GSC knowledge base
Google Search Console knowledge base
Practical Search Console guides for traffic drops, low CTR, indexing, access and data limits. Follow checks, examples and official sources.
What problem are you encountering now?
These are operational guides. Indexing, crawling, device, and performance checks need to be done in the corresponding tools; they do not mean AISEO Site has automatic detection.
Basics and permissions
Verify the correct property, control account permissions, and explain the boundary between GSC and GA4.
Connect Search Console: first verify the correct site scope
First confirm website access in native Search Console; then connect AISEO Site using the same Google account and the correct property. Ownership verification, Google sign-in, read permission and available search data are separate checks. Follow the step where your setup actually fails instead of creating new properties repeatedly.
Steps · error branches · checklistDomain property or URL prefix: decide by URL scope
Choose the property that contains the URLs you need to measure. Use a Domain property for a domain-wide view, or a URL-prefix property for one protocol, host or directory. Keeping both is useful when their roles are explicit; adding their totals together is not.
Steps · error branches · checklistGSC permissions and handover: grant only the access scope needed
Grant access for a named task, then test it from the recipient's account. For a handover, preserve a verified business owner and review property roles, ownership verification methods and third-party app access separately. Removing an email from one screen may not close every route back in.
Steps · error branches · checklistGSC clicks and GA4 sessions differ: reconcile layer by layer
Compare Search Console clicks with GA4 Google organic sessions to find patterns, not to force equal totals. First align the site, dates and channel. Then isolate the landing pages where the discrepancy changed and test collection there. A difference between clicks and sessions is not a count of lost people.
Steps · error branches · checklistDomain migration: URL mapping before the address change tool
A migration starts with an old-to-new URL map and a tested destination. Moving a host without changing URLs, changing paths and changing domains need different actions. Verify redirects and the new pages before using any eligible Change of Address request; then monitor both properties without promising a recovery date.
Steps · error branches · checklistSearch performance
Locate metric changes to queries, pages, countries, and devices, rather than only looking at site-wide averages.
Clicks, impressions, CTR, ranking: understand the calculation before looking at changes
Read clicks, impressions, CTR and average position together. A changing total can reflect a changing mix of queries rather than better or worse performance for the same audience. Calculate ratios from their underlying counts, then check one comparable page-query segment before deciding what to change.
Steps · error branches · checklistHigh impressions, low clicks: first distinguish four types of causes
Low CTR needs a diagnosis before an edit. Determine whether the page is exposed to low-ranking or unsuitable queries, whether the result already answers the task, or whether its promise is unclear. Test one defensible change within a comparable segment; there is no universal target CTR.
Steps · error branches · checklistSearch traffic drop: locate by scope and time
Locate the lost clicks by date and page before naming a cause. Restore confirmed availability, security or indexing faults first. When the site works, investigate demand, query mix and competing results. An update occurring at the same time is context, not proof that it caused the loss.
Steps · error branches · checklistGSC regex filters: validate rules with positive and negative examples
A regex filter is useful only when it classifies the intended inputs correctly. Start with a small rule, test positive and negative examples, then inspect the returned rows. A rule being accepted does not prove it is accurate, and filtering does not reveal hidden queries.
Steps · error branches · checklistBrand and non-brand queries: build a dictionary before comparing
Build a documented brand dictionary before separating brand and non-brand search. Include genuine names and spelling variants, keep ambiguous words unresolved, and compare periods with the same rule. Brand growth is useful evidence of demand, but does not prove that an SEO article caused it.
Steps · error branches · checklistCountry and device: find the users actually affected
Country and device differences tell you where to investigate, not why performance changed. Fix one page and task, compare sufficiently sized segments, then test the relevant language, device and business conditions. A country label is not evidence of a visitor's nationality or preferred language.
Steps · error branches · checklistDiscover data: missing report does not equal zero impressions
Discover has no manual article submission that guarantees exposure. Check whether its report is available, then review the page's eligibility, image and reader experience. Missing reporting is an unknown state; a recommendation spike is not evidence of dependable future traffic.
Steps · error branches · checklistGoogle News and Web news results: do not mix definitions
Separate Google News platform traffic from the News tab in Google Search. The Google News performance report covers news.google.com and the News apps; the Search performance report's News search type covers the search tab. Compare each source within its own scope and account for a story's life cycle.
Steps · error branches · checklistIndexing and URLs
Distinguish discovered, crawled, allowed to index, and actually indexed; prioritize important pages.
