Google quietly disclosed on April 3, 2026 — in a 47-word changelog entry on the Data Anomalies page, with no email and no Search Liaison post — that Search Console over-reported impressions from May 13, 2025 onward. For roughly 11 months, your year-over-year organic search reports have been built on a layer Google itself now admits was broken.

That bug is one of two distortions. AI Mode results were blended into Search Console “Web” totals on June 17, 2025, with no native filter to separate them. The fix is two-pronged: switch your primary KPI to clicks (Google says clicks were not affected), then reconcile against first-party server data — where neither defect can reach.

Four Documented Breaks in One Reporting Layer#

Search Console did not quietly degrade once. It degraded on at least four separate dates, each documented, none of them flagged inside the dashboard your team opens every morning.

• May 13, 2025 — the impressions bug starts. Google’s own Data Anomalies page now confirms this as the first day Search Console began over-counting impressions across properties.
• June 17, 2025 — AI Mode merges into “Web.” Search Engine Land reported that AI Mode clicks, impressions, and position were blended into Search Console totals with no filter to separate AI Mode from traditional results. AI Mode now reaches roughly 75 million daily users.
• September 2025 — the &num=100 URL parameter is removed. Rank trackers and SEO tools that relied on viewing 100 results per page lost a long-standing signal, and reported impression patterns shifted again as a side effect.
• April 3, 2026 — the fix begins. Google’s 47-word changelog announces the rollout of a corrected impression count. Stores will see impressions drop after this date — not because traffic changed, but because the inflation finally stopped.

Across 17 analyzed properties (ranging from 17,000 to 5.55 million annual clicks; 2.52 million to 292 million impressions), every single property showed impressions growing faster than clicks during the bug window — a near-mathematical signature that the impression layer was over-counting (Passionfruit Labs research, 2026).

If your YoY content marketing report shows “impressions up 40%, clicks flat,” that is now the expected pattern. It is not a content-marketing victory. It is what an inflated denominator looks like.

Clicks Survived. Impressions and CTR Did Not.#

Google has been explicit that the bug affected impressions, not clicks. That detail matters more than it sounds.

Click counts are tied to actual user actions — a real outbound click from a Google result surface to your URL. Impressions are a Google-side count of “how many times your URL appeared on a result surface.” The first is observable; the second is calculated. Calculations go wrong silently. Clicks do not.

Click-Through Rate inherits the broken denominator. CTR equals clicks divided by impressions, so even if clicks held steady, CTR fell every time the impression count was inflated. Seer Interactive measured organic CTR for queries with AI Overviews dropping from roughly 1.76% to 0.61% — a 61% decline at similar impression counts. Roughly 50% of searches showed an AI Overview by mid-2025 (Safari Digital), so the affected query set was not a small slice.

That single metric movement was used by countless agencies to justify pivots — “AI Overviews are killing CTR, we need to invest in different content.” Some of that may be true. Some of it is the inflated denominator talking. You can no longer tell which from Search Console alone.

You may also be interested in: Gemini Just Overtook Perplexity in AI Referrals — Your WooCommerce Store’s “Direct Traffic” Bucket Is Hiding Three Different AI Platforms

The Real Question Isn’t Impressions Per URL. It’s Revenue Per Landing Page.#

The question isn’t “how many impressions did this URL get?” The question is “how much revenue did this landing page generate?”

The first question routes through Search Console, where two structural defects now live. The second question routes through your WooCommerce database — orders, line items, customer email — joined to organic landing-page entry points captured at the WordPress hook layer. That second question has a clean answer, and Google’s bug never touched it.

For a content-led WooCommerce store, this is a simpler conversation than it sounds. Each blog post or category page has a URL. Visitors land on that URL from organic search. Some of those visitors place orders within a defined attribution window. The math is “orders attributable to this landing page × order value.” That number is generated entirely from data on your server, not Google’s.

Run that calculation per landing page and you are no longer arguing with leadership about whether content is working based on a metric Google has already admitted is wrong. You are reporting on revenue, sourced from your own database.

Reconcile Against First-Party Server Data#

The architectural pattern is straightforward. Capture every page-view event server-side at the WordPress hook layer rather than relying on a browser pixel that ad blockers (31.5% of users globally, per Statista) and Safari’s 7-day cookie window keep dismantling. Capture every WooCommerce order event the same way. Stream both to BigQuery. Join landing page → first organic session → order in SQL.

Now you have a second source of truth that:

• Is not subject to ad blocker loss — events originate on your server, not in the browser
• Is not subject to Safari’s 7-day cookie window — first-party session storage on your subdomain
• Is not subject to Search Console’s blended AI Mode totals — referrer-level data is yours
• Is not subject to whatever Google quietly miscounts next — the data lives in your warehouse

Annotate the four bug-window dates on every dashboard you ship to leadership: May 13, 2025; June 17, 2025; September 2025; and April 3, 2026. Then export a snapshot of pre-fix Search Console data this week — once it’s gone, agencies will be reasoning about a moving baseline forever.

You may also be interested in: WooCommerce Events to BigQuery Without GA4: The Direct Pipeline

Here’s How You Actually Do This#

Transmute Engine™ is a first-party Node.js server that runs on your own subdomain (e.g., data.yourstore.com). The inPIPE WordPress plugin captures WooCommerce hooks and sends events via API to your Transmute Engine server, which streams them simultaneously to GA4, Facebook CAPI, and BigQuery — where you can join organic landing pages to actual orders without ever asking Google’s reporting layer for the truth.

That’s the architectural move. The impressions bug couldn’t reach first-party server data, the AI Mode merge can’t reach it either, and whatever Google miscounts next won’t reach it.

Key Takeaways#

• Search Console over-reported impressions from May 13, 2025 through early April 2026. Google disclosed it in a 47-word changelog entry on April 3, 2026 — no email, no announcement post.
• AI Mode has been blended into Search Console “Web” totals since June 17, 2025 with no native filter. “Impressions” now bundles traditional organic, featured snippets, AI Overviews, and AI Mode together.
• Clicks were not affected. Impressions and CTR were. Switch primary KPI to clicks for the foreseeable future and treat impression-based comparisons as directional only.
• Annotate the four discontinuity dates on every YoY dashboard: May 13, 2025; June 17, 2025; September 2025; April 3, 2026. Export a pre-fix snapshot before April data fully replaces the historical curve.
• Reconcile organic performance against first-party server data in BigQuery. Revenue per landing page is the layer Google’s bug couldn’t touch — and the only one your leadership deck can actually defend.

Frequently Asked Questions#

Build your reporting on data Google can’t quietly break. Visit seresa.io to see how Transmute Engine puts revenue per landing page next to organic discovery — without GTM.