Signals, Diagnosis and Fixes for a Google Penalty in 2026

A Google penalty is a loss of visibility in search results, caused either by a manual action notified in Search Console or by an algorithmic demotion that sends no message. For a B2B site, a sudden drop in rankings quickly translates into fewer leads, or even partial deindexing.

This guide is for the marketing manager or SEO who needs to qualify the drop without false positives, choose the fixes and track the recovery. It complements the overview of search engine optimisation.

Understanding Google sanctions: causes, risks and the plan to recover

A sanction follows a guideline violation or a reassessment of the quality of a set of pages. It sits downstream of ranking, whose factor families are described in the article on how Google ranks pages.

With core updates and anti-spam enforcement, re-evaluations are easily confused with sanctions. The difference is decisive: a manual action comes with an explicit reason and a defined process, whereas an algorithmic demotion is diagnosed through triangulation. The risk shows up in both visibility and revenue:

• Demotion: ranking losses across one or more query groups.
• Deindexing: strategic pages vanish from results, cutting acquisition on key intents.
• Lost leads: when organic is a major channel, the drop feeds directly into the pipeline.

Manual action vs algorithmic filter: identify the issue before you act

Before making changes, establish a binary diagnosis: manual action (with notification) or algorithmic demotion (without notification). This choice determines the method and the way out.

Understanding a manual action: signals, messages and limits in Search Console

A manual action, applied after a review by a Google team, is signalled by a notification in Search Console under Security & Manual Actions > Manual actions, which specifies the issue type and, in most cases, example URLs affected.

Reasons include unnatural links, user-generated spam, thin content, cloaking, hidden text and hacking. These practices and how they are detected are covered in the guide to black hat practices; this page deals with the sanction and the way out.

Algorithmic filter: how to distinguish it from normal ranking volatility

An algorithmic demotion is inferred from symptoms: simultaneous drops in impressions and clicks, a decline across a set of queries, a fall concentrated in one directory. Also check the opposite scenario, stable rankings with falling click-through rate: zero-click searches account for 60% of searches (Semrush, 2025), and a fall in sessions can come from AI Overviews rather than a sanction.

Then build a timeline, which points you to the root cause without proving causation:

• the start date of the decline, across impressions, clicks, rankings and indexing;
• internal changes: migration, redesign, deployments, removals, link campaigns, page generation;
• external events: volatility periods, core updates, spam updates.

Why a sanction affects rankings: common causes of a drop

Causes fall into three families with the same effects: ranking losses, less organic traffic, indexing failures, and sometimes deindexing.

• Content: duplication, similar pages generated at scale, thin pages, artificial keyword repetition.
• Links: rapid, unnatural acquisition, repetitive anchors, site networks, undeclared paid links.
• Technical and abuse: hacking, injected spam pages, deceptive redirects, inconsistent canonicals or accidental noindex.

The line between excessive optimisation and manipulation is covered in the article on over-optimisation. For links, what matters is naturalness and quality rather than volume, a principle developed on the page dedicated to link building.

Panda and Penguin: what these filters really changed (and what still applies)

Panda (2011) and Penguin (2012) have been integrated into core ranking systems, but their two targets remain the main causes of demotion: weak content and artificial popularity.

Panda's legacy and Helpful Content: the content quality logic

Panda targeted content quality; the Helpful Content system, which assessed how useful content was to readers, was integrated into the core algorithm in March 2024. Symptoms: traffic drops, declines across many informational queries, less efficient indexing. Remedies: reduce duplication, consolidate similar pages, enrich existing content, remove pages with no value, then wait for gradual re-evaluation.

Penguin's legacy: the unnatural links logic

Penguin targeted paid links and over-optimised anchors. It has run in real time since 2016 and discounts suspicious links rather than penalising the whole site; Google does not name any "over-optimisation penalty". The commercial impact remains similar: losses on competitive queries. Only an "unnatural links" manual action involves an explicit procedure.

The March 2024 spam policies: three abuses now targeted

In March 2024, Google added three practices to its spam policies that can trigger a manual action or a demotion:

• Scaled content abuse: producing pages at scale, with or without AI, to manipulate rankings rather than help readers.
• Site reputation abuse: hosting third-party content that exploits a domain's authority on an unrelated topic, a policy enforced from May 2024.
• Expired domain abuse: buying an expired domain to exploit its history for unrelated content.

A site that publishes at scale with AI or hosts partner content must therefore govern what it publishes.

How to spot a Google penalty: a step-by-step diagnosis

A reliable diagnosis segments the drop and cross-checks Search Console (pre-click) with analytics (post-click), never a single metric.

Define the scope: site-wide, directory-level, page types and affected queries

Start by answering four questions:

• does the drop affect the whole site or a directory (/blog/, /products/)?
• does it affect a template, i.e. a set of similar pages?
• does it mainly affect non-brand queries, or the brand as well?
• is it worse on mobile or on desktop?

A drop by template points to a structural cause; a drop by query family, to a mismatch with search intent.

Check Search Console and read the before and after curves

Search Console data is not real-time: read trends over several weeks, isolating the break date. Each symptom points to a likely cause and a report to check:

Rule out alternatives: migration, redesign, seasonality and competition

Before removing content or submitting a disavow file, eliminate false positives:

• migration or redesign: redirects, canonicals, internal linking, lost pages;
• measurement issues: duplicate tags, consent, GA4 channel grouping;
• seasonality and intent shifts;
• competition: new entrants, fresher content, richer SERPs.

Action plan to recover from a Google penalty: prioritise for a lasting comeback

An effective plan follows a "stabilise, clean up, requalify, measure" logic. Limit parallel workstreams and keep a log: without it, neither attributing effects nor filing a reconsideration request is possible.

