A traffic decline does not establish that a site has a manual action. Before preparing a reconsideration request, confirm the notice in Search Console’s Manual actions report. Without a notice, investigate other causes: technical failures, changing demand, migrations and ranking system changes require a different diagnosis.
Reconsideration makes sense after fixing the reported problem and collecting evidence. It is not a form for requesting better rankings or a general mechanism for reversing algorithmic updates.
Confirm the incident and the correct property
Verify that you are reviewing the right website with sufficient access. Save the date, notice text, scope and examples in a private case record. The official report documentation explains how to inspect the issue and request review after resolving it.
| Situation | First check | Workstream |
|---|---|---|
| A manual action notice | Reported type and scope | Address the violation and document review |
| A security notice | Security issues report | Resolve the vulnerability and follow that procedure |
| Decline without either notice | Performance, indexing and site changes | Technical, editorial and demand diagnosis |
| Analytics alone declines | Measurement compared with Search Console | Verify data collection before attributing an SEO loss |
The Security issues report has a different purpose and incidents may coexist. Do not assume a website is clean because one report shows no problems. Investigate specific symptoms without mixing the procedures.
Create a case record that preserves context
Before changing content, assign investigation, implementation and verification. Preserve copies and a decision log explaining what existed and what was corrected. The document need not be long: it needs traceable evidence and an owner for every task.
- Affected URL or pattern and detection date.
- Observed behaviour and reproduction steps.
- Relevant policy and reference to the notice.
- Identified cause and source of the configuration or publication.
- Correction, date and subsequent validation.
- A control preventing the pattern from returning.
Do not include passwords, access credentials or private user information. Share only what the team needs to verify the correction. For linked documents, check that permissions support the intended review without exposing unrelated data.
Fix the cause rather than only the example URLs
Use examples as the beginning of a pattern investigation. If a template inserts hidden content into hundreds of pages, editing three URLs does not repair the generator. If an editorial process creates pages with no independent usefulness, changing titles does not correct the process.
Compare the incident with the relevant spam policies. Remove the behaviour first, then validate affected content and preventive controls. Avoid universal remedies: redirects, content and link problems do not automatically require the same tool.
Do not delete links or prepare disavow files simply because traffic dropped. Such action needs a specific diagnosis of links and the applicable notice. Blind intervention can remove legitimate references without addressing the original problem.
Hypothetical example: an automated module recreates pages
Imagine a notice related to scaled content lacking value. The team removes several identified pages, but an automated job recreates them the following day. The work should include disabling or correcting that rule, reviewing the module’s full inventory and deciding which pages deserve to remain.
The case record might contain before-and-after samples, the correction date and the criteria applied to the inventory. It should not claim that Google approved the work or a client recovered traffic: this is an illustration of translating a reproducible cause into verifiable tasks.
Turn the notice into a scoped investigation
Record the incident type without translating it into a generic “penalty” label. Separate what the notice establishes from hypotheses about its origin. One column can contain the reported pattern and another the team’s observations. This prevents an initial assumption from determining every subsequent change.
If a section is affected, inventory that section and the templates or processes shared with others. Do not assume URLs visible in a screenshot describe the entire scope. Equally, do not assume every page has precisely the same defect. The investigation should establish the behaviour and where it repeats.
Preserve different versions when behaviour varies by device, direct access or referral path. Record reproduction steps and dates. If a problem can no longer be reproduced, document that limitation and investigate change history. Do not fabricate a screenshot recreating an earlier state or declare a cause without evidence.
Separate correction from general optimization
Removing a violation and improving editorial quality may happen together, but they need different acceptance criteria. Classify tasks as root-cause correction, affected-content repair, prevention or additional improvement. This helps prioritize essential work and avoids claiming a problem was resolved when it received only cosmetic changes.
| Task | Expected evidence | Insufficient evidence |
|---|---|---|
| Correct a generator | Changed rule and affected-record tests | Editing a sample without reviewing the rule |
| Review repetitive content | A documented decision for each pattern | New titles with the same underlying answer |
| Remove problematic navigation behaviour | A reproducible journey without the failure | A normal-looking home page |
| Prevent recurrence | A testable control and accountable owner | A promise it will never happen again |
This table proposes a working process, not universal Google requirements. The particular incident determines the necessary checks. If a third party caused the behaviour, document the changed dependency and your own verification as well as the supplier’s response.
Test the cause before and after correction
Select a representative sample, including pages absent from the initial examples but sharing the pattern. Save expected and observed outcomes for each test. Someone other than the implementer should be able to repeat the journey using clear instructions and obtain the same result.
When removing a module, check caches, alternative templates and automated publishing. When correcting a content rule, test a new record, an old record and an incomplete case. For navigation-dependent incidents, test direct entry and journeys through internal links. These checks reduce the risk of declaring a cause fixed while it remains active elsewhere.