URL inspection: look at historical indexing and live testing separately
URL Inspection contains two different kinds of evidence: the version Google indexed and a test of the page available now. Read their timestamps and results separately. A live test passing, or an indexing request being accepted, does not establish that the latest content is already indexed.
Steps · error branches · checklistPage indexing report: find important pages first, do not chase all green
The goal is to index useful canonical pages, not every URL or an all-green report. Define which pages should be searchable, classify the exclusions, and inspect important examples individually. A report sample is not a complete inventory, and validation is not immediate indexing.
Steps · error branches · checklistCrawled but not indexed: check value and signals page by page
‘Crawled, currently not indexed’ confirms a crawl without selection for indexing; it does not identify one root cause. For an important page, verify current delivery, assess its distinct reader value and compare duplicate alternatives. Repeated submission is not a substitute for that review.
Steps · error branches · checklistDiscovered but not indexed: check discovery paths and crawling conditions
‘Discovered, currently not indexed’ means Google knows the URL but has not yet crawled it. Check useful discovery paths, host health and unnecessary URL growth. The label alone does not prove a crawl-budget shortage or that Googlebot tried to fetch this specific page.
Steps · error branches · checklistGoogle chose a different canonical: align five types of signals
When Google selects another canonical, compare the actual pages and their signals before editing the tag. Canonical annotations express a preference for duplicate or very similar content. Align the intended destination's health, redirects, annotations, internal links and sitemap; keep genuinely distinct language pages distinct.
Steps · error branches · checklistnoindex and robots.txt: fix the correct blocking layer
Robots.txt controls crawling; noindex controls indexing when Google can read the instruction. First decide whether the URL should remain public and searchable, remain public but excluded, or require authentication. Apply the control for that purpose and inspect the delivered response, not only the source template.
Steps · error branches · checklist404, Soft 404, and 5xx: handle by business destination
Choose the response according to the page's real business state. Restore an accidentally broken page, redirect a moved page to its equivalent, or return a genuine removal response when no replacement exists. A 200 response with an empty or error body does not restore the reader's task.
Steps · error branches · checklistVideo indexing: first confirm the page is a watch page
A page being indexed does not mean its video is indexed. First confirm it is a genuine watch page, then test the player, thumbnail and media access. Use the video's reported reason to choose the repair; structured data cannot replace visible, playable content.
Steps · error branches · checklistCrawling and discovery
Check sitemap, robots, and host responses so important pages have a stable discovery path.
Sitemap: submission, processing, and indexing are three states
A sitemap supplies discovery hints, not an indexing command. Build a truthful list of intended canonical pages, verify the public file and its URLs, then track submission, Google processing and page indexing separately. Lastmod should describe a real significant change, not a daily freshness script.
Steps · error branches · checklistSitemap cannot be crawled: first see what Google gets
For ‘Couldn't fetch’, first identify the exact file Google tried to read and what was delivered. Separate network/access failure from XML parsing failure. A browser's 200 response can still contain a challenge or HTML error; repeated submission does not repair either problem.
Steps · error branches · checklistCrawl stats: host failures take priority over crawl volume
Start with host availability and failing responses, not the number of crawl requests. A higher count can come from repeated resources or duplicate URLs. Combine the native Crawl Stats report with dated release and server evidence to determine whether important content was actually reachable.
Steps · error branches · checklistrobots.txt report: verify host and actual rules
Check the robots file for the exact protocol and host, then distinguish fetch success, parsing and rule matching. The native report shows Google's fetched version; your live file may be newer. Robots restrictions are crawl instructions, not authentication or a dependable removal method.
Steps · error branches · checklistPage experience
From real user metrics to lab diagnostics, verify that performance changes have not broken the business.
Enhancements
Check visible content and structured data, and understand the applicable conditions for products, videos, and AMP.
Rich results: valid markup does not equal guaranteed display
Treat rich-result readiness as several checks: suitable content, truthful fields, valid implementation, indexable delivery and observed display. A successful validator covers only part of that chain. Add supported markup for the real page task, never fabricated reviews or invisible answers.
Steps · error branches · checklistProduct snippets and merchant listings: first distinguish page eligibility
Choose product markup for the page's actual purchase or editorial task, then reconcile the same item across visible content, structured data and any Merchant Center source. Product-snippet validity, merchant-listing eligibility and feed approval are separate states; none proves exposure or sales.
Steps · error branches · checklistAMP report: only troubleshoot existing AMP pages
Investigate AMP only if your site actually maintains AMP URLs. Decide whether to keep and repair them or retire them through a planned migration. A missing AMP report is not a reason to create AMP pages, and a valid AMP document does not guarantee ranking or search distribution.
