11 - Dec 29, 2025
- March 2026 spam update: rolled out March 24, 2026, completed in roughly 19.5 hours (the fastest spam update on record)
- March 2026 core update: Mar 27 - Apr 8, 2026
- May 2026 core update: started May 21, 2026
Core updates still take 2-3 weeks to fully roll out; spam updates can now finish in under a day. If your drop falls inside one of these windows, treat it as algorithmic.
How to confirm: Cross-check the official Google Search Status Dashboard and the Search ranking updates list against the day your impressions started falling.
3. Reclassified as a duplicate of another page
Google’s canonical consolidation decided your page is a duplicate of another URL, transferred the ranking signals there, and dropped yours from the index.
How to confirm: URL Inspection → “Page indexing” → compare User-declared canonical vs Google-selected canonical. If Google selected a different URL, you were consolidated, and URL Inspection will say “Duplicate, Google chose different canonical than user”.
4. Persistent server 5xx / extended downtime
If the page returns 5xx (or times out) for an extended period, Google drops it to avoid serving a broken result. It usually re-indexes within days to a couple of weeks once stable 200s return and the URL is re-crawled.
Shortest path to fix
Step 1: Always check Manual Actions first
Search Console → Security & Manual Actions → Manual ActionsThree possible outcomes:
- “No issues detected” → not manual. Go to Step 3 for the algorithmic path.
- Site-wide notice (e.g., “Pure spam”) → severe, the whole property is affected.
- Partial match notice (e.g., “Thin content with little or no added value”) → only specific sections.
Each notice includes the violation type, affected scope, and example URLs. Read the sample URLs carefully — they tell you exactly what Google objected to.
Step 2: If manual, fix the violation then request reconsideration
Fix the exact thing the notice names:
- Scaled / spam content: delete or substantially rewrite every low-quality page (do not just
noindexthem and walk away — a reviewer wants to see the spam gone) - “Unnatural links to your site”: submit a disavow file (see FAQ — this is the one case where disavow genuinely applies)
- Cloaking: confirm every user agent, including Googlebot, gets the same content (test with URL Inspection → “View crawled page”)
- Thin affiliate: add genuine first-hand reviews, original data, or testing
Then: Search Console → Manual Actions → Request review. In the form, be specific:
1. Issue: the exact violation type Google flagged
2. What we changed: concrete fix actions with example URLs and dates
3. How we prevent recurrence: the process or staffing change you put in placeReviews are processed manually and typically take 2-6 weeks (occasionally a few days, occasionally up to a month). A first rejection is normal; read the feedback, fix what they point to, and resubmit. Do not submit a half-fixed site — repeated thin requests slow you down.
Step 3: If not manual, correlate your drop date with update timing
Pin down the day impressions started falling (Performance report → set the date range around the drop), then compare it to the rollout windows listed above. If your drop sits inside a confirmed core or spam update window, you are looking at an algorithmic re-evaluation, not a technical bug.
Drop date: 2026-05-22
Matching window: May 2026 core update (started May 21) → algorithmicStep 4: Algorithmic demotion = site-wide quality overhaul
Technical fixes and re-submitting URLs do nothing for an algorithmic drop. Only content quality moves it, and only at the next core update. Concrete directions:
- Cut the thinnest pages: find pages under
< 400words or with no unique signal, then either substantially improve them,noindexthem, or return410. Bulk-thin pages drag down the whole property’s quality assessment. - Add real density to the survivors: original screenshots, first-hand testing, exclusive data, honest comparison tables — the things a competitor cannot copy.
- Build topical authority: cluster content tightly around core topics so Google reads the site as an authority, not a grab-bag.
- Earn editorial links: mentions and links from genuinely relevant, reputable sites.
- Stop publishing un-reviewed AI output: Google’s policy targets “scaled content abuse,” and the helpful-content signal is now continuous. Every AI-assisted draft needs human fact-checking and first-hand experience added before it ships.
