Missing queries in Search Console: privacy and limits
Read the conclusion first
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.
Applicable conditions
Applies to report reconciliation, AI input, and cross-window comparison.
Before you open the report
The request scope, response status, returned dates/rows and any report privacy or availability notes.
Use one concrete URL or report first. Keep account credentials and private visitor data out of shared notes.
Steps and verification
Write a measurement contract beside the chart
Record the exact property string, date range and timezone, search type, dimensions, filters, data state and collection time. Add whether the dataset came from the interface, Search Analytics API or native bulk export. Keep that information with a downloaded file, not only in the analyst's memory.
Teaching comparison: an unfiltered property total and a page-query extract for mobile users in one country answer different questions. Their totals can differ without a broken integration. First align the definitions; only then investigate a remaining discrepancy.
Separate visible detail from the total
Search Console protects rare query text. Query-level rows therefore cannot be treated as a complete list of every search. The API also returns top rows subject to internal limits, even with pagination; more collected pages do not establish complete coverage.
An aggregate total and the sum of visible query rows need not match. Preserve both with their labels. Do not distribute the difference across guessed keywords, infer hidden search phrases, or label an anonymized query row as a literal empty search.
Distinguish zero, missing, preliminary and failed
Zero is a successful observation within the specified scope. Missing means a row or interval is absent. Preliminary means recent data can change. Failed means the collection did not establish a result. A dashboard should expose those states instead of using zero as a universal fallback.
If a report is unavailable, check property eligibility and data volume before describing a performance decline. If a request failed, retain the error and last successful collection. For a partially collected period, show the actual cutoff and do not compare it to a fully collected period without a warning in the comparison itself.
Make analysis accountable to the supplied rows
Give an AI assistant the dataset definition and limitations with the rows. Ask it to separate observations, possible explanations and checks that need evidence. A top-page list can show which returned pages changed; it cannot prove why every page on the site changed.
Close an analysis with the supported conclusion, unresolved questions and the next discriminating check. In AISEO Site, a limited snapshot should remain labelled as limited when it is exported or analysed. Neither a confident explanation nor a larger spreadsheet restores data that the source did not provide.
Common mistake: AI automatically fills in explanations for missing data
Reasons without evidence must remain assumptions or unknown.
How to decide when you encounter different situations
Match your actual observation to a row before changing the page. If it does not match, collect the missing evidence first.
| What you see | Next step | How to confirm |
|---|---|---|
| Valid request returns zero | Show true 0 | Preserve request definition |
| Data was not returned | Mark as missing or not applicable | Must not default to zero |
| Request failed | Show error and next action | Do not use simulated results |
Separate a returned zero from a failed or filtered result
Teaching scenario. Names, URLs and numbers illustrate the method; they are not customer or measured site results.
- What you observe
- Teaching example: summary clicks are 100, visible query rows sum to 70, and a separate API request fails.
- What it establishes
- The 30-count gap is not a recoverable list of hidden keywords. The failed request supplies no valid zero; neither gap should be filled with invented rows.
- What to do next
- Preserve summary and visible detail as separate views. Label failed, unavailable, filtered and valid-zero states explicitly, then use only compatible observations for comparisons.
What counts as a checked result
Inspect the state labels and denominators in exports and AI prompts. Missing dates and unknown queries must remain visible limitations in the final analysis.
If a required check fails, keep the error and return to the matching decision branch. A blank record means unverified, not passed.
Leave with a checked result, not another vague instruction
- Your deliverable
- A measurement contract and distinct zero/missing/preliminary/failure states beside every conclusion.
- Stop and resolve this first
- Do not fill unavailable inputs with zero or distribute a hidden-query gap to guessed terms.
Record missing evidence in the worksheet below. An unresolved test is useful information; a blank field is not a pass.
Checklist for this page
Fill in your own check results, then decide the next step. Blank means not yet verified, not normal. Content is saved only in the current local session and is not submitted to the server or AI.
Frequently asked questions
Why does the product show top rows?
Current API and request volume are limited; a limited list is not a complete site-wide query database.
Can AI reconstruct anonymized queries accurately?
No. It can propose hypotheses, but those are not retrieved search records and must not be inserted as facts.
Official materials and usage boundaries
This page provides operational judgment and checking methods, and does not show customer success cases. Calculation examples and request samples are teaching materials, not actual website metrics. Google's reports, conditions, and interfaces may change; please refer to the current official documentation.
Read the corresponding Google official documentationPerformance report definitions and data boundaries ↗
Continue with the relevant tool
Read AISEO Site's visible page and query rows with their coverage limits. Missing queries cannot be reconstructed from the difference between totals.
See how it works →