Currently Available: Need a skilled Software Developer for your next project?
Categories
SEO, GEO and AEO

Handling Failed Search Console BigQuery Bulk Exports

Google Search Console writes search performance data to BigQuery, Google’s service for storing and analyzing data, through its daily bulk export. When a day appears to be missing, check the ExportLog table first. It records successful writes, including the activity date and destination table. Search Console exports its site-level and URL-level performance tables separately, so check coverage for each one.

A missing log entry means Search Console has no record of a successful export for that table and date. The log does not explain why the export failed. Search Console retries some failures, and it can later revise data for a date that already had a successful export. Track the activity date and the time Search Console last wrote data. These dates and times help you tell a gap from a late retry or a revision.

Check the export log before treating a date as failed

Search Console schedules exports daily, but timing varies, without publishing a fixed completion time. Search Console can write the site-level table (searchdata_site_impression) and URL-level table (searchdata_url_impression) at different times. Their dates use Pacific Time. Investigate a missing date only after your chosen grace period, the time you allow for an export to arrive, has passed. Google does not specify a freshness service-level agreement or a required alert threshold.

Check ExportLog to see whether Search Console recorded a successful save for the expected date and table. An impressions table with no rows for that date does not prove the export failed. The log directly records successful saves. The table reference also explains that failed attempts are not recorded in ExportLog, so the log cannot show the failure reason.

If a date appears to be missing, check Search Console’s export status for the latest attempt. Then search Cloud Logs Explorer for search-console-data-export. The status page may show details for a non-transient error, but it does not provide a complete history of failed attempts. Search Console sends email about export errors and their resolution, but it does not email about errors it considers transient. Treat email as an alert, not as a full record of export errors. Google’s export monitoring guidance describes these diagnostic limits.

The query below compares successful log entries with the dates you expect for both tables. Replace the project and dataset names, set first_date to the start of your monitoring period, and check the namespace values in your ExportLog before using the table names shown. The three-day buffer is an example monitoring choice, not a Google delivery deadline.

DECLARE first_date DATE DEFAULT DATE '2026-11-01';
DECLARE through_date DATE DEFAULT
  DATE_SUB(CURRENT_DATE('America/Los_Angeles'), INTERVAL 3 DAY);

WITH expected AS (
  SELECT
    namespace,
    data_date
  FROM UNNEST([
    'searchdata_site_impression',
    'searchdata_url_impression'
  ]) AS namespace
  CROSS JOIN UNNEST(
    GENERATE_DATE_ARRAY(first_date, through_date)
  ) AS data_date
),
successful AS (
  SELECT DISTINCT namespace, data_date
  FROM <code>your-project.your_dataset.ExportLog
  WHERE agenda = 'SEARCHDATA'
)
SELECT
  e.namespace,
  e.data_date AS missing_successful_export_date
FROM expected AS e
LEFT JOIN successful AS s
  USING (namespace, data_date)
WHERE s.data_date IS NULL
ORDER BY e.data_date, e.namespace;

Use the query results to find dates to investigate; they do not explain why an export is missing. Check each table on its own, then check Search Console status and Cloud Logs Explorer for details about the latest error. For more on the exported fields, see the BigQuery Search Console schema.

Retries have a limited window

For a temporary error, such as a connection problem, Search Console retries immediately and does not send a notification. When an error needs a fix, such as a permissions problem, Search Console does not immediately retry that table’s write. It tries again on the next scheduled daily export and keeps the missed date’s data for about a week.

The week is an approximate period for recovering missed data. Search Console does not promise to export older data indefinitely. In Google’s documented example, access returns on day nine. Search Console no longer exports data from the first two days, but it can export later dates after access returns. The management guidance also says that continued failures for about a month can cause Search Console to stop the bulk export entirely. These are approximate periods, not exact deadlines.

Use the reported error to decide what to check. Google lists issues involving quota, billing, permissions. It also lists problems with dataset location, schema, and partition expiration. It recommends partition expiration of at least 14 days; a shorter setting can prevent an export to a partition that has expired or is about to expire. For Deadline exceeded, Google says the cause is probably on its side. For Unknown error, it advises waiting for the next scheduled export to see whether the issue resolves.

After you fix a reported configuration problem, use Search Console’s Test report to check some settings. A successful test does not run an export or check every condition, including quota. Confirm recovery by checking for the next successful ExportLog entry. The Search Console export error guidance describes the limits of the test and the listed errors.

Older dates can be revised

A retry exports data for a date that had no successful export. A revision updates data for a date Search Console already exported, for example after Search Console corrects a processing error. Search Console commits a revision atomically, meaning it updates all data for that day in one commit. The revision appears in ExportLog and normally does not trigger an email.

Three fields help explain what happened:

  • data_date is the date of the Search Console activity, in Pacific Time.
  • namespace identifies the table Search Console wrote.
  • epoch_version is 0 for the first successful save of a table and date; Search Console increases it when it updates that table and date.
  • publish_time records when the write completed, also in Pacific Time.

A newer publish_time alone does not mean Search Console delivered a newer activity date. It could reflect a revision to an older date. When you investigate a late write, compare data_date, namespace, and epoch_version to distinguish a missing-date retry from a revision.

Monitor coverage and write time separately

A useful check of how current the exports are answers two questions: what is the latest date each table covers, and when did Search Console last write data? For each namespace, track the latest data_date and the latest publish_time separately. After your grace period, check for missing dates and for changes to epoch_version on older dates.

If one table has a gap and the other is current, investigate the table with the gap. Search Console can write the two tables at different times, so a gap in one does not mean both exports failed. If both tables stop advancing, check the latest Search Console status and Cloud Logs Explorer. If a date is still missing from ExportLog after about a week, restoring access might not recover it. The documented retry window is limited, and Google does not describe a supported way to manually export dates it has dropped.

What I'm building

Delegate tasks. Get software.

Give Vroni a GitHub issue, bug report, spec, or rough idea. It reads the repo, plans the change, writes code, runs checks, and works toward a review-ready pull request.

Take a look at vroni.com

Email updates

Usually a new article and a few links I found interesting.

No spam. Unsubscribe with one click.

Leave a Reply

Your email address will not be published. Required fields are marked *