Realistic recovery timeline: months, not weeks, and visible only when the next core update re-evaluates the site.
Step 5: Compare yourself against the current top 3
Open your previously well-ranking queries and study who ranks now:
- Word count, depth, and signal density
- Backlink profile and domain authority
- First-hand experience, original media, data
Be honest about the gap. Whatever the top results have that your page lacks, that is your work list.
Step 6: If it is a technical regression, fix it and request indexing
The simplest case. If URL Inspection shows a stray noindex, a robots.txt block, a 404/5xx, or a wrong canonical, fix the root cause, then URL Inspection → Request indexing. These usually return within days to two weeks once re-crawled.
How to confirm it is fixed
- Manual action: the Manual Actions report flips to “No issues detected” and you get a resolution message. Indexing then returns over the following days to weeks.
- Algorithmic: there is no “fixed” notification. Watch Performance impressions recover after a subsequent core update; partial recovery often precedes full recovery.
- Technical: re-run URL Inspection → it should report “URL is on Google” and “Indexing allowed: Yes” once re-crawled.
When it is not on you
A core-update demotion often only reverses at the next core update — the timing is not in your hands. Patience plus genuine quality work is the real fix; panicked thrashing of canonicals and content resets Google’s assessment and slows recovery.
Easy to misdiagnose
- Assuming any deindex = manual action. Most are algorithmic and silent.
- Believing repeated “Request indexing” recovers an algorithmic drop. It cannot. Algorithmic recovery needs quality work plus a future core update.
- Thinking a domain change skips the penalty. Algorithmic quality history follows the content and links; a manual action follows the entity and can be re-applied.
- Thinking more publishing “washes out” the problem. Bulk low-quality publishing strengthens the spam and low-quality signal.
- Reaching for the disavow tool by default. It only matters for an unnatural-links manual action; SpamBrain already discounts most junk links automatically.
Prevention
- Don’t ship thin, over-affiliated, or templated pages; slow with unique value beats fast without.
- Diversify topics across clusters so one core update can’t nuke the whole site.
- Follow the Search Central blog and a tracker like Search Engine Roundtable for update direction and timing.
- Run a quarterly content audit; proactively
noindexor delete thin pages. - Monitor impressions in Search Console; investigate any 30%+ drop immediately and note whether it lines up with a known update.
FAQ
Q: How long does a reconsideration request take? A: Typically 2-6 weeks as of June 2026 — occasionally a few days, occasionally up to a month. Requests are reviewed by humans, so resubmitting daily doesn’t speed it up. Fix thoroughly, then submit once and wait.
Q: Can deindexed pages come back? A: Yes. A manual action usually clears within days to a few weeks after your reconsideration is accepted and the site is re-crawled. An algorithmic drop only recovers when a future core update re-evaluates the improved site — plan on months, not weeks.
Q: There’s no separate “Helpful Content Update” anymore? A: Correct. Google folded the helpful-content signal into core ranking in March 2024, so it runs continuously instead of as a periodic launch. There is no standalone “HCU recovery”; recovery now rides the next core update.
Q: Can a manual action and an algorithmic drop happen at the same time? A: Yes. Fix the manual action first (it’s visible in Search Console and has a defined recovery path), then do the site-wide quality work for the algorithmic side.
Q: Does the disavow tool help? A: Only for an active “Unnatural links to your site” manual action (or a strong, evidence-based belief one is imminent). Google’s John Mueller has repeatedly said disavow is not part of normal site maintenance — SpamBrain already ignores most spammy links, and a sloppy disavow file can hurt by removing good links.
Q: How do I find the exact day my traffic dropped? A: Search Console → Performance → set a wide date range and watch the impressions line. The first day it falls off a cliff is your drop date; compare it to the update windows above to classify the cause.
Related
Tags: #SEO #Google #Search Console #Indexing #Troubleshooting #Deindex