For the third time in roughly three weeks, Google has confirmed that Search Console’s Discover performance report lost data permanently. A logging error on May 21, 2026 caused clicks and impressions to go unrecorded, and Google has stated the data will not be recovered. For SEOs managing publisher accounts, the incident is not a one-off anomaly; it is the latest point in a pattern that is eroding Search Console’s credibility as a reliable Discover reporting surface.
Google’s statement, reported by Barry Schwartz at Search Engine Roundtable, drew a distinction that matters operationally: “this issue affects data logging only.” The underlying ranking signal in Discover was not disrupted. The site’s content continued to surface to users as normal. What was lost was the record of those interactions, which means any decision made from late-May Discover data in Search Console is built on an incomplete foundation.
That distinction, however, does not eliminate the cost. Discover performance data is used to gauge content resonance, validate publishing cadence, and report traffic trends to stakeholders. Losing a day of that record is not a ranking problem; it is a decision-input problem. The absence of accurate data still breaks the feedback loop publishers rely on.
The May 21 incident follows confirmed logging failures on May 7 and May 8. All three gaps are permanent. Google has not indicated any recovery mechanism for historical Discover data once a logging error closes. The compounding nature of three irrecoverable gaps within twenty-one days creates a statistically unreliable May baseline for any Discover analysis conducted now.
This sits inside a longer reliability arc. Google previously disclosed losing approximately 50 weeks of data on another Search Console surface, a disclosure that drew widespread attention in the SEO community. As of this writing, the Search Console link report is also broken, a separate concurrent issue that Search Engine Roundtable has noted. Two distinct Search Console data surfaces failing at the same time is not a single engineering failure to be explained and moved past.
Discover’s share of publisher referral traffic has grown substantially over the past two years. For some sites, particularly those in news, lifestyle, and content-heavy verticals, Discover now delivers a material portion of overall sessions. The combination of growing Discover dependence and degraded Discover measurement creates a specific reporting risk: publishers may be flying blind on their highest-growth traffic channel at precisely the moment it warrants closer attention.
The operational response for SEOs running publisher accounts is straightforward. Search Console should not serve as the canonical record for Discover performance right now. Server-side logs and Google Analytics 4 session data (filtered by source to capture Discover-referred traffic via the google / discover channel grouping) should be treated as the primary record. Week-over-week comparisons anchored to the May 7-8 or May 21 reporting periods should carry an explicit caveat in any stakeholder report. Presenting May performance without noting the three confirmed data gaps risks mis-attributing content strategy changes to a signal that was simply not recorded.
Google has not released a postmortem explaining the common cause across the three May incidents, nor has it announced a fix or a data-integrity guarantee going forward. The absence of a disclosed root cause means publishers have no basis for assuming May’s pattern will not continue into June.
Pull your GA4 Discover channel data for May now, flag the three gap dates in your reporting, and set Search Console as a secondary source until Google confirms the logging errors have been resolved.
Reported by Search Engine Roundtable (Barry Schwartz), published May 27, 2026.