Avoid simultaneously redesigning the entire site unless it is necessary for the correction. Unrelated changes make regressions harder to interpret. Maintain a version record and a way to repair a failed release without restoring the behaviour responsible for the incident.
False signals that do not justify reconsideration
A ranking decline for one word, a missing page or an alert from a commercial tool does not replace a Search Console notice. Check the property and date before diagnosing a manual action. Without one, the work may still matter, but it belongs to a different investigation.
A recently acquired domain also requires checking its actual state and available history. A purchase does not prove the site is harmless, and a historical issue is not necessarily still active. Describe only confirmed facts. When a security incident exists, follow its procedure and coordinate the technical repair rather than hiding it inside a generic letter.
Timing near a public update does not establish causation either. Separate site changes, measurement and demand. An investigation can retain several plausible explanations without forcing a single answer. An inapplicable reconsideration request cannot replace that work.
Hypothetical example of useful review evidence
Imagine a module publishing resources without sufficient information, and the team identifies that rule as the cause relevant to the notice. Useful evidence could describe the module identifier, revised eligibility rule, reviewed inventory and sanitized examples. Credentials and private data are unnecessary to show what changed.
The request should not claim “we removed all useless pages” without explaining how the scope was determined. It can describe the review process and resolved exceptions using actual case facts only. If earlier attempts failed, document the final correction and relevant lesson without overwhelming the reviewer with an unnecessary chronology.
An evidence link should be understandable on its own. Label the sample, explain which behaviour it illustrates and ensure the intended reviewer can access it. Avoid attaching an entire unfiltered export and expecting the reader to infer which rows matter. A concise, relevant record is more useful than a large volume of unexplained files.
Manage the waiting period without losing control
While the request is reviewed, preserve the corrected configuration and monitor processes capable of recreating the issue. Assign someone to review notifications and record the decision. A supplier change or backup restoration should not silently reinstate a removed rule.
If the response reports remaining problems, expand the inventory and reproduce the pattern before submitting again. If the action is removed, verify the report and continue observing access and performance. In both cases, distinguish the review decision from later commercial outcomes: a favourable decision does not establish that every page recovered its previous performance.
Use the completed case to improve release controls rather than to create a universal recovery promise. The next incident may have a different cause and scope. What transfers is the method: confirm the notice, collect evidence, correct the pattern, verify independently and communicate only the work actually completed.
Write a concise, specific and verifiable request
The request should explain the problem and completed work without requiring a reviewer to interpret generic statements. Summarize the real cause, corrections and validation. Use facts from the case and remove promises you cannot demonstrate.
- Describe the behaviour that violated the reported policy.
- Explain its origin and what changed in the system or process.
- Identify the reviewed scope, including similar patterns.
- Provide relevant, accessible evidence of correction.
- Describe the control preventing recurrence.
“We improved our SEO and ask for another chance” does not explain what was resolved. A standard letter listing work that never happened is equally unhelpful. Precision makes the case reviewable; length alone does not add evidence.
Validate the website before requesting review
When practical, have someone other than the implementer perform a separate check. Reproduce the problematic journey, sample representative pages and confirm that valid pages remain accessible. Test mobile and desktop if behaviour depended on device or origin.
Look for scheduled jobs, plugins, rules and imports that could reintroduce the issue. Define a check after the next automated publication. Complete any remaining corrections before declaring the case resolved.
After submission, track the decision and retain control
Save the request and confirmation, assign follow-up ownership and wait for a decision before submitting again. A rejection requires reading what remains unresolved and extending the investigation; resending unchanged wording provides no new information.
Removal of an action does not guarantee previous rankings or a particular recovery date. Afterwards, assess accessible pages, relevant queries and commercial outcomes over comparable periods. Without a manual action, use an organic traffic recovery process and Google’s traffic decline diagnostic guidance to investigate the appropriate cause.
Maintain a task-status column separating planned, implemented and independently verified changes. Only completed work belongs in the request as completed evidence. A provider saying a correction is scheduled does not establish that the published website has changed, so check the actual accessible result before describing it.
Frequently asked questions
Does a large traffic drop mean Google applied a manual action?
No. Confirm the notice in the correct property’s Manual actions report. Without a notice, investigate technical errors, demand changes, measurement and ranking systems.
Can I request reconsideration after an algorithmic update?
The reconsideration process described here applies to a confirmed manual action. An algorithmic change without that notice needs a different investigation, not a reconsideration letter.
Is fixing the example URLs enough?
Investigate the pattern and correct the problem across its actual scope. Examples help locate it, but a template or process may affect additional pages.
Will rankings return when the action is removed?
Previous rankings are not guaranteed. Eligibility, competition, content and other factors still matter; assess performance after the decision.