Every monthly report your agency sent since May 2025 included a lie you didn't know you were telling.
On April 3, 2026, Google quietly updated its Data Anomalies page with a 47-word statement confirming that a logging error had been inflating impression counts in the Search Console Performance report since May 13, 2025. That's 11 months of inflated data in the single most widely used organic search measurement tool in the world.
The GSC impressions bug didn't touch clicks. It didn't change your rankings. But it poisoned every impression-based KPI, every CTR calculation, every visibility trend line, and every client report built on Search Console impressions during that window. And if your agency used impressions as a primary success metric, you now have 11 months of client conversations to revisit.
I'm not writing this to pile on Google. Bugs happen. I'm writing this because most of the coverage explains what happened. Almost none of it tells agencies what to do about the reports they already sent.
Let's fix that.
What Exactly Is the GSC Impressions Bug?
A logging error caused Google Search Console to over-report impressions starting May 13, 2025. Google confirmed the issue on April 3, 2026 and is rolling out corrections over several weeks. Clicks and other engagement metrics were not affected. The decrease you see in impressions reflects more accurate data, not a drop in actual search visibility.
Google's official statement on its Data Anomalies page read: a logging error prevented Search Console from accurately reporting impressions from May 13, 2025 onward. A Google spokesperson told Search Engine Land that bug fixes are being implemented to ensure accurate reporting.
Here's the critical part most people miss: Google did not disclose the magnitude of the inflation. We still don't know if impressions were inflated 5% or 50%. The correction is rolling out progressively, not in a single clean reset. Some properties may see changes sooner than others.
SEO consultant Brodie Clark actually flagged anomalies across merchant listings and Google Images filters on March 30, 2026, four days before Google's acknowledgment. Glenn Gabe at G-Squared Interactive raised the alligator-mouth pattern question: impressions surging while clicks stayed flat. Many in the industry had attributed that divergence to AI Overviews. Turns out at least part of it was a bug.
Why Is This Worse Than a Normal Data Anomaly?
This isn't a one-week glitch. This is 11 months of corrupted denominator data layered on top of two other measurement disruptions that happened during the same window. Any impression trend line spanning mid-2025 to mid-2026 contains at least five documented discontinuities.
Here's the timeline every agency needs to annotate:
| Date | Event | Impact |
|------|-------|--------|
| May 13, 2025 | Logging bug begins | Impressions start inflating across all properties |
| June 17, 2025 | AI Mode data merged into GSC | New impression/click stream blended into "Web" with no filter |
| Sep 12, 2025 | &num=100 parameter killed | Impressions from third-party crawlers removed, causing sudden drop |
| Oct 19, 2025 | GSC data freeze | Performance reports stopped updating for several days |
| April 3, 2026 | Google confirms bug and begins fix | Impressions start correcting downward over several weeks |
The layering effect is what makes this so damaging. The September &num=100 shutdown masked part of the logging bug's inflation. If both problems ran simultaneously from May through September, the September correction only peeled away one layer while the other continued undetected for another seven months.
And then there's the AI Mode problem.
How Does AI Mode Data Make This Worse?
On June 17, 2025, Google began merging AI Mode clicks, impressions, and position data into Search Console's Performance report under the existing "Web" search type. As of April 2026, there is still no native filter to separate AI Mode data from traditional organic results.
Glenn Gabe summarized the situation directly: "We now have the 10-blue links, featured snippets, AIOs, and now AI Mode all grouped together in the performance reporting under Web Search. Good luck trying to figure this out in GSC."
AI Mode now reaches 75 million daily active users globally. That is not a small data stream being blended in. Position data is calculated differently for AI Mode surfaces. Follow-up questions reset positions to 1. The SERP you think you're measuring and the one GSC is actually reporting on are not the same thing.
So the trust problem compounds: inflated impressions from a logging bug, plus a growing AI Mode data stream you can't isolate, plus the ghost of the &num=100 parameter shutdown still affecting year-over-year comparisons. Three layers of noise in one metric. And agencies have been reporting that metric to clients as proof of progress.