Phase 1: secure and stabilise

If Search Console flags a security issue, or unknown pages appear in the index, stop the bleeding before any optimisation:

• clean up the site: injected pages, compromised accounts, vulnerable plugins;
• remove parasitic pages and close spam vectors (forms, unmoderated user-generated content);
• review redirects to eliminate deceptive redirects and long chains.

Phase 2: clean up content through pruning

Work in batches (clusters, directories, templates). Content pruning is decided on observable criteria, not gut feel:

Phase 3: clean up your link profile and disavow if needed

For link-related issues, the standard process is as follows:

• inventory backlinks, Search Console providing a partial but useful view;
• identify risky domains and anchors: spikes, networks, repetitive anchors;
• request removals when possible, documenting each request;
• reserve disavow for "unnatural links" manual actions or strong evidence of toxicity.

Disavowing by default loses useful signals, whereas Penguin already discounts most suspicious links.

Phase 4: fix indexing and crawling

Align internal links, redirects and canonicals, reduce indexable parameter URLs, check that noindex does not reach business pages and monitor the gap between submitted and indexed URLs. Server logs show what Googlebot actually crawls.

Phase 5: re-validate and monitor the recovery

Timelines vary depending on the issue type and the depth of the fixes; no duration is guaranteed. Each case has its own way out:

Look for exit signals: stabilised exclusions, returning impressions and non-brand queries, converting pages climbing back.

Lifting a manual action: get it removed without wasting time

A manual action should be handled like a case file. Google expects the issue to be fixed across the full scope (not just a few examples), actions that match the stated reason and preventive measures. The reconsideration request follows a fixed structure:

• a summary of the reason and scope: site, directory, page types;
• the list of actions taken, dated and grouped by category;
• the evidence: fixed URLs, removed pages, anti-spam measures in place;
• the commitment to prevention: publishing workflow, quality control, link review.

Stay factual. If rejected, look for blind spots (old directories, subdomains, orphan pages); if partially lifted, isolate the still-affected sections and start again.

Measure the recovery and prevent recurrence

Measuring recovery stops you restarting production before the root cause is fixed. Search Console measures before the click (coverage, returning non-brand queries); GA4 measures after it, on the Organic Search channel: engaged sessions and demo requests. Tie it to value with reporting focused on SEO return on investment.

The most common mistakes during the fix phase are well known:

• relying on a drop in sessions without checking impressions, click-through rate and rankings;
• deleting hundreds of pages without mapping intent and cannibalisation;
• changing content, links and technical setup at the same time, making attribution impossible;
• not documenting fixes, which weakens any reconsideration request.

Prevention follows the cluster's cadence: alerts every week (security, manual actions, indexing), decisions every month (strategic pages, new content, new links), a portfolio review every quarter. Three safeguards complete the routine:

• editorial: standards of originality and usefulness, review of AI-produced content;
• link building: links earned for editorial reasons, diversified anchors, investigation of any backlink spike;
• technical: security monitoring, checks on canonicals, redirects and noindex after every release, following the foundation described in our technical SEO guide.

What to do if you are stuck: targeted fixes, a redesign, or a domain change

When fixes stall, three paths are open, with very different levels of risk:

Prioritise fixes over production when indexing is unstable, when click-through rate collapses on landing pages or when a manual action is active: publishing more would amplify the problem. The first quick win: restore existing pages that were indexed and historically performed well.

Depending on your situation

Where to start depends on what you observe:

• You do not yet know whether it is a sanction: first read ranking changes and their causes with the guide to site ranking analysis.
• You want to track recovery query by query: set up a query basket and alert thresholds with the article on your position in Google.
• Indexing is deteriorating without a message: check crawling and robot directives with the guide to Googlebot.
• Your content lacks proof of expertise: strengthen page credibility before republishing with the article on E-E-A-T.

To qualify the sanction (manual action or demotion) and isolate the sections to fix, the Incremys 360° SEO & GEO audit consolidates technical, semantic and competitive findings into a prioritised roadmap.

FAQ: penalties, diagnosis and recovery

What is a penalty, and why does it matter?

A Google penalty is a loss of visibility due either to a manual action for guideline violations or to an algorithmic reassessment. It matters because clicks are concentrated in the top positions: losing a few places drains a page of its traffic.

What is the impact on rankings and revenue?

For rankings, the impact ranges from demotion on a few queries to deindexing of entire pages. For revenue, the effect is indirect but fast: fewer pages visible in high-click-through-rate areas means fewer sessions and fewer leads.

How can you tell whether the drop is from a manual action or an algorithmic filter?

A manual action is identified by a message in the Security & Manual Actions section of Search Console. With no message, suspect an algorithmic demotion and run a before-and-after analysis of impressions, rankings and click-through rate, segmented by directories, pages, queries, countries and devices.

How long does recovery take?

Recovery time depends on the type of sanction and the depth of the fixes, with no guaranteed timeline. A manual action is lifted once a reconsideration request is accepted. An algorithmic demotion fades as re-evaluations happen, sometimes across several updates.

What mistakes should you avoid during the fix phase?

The mistakes to avoid are unmeasured mass deletions, preventive disavows without evidence and multiple simultaneous changes. Document every action and keep a clear timeline, to understand what produces an effect and to support any reconsideration request.

Which practices reduce the risk of recurrence?

The risk of recurrence falls with editorial standards, link building based on natural and diversified links, and technical checks after every release. A weekly alert routine and a monthly decision routine complete this governance.

Should you consider a major redesign or changing domains?

A redesign or domain change is only justified if targeted fixes are impossible: major technical debt, unmanageable architecture, recurring spam you cannot control. These options increase risk (migration, temporary losses, measurement complexity) and never replace fixing root causes.