Steps · error branches · checklistSecurity and links
Handle security and manual actions according to clear notifications, and understand links and removal reports cautiously.
Manual actions: first fix the issues in the notification
Use the native Manual Actions notice to define the actual issue and scope. Repair the reported violations across the affected pages, verify them, then submit an honest reconsideration record. A traffic decline without a notice does not establish a manual action, and acceptance of a review request does not establish removal.
Steps · error branches · checklistSecurity issues: control the incident first, then restore search
Treat a confirmed security issue as an incident affecting users and systems. Preserve evidence, contain harmful delivery and repair the entry point before seeking a search review. Deleting Google's example URLs or restoring the homepage alone does not establish that the whole site is clean.
Steps · error branches · checklistRemovals tool: temporary hiding and permanent deletion are separate
A temporary removal hides a result from Google Search; it does not delete website content or make a permanent exclusion. Correct the source first, select the narrow intended URL scope, and verify the durable method separately. Sensitive material needs actual removal or access control, not only a search request.
Steps · error branches · checklistLink report: see discoveries and associations, not a full-web backlink database
Use the Links report to investigate entry points and source context, not as a complete live backlink database or an automatic quality score. Check important internal routes and sample external sources. Unfamiliar links alone do not justify a blanket disavow action.
Steps · error branches · checklistDevelopment and data
Permissions, APIs, pagination, quotas, and data warehouse: readable data does not equal the full backend.
Search Console API: what it can read, what it cannot do for you
Choose an API method for a defined operation and verify its authorization and property scope. Search Analytics reads performance, URL Inspection reads indexed information, and other resources have their own methods. The API is not a replica of every GSC report or a general ordinary-article indexing submission service.
Steps · error branches · checklistSearch Analytics queries: dimensions, filters, and pagination
Define what one result row means, then keep the property, dates, type, filters and aggregation stable across pagination. More pages do not guarantee complete data. A page-query mapping requires joint dimensions; separate top-page and top-query lists cannot be joined by row order.
Steps · error branches · checklistAPI quotas: caching and backoff, not unlimited retries
A quota failure needs a record of the endpoint, property, user, project, request rate and error response. Search Analytics load limits and URL Inspection request quotas are different. Repeatedly retrying every failed job can deepen the failure; first reduce the workload that caused it.
Steps · error branches · checklistBulk export to BigQuery: determine data and cost responsibility first
Use bulk export when you need a durable, queryable Search Console dataset, and budget for Cloud storage and queries. Setup is not a successful export. Verify the first saved date and both performance tables, then aggregate repeated rows correctly. The setup does not backfill your earlier history.
Steps · error branches · checklistQueries fewer than the total: missing, zero, and errors are separate
A number is useful only with its scope: property, dates, search type, filters, aggregation and collection method. A missing query, failed request or unavailable report is not zero. Establish those boundaries before comparing dashboards or asking AI to explain a change.
Steps · error branches · checklistCheck guide coverage and product capabilities separately
SEO operations adds bounded technical checks, saved index-version reads, laboratory records and human content review. Native Google security, Discover, News and enhancement reports still require their own tools. A guide explains a method; it does not establish every report is connected.
View real product capabilitiesMore search and ecommerce guides
Practical checks for product visibility, multilingual websites and search measurement.
AI SEO Agency Workflow: Automate vs Review
Use a risk-based workflow to decide which SEO tasks an agency can automate, which need approval, and what evidence every client release should include.
Read the guideGoogle Showing the Wrong Product Price? What to Check
Changed a product price but Google still shows the old one? Learn what to check on your store and in Merchant Center, and how to prevent repeat mismatches.
Read the guideGoogle Shopping Wrong Color or Size? Fix Variant Links
Google Shopping opens the wrong color or size? Trace one SKU from its product record to the landing page and cart, then fix the source of the mismatch.
Read the guideGoogle Shows the Wrong Language? Hreflang Checks
Google showing the wrong language version? Check page content, canonical, hreflang and redirects with a practical bilingual page-pair worksheet.
Read the guideApproved Products Not Showing on Google?
Merchant Center approval is not a promise of impressions. Check the right country, free listings or ads, product performance, and the shopper's landing page.
Read the guideMultilingual SEO Audit Case Study: 10 Page Pairs
See how AISEO Site audited 10 English and Chinese page pairs, verified canonical and hreflang signals, and avoided unnecessary SEO changes.
Read the guideTrack AI Overviews in Google Search Console
Use Search Console's Generative AI report without mixing citations, total search clicks, analytics referrals or crawler logs. A practical measurement guide.
Read the guide