What Should Agencies Do Right Now?
Stop treating this as a "Google problem." Your clients don't care that Google had a bug. They care that the numbers you showed them were wrong. This is a trust-repair exercise, and the agency that communicates proactively wins.
Here's the action plan:
Step 1: Export Your Pre-Fix Data Today
Capture daily data from every GSC property before the fix fully propagates. You need the pre-fix numbers to calculate property-specific inflation once the correction stabilizes. This data will not be available again after impressions normalize. Use the Search Console API if you manage more than a handful of properties, since the GSC interface caps exports at 1,000 rows.
Step 2: Annotate Every Dashboard
Add these four annotations to every reporting dashboard that touches Search Console data:
- May 13, 2025 - GSC impressions logging bug begins
- June 17, 2025 - AI Mode data merged into Web Search totals
- Sep 12, 2025 - &num=100 parameter killed (impression baseline shift)
- April 3, 2026 - Google confirms bug, fix rollout begins
Anyone reviewing historical data needs to understand these discontinuities at a glance. Without annotations, the impression drop looks like an SEO failure. With annotations, it's a measurement correction.
Step 3: Send a Proactive Client Communication
Don't wait for clients to see the drop and panic. Get ahead of it. Your communication should cover:
1. What happened: A confirmed Google logging error inflated impressions for 11 months
2. What it means: Reported impressions were higher than reality. Clicks, traffic, and conversions were unaffected
3. What you're doing: Cross-referencing click data with GA4 sessions to verify actual performance was stable
4. What changes going forward: Shifting primary reporting KPIs from impressions to clicks and conversions
Keep it factual. Link to Google's Data Anomalies page. Show your GA4 organic sessions alongside GSC clicks to demonstrate that actual performance didn't change.
The agencies that send this email before the client asks the question are the ones that keep the client.
Step 4: Re-Baseline Your CTR Benchmarks
Every CTR analysis conducted between May 2025 and the full correction is based on an inflated denominator. CTR = clicks / impressions. If impressions were artificially high, CTR appeared artificially low.
Once the fix fully propagates:
- Pull 30 days of clean post-fix data
- Compare click counts (stable) against corrected impressions
- Recalculate CTR benchmarks per page, per query cluster
- Use these new benchmarks as your baseline going forward
Do not try to retroactively "correct" the 11-month window. The inflation magnitude varied by property, and you'll never get a clean number. Draw a line, move forward with accurate data.
Step 5: Run a Full Site Audit Against Clean Data
With impressions normalizing, now is the time to run a complete SEO audit against data you can actually trust. Your click data was accurate the entire time. Cross-reference it with your corrected impression baselines to identify pages that were genuinely underperforming versus pages that just looked bad because of inflated impression denominators.
A free audit can give you a quick snapshot of where your technical SEO stands while you work through the data cleanup.
Why Should Impressions Never Be Your Primary Agency KPI Again?
Impressions are a derived metric from a system you don't control, measured by a tool that just proved it can be wrong for 11 months without anyone at Google noticing. That's the sentence every agency owner needs to sit with.
Impressions tell you about potential visibility. Clicks tell you about actual engagement. Conversions tell you about business impact. The further up that chain you build your reporting, the more vulnerable you are to measurement failures outside your control.
Here's my reporting hierarchy recommendation for agencies going forward:
1. Revenue / leads from organic - the only metric clients truly care about
2. Organic clicks - stable, verified, unaffected by this bug
3. Click-through rate - only useful when denominator (impressions) is trustworthy
4. Impressions - directional signal only, never the headline metric
5. Average position - context metric, not a KPI
Impressions still have value as a directional indicator. But the moment you put "impressions grew 23% this quarter" as the lead stat in a client report, you're building your credibility on a foundation you can't verify.
