Why Google Search Console impressions fell (and why that’s good)

In mid-September, many SEO professionals began noticing unusual drops in Google Search Console (GSC) data. 

Impressions were down, average positions shifted, and the number of reported queries changed overnight. 

This wasn’t the result of a ranking update but a reporting change – one that redefines how we interpret GSC visibility data going forward.

What happened to GSC impressions

The drop traces back to Google’s quiet decision to stop supporting the &num=100 parameter, which allowed tools and crawlers to retrieve up to 100 results per query. 

Once support ended, those extra data points disappeared – and with them, many of the impressions and positional averages SEOs had been tracking.

While search results themselves didn’t change, GSC reporting did. 

Third-party platforms lost signal beyond Positions 1–20, and GSC data recalibrated to reflect real user activity rather than automated crawler noise.

Why this change makes sense

As I’ve spoken with SEOs and reviewed multiple sites, it’s clear we need to reset how we interpret impressions: treat the data you see now as the new normal.

Discontinuing support for the &num=100 parameter makes sense. 

I saw the same principle in action at my community garage sale last weekend – fewer choices led to better results.

It’s also easier for Google to show 10 or 20 results per search than to generate 100. 

While this shift disrupted tools built on the now-unsupported parameter, it didn’t affect real users. 

It did, however, catch the attention of SEOs who monitor every fluctuation in how Google reports data.

We saw no visible changes in the SERP, but Search Console told a different story:

• A drop in impressions.
• Changes to average position.
• An increase in the number of queries ranking in Positions 1–20.

Those shifts changed how we think about GSC data.

The alligator effect

Earlier in the year, we observed another pattern – the alligator effect. 

Starting around February, GSC charts showed rising impressions and steady clicks, creating a shape that resembled an open alligator’s mouth. 

Many assumed the growth was tied to AI Overviews and an increase in zero-click searches.

When Google ended support for the &num=100 parameter on September 12, that alligator finally closed its mouth.

Experts now believe automated crawlers inflated impression counts. 

The post-change drop reflects a more accurate baseline and changes how impressions should be interpreted going forward.

Some impressions may even have reflected exposure in large language models (LLMs), like ChatGPT, that used third-party tools to scrape Google results. 

While unconfirmed, it suggests those inflated impressions were signals of visibility – just not on Google’s SERP.

Dig deeper: SEO in the black box era: Why reports will look more like Mad Men than Search Console

What this means for reporting

Despite the fluctuations, GSC remains the most reliable source for keyword ranking data – especially now that third-party tools no longer capture results beyond Positions 1–20.

For most reporting contexts, annotate the measurement change in GSC and use the current impression and average position levels as your new baseline.

Suggested annotation:

• “The data reported in Google Search Console (GSC) from 9/13/2025 onwards is the most accurate accounting of how your brand appears in Google organic search. GSC stopped supporting the &num=100 parameter around 9/12/2025 resulting in impressions generated by third-parties and automated crawlers being removed from reporting.”

If impressions or average position are used in broader models or long-term trend analyses, adjustments between Feb. 1 and Sept. 12, 2025, may be appropriate.

Option 1: Simple method

Use prior-year impressions and average position up to the start of volatility in early February. 

When impressions began to rise sharply (Feb. 1–Sept. 12, 2025), revert to prior-year values, then use GSC data as reported from Sept. 13 onward.

Option 2: Advanced method

Rebuild historical impressions and average position using trend data and adjustment factors. Evaluate:

• Differences in fluctuation by query type or ranking position (branded, non-branded, long-tail).
• Metrics such as GSC clicks, which were less correlated with impressions during the affected period.
• Other factors unaffected by impression volatility.

What to expect from GSC going forward

Treat the post-change figures as the new baseline. The numbers you see now reflect how GSC will report visibility from here on.

• Impressions will stabilize at lower levels compared with earlier 2025 trends.
• Average position will level out, since it’s calculated relative to impressions.
• The number of unique queries reported in Positions 1–20 will hold steady. With less long-tail data, GSC will show more queries ranking in those positions.
• Clicks and traffic should remain consistent, confirming that user engagement with your listings hasn’t changed.

Why impressions and average position still matter

Even with the recalibration, impressions and average position are still key indicators of visibility and progress – just measured more accurately than before.

• Consistency: GSC metrics now reflect real search activity instead of automated crawler data, making them more reliable for SEO measurement.
• Visibility tracking: Still useful for identifying when optimizations begin to take effect.
• Stability: Data for keywords ranking in the top 20 positions show steadier trends than long-tail terms, since positions beyond 20 are no longer captured.

A clearer baseline for visibility

The impression and average position levels you see now in GSC represent a more accurate view of real user search activity. Treat these as your baseline for reporting.

If impressions or average position are used in models or performance controls, normalization methods can adjust historical data. But in most cases, it’s best to move forward with current figures as-is.

Google simply made fewer results available. Searchers still find what they’re looking for, and third-party tools will adapt.