What Does This Mean for Year-Over-Year Reporting?
Any YoY comparison that touches May 2025 through the full correction window is unreliable for impressions. Comparing Q2 2026 against Q2 2025 means comparing corrected data against inflated data. You'll see a decline that isn't real.
The practical approach:
- For impressions: Skip YoY comparisons during the affected window. Use month-over-month within the post-fix period once data stabilizes
- For clicks: YoY comparisons remain valid. Clicks were unaffected. Use these as your continuity metric
- For CTR: Recalculate only against post-fix data. Historical CTR from the affected window is unreliable
- For traffic and conversions: GA4 organic session data and conversion data are completely unaffected. Lead with these
The healthiest thing an agency can do right now is lean into the metrics that were accurate all along and stop apologizing for the ones that weren't your fault.
The Bigger Lesson: Single-Source Dependency Is a Risk
I've been saying this since I started building Vantacron: relying on a single data source for business decisions is a structural weakness. This bug proved it.
The combination of the impressions logging error, the AI Mode data merge, the &num=100 shutdown, reporting freezes, and missing bulk export data across 2025-2026 represents the most turbulent stretch for GSC data quality in recent memory.
The fix isn't abandoning GSC. It's the most important free SEO data source available. The fix is triangulating: GSC clicks + GA4 sessions + conversion data + rank tracking from independent tools. When one source has an anomaly, the others catch it.
If your reporting stack collapses because one tool has a bug, that's not a Google problem. That's an architecture problem.
Checklist: GSC Impressions Bug Response for Agencies
- [ ] Export daily GSC data from all properties before fix fully propagates
- [ ] Add annotations at May 13 2025, June 17 2025, Sep 12 2025, and April 3 2026
- [ ] Send proactive client communication explaining the bug and your response
- [ ] Cross-reference GSC clicks with GA4 organic sessions to verify real performance
- [ ] Pause all impression-based YoY comparisons for the affected window
- [ ] Shift primary reporting KPIs to clicks, conversions, and revenue
- [ ] Recalculate CTR benchmarks once 30 days of clean post-fix data is available
- [ ] Run a technical audit against clean data to identify genuine performance issues
- [ ] Document your response process for future data anomaly events
- [ ] Review your reporting stack for single-source dependencies
Frequently Asked Questions
How long did the GSC impressions bug last?
The GSC impressions bug ran for approximately 11 months, from May 13, 2025 through Google's confirmation and fix rollout starting April 3, 2026. During this entire period, impression counts in the Search Console Performance report were inflated due to a logging error. Clicks and other engagement metrics were not affected by the bug.
Were Google Search Console clicks affected by the impressions bug?
No. Google explicitly confirmed that clicks and other metrics were not affected by the logging error. Only impression counts were inflated. This means your organic traffic data in both GSC and GA4 remained accurate throughout the 11-month window. Agencies should use click data as their reliable performance baseline during this correction period.
How should I explain the GSC impression drop to clients?
Tell clients directly: Google confirmed a logging error inflated impression counts for 11 months, and the correction means impressions are returning to accurate levels. Show them that clicks and GA4 organic sessions remained stable. Frame the drop as improved measurement accuracy, not declining performance. Proactive communication before clients notice the drop builds trust.
Can I trust Google Search Console impressions going forward?
Once the fix fully propagates over the coming weeks, impression data should return to accurate reporting. However, AI Mode data remains merged into Web Search totals with no filter to separate it, which adds ongoing complexity. Use impressions as a directional signal, but anchor your primary KPIs on clicks, conversions, and revenue from organic search.
What is the alligator graph pattern in GSC?
The "alligator graph" describes a pattern where GSC impressions surge upward while clicks stay flat or decline, creating a shape resembling an open alligator mouth. Many SEOs attributed this to AI Overviews reducing click-through rates. The confirmed logging bug reveals that at least part of the impression surge was artificial inflation from the measurement error, not just behavioral change.