Did your Google Search Console impressions fall sharply in mid-September 2025 while clicks held steady? You are not alone, and you have probably not lost rankings. The most likely cause is that Google stopped supporting the num=100 parameter. Rank trackers and scrapers used it to load 100 results on one page. Fewer of those automated page loads means fewer impressions logged for results far down the list.
This guide explains what changed and why it shows up in your reports. It also covers how I’d read and present your Search Console data from now on.
What Happened to the num=100 Parameter
For years you could add &num=100 to the end of a Google search URL. You then got up to 100 organic results on one page instead of the usual ten. Google took the results-per-page option out of its visible settings long ago, but the URL parameter kept working.
Around 10 September 2025, people across the SEO industry noticed it had become unreliable. Within days it had stopped working for most searches. Google ignored the parameter, and you got a normal page of results.
What Google Said
Google made no announcement, blog post or help page update. When the trade press asked in mid-September, a Google spokesperson gave a one-line reply: “The use of this URL parameter is not something that we formally support.”
That statement confirms the parameter is not a supported feature. It doesn’t explain why it stopped working, and it says nothing about Search Console. So any explanation of the reporting changes is a reading by people in the industry, not a confirmed statement from Google. I always make that difference clear when I talk to clients or managers.
Who Actually Used It
Almost no normal searchers typed &num=100 into their browser. The heavy users were automated systems:
- Rank tracking tools that check where a site ranks for thousands of keywords every day.
- SEO platforms that build keyword databases and visibility scores from search results.
- Scrapers and data collectors, including some that gather search results for research or AI products.
One request with &num=100 gave a tool the top 100 results at once. Without it, getting the same depth takes ten separate requests, one per page of results. That is ten times the requests for the same data. It’s why many tool providers spent September cutting tracking depth, slowing updates or changing their prices.
Why Search Console Impressions Dropped After the num=100 Change
To understand the drop, you need to know how Search Console counts an impression (one time your result was shown). Google’s help page on how impressions, position and clicks are calculated explains it. For a standard result, an impression counts when the result appears in the current page of results. It doesn’t have to be scrolled into view.
Now picture a rank tracker loading a 100-result page for one of your keywords. Say your page sat at position 64. It was on the “current page of results”. So that load could be recorded as an impression for your site at position 64. No human ever saw it.
Multiply that by every tool checking every keyword every day. Sites with lots of deep rankings were collecting a large number of impressions that came from machines, not people.
Why the Drop Shows Mostly on Desktop
Most rank trackers and scrapers ask for desktop results. So when the parameter stopped working, the effect was mostly in desktop impressions. Many site owners saw desktop impressions fall from around 10 to 12 September, while mobile impressions looked mostly the same.
Why It Is a Theory, Not a Confirmed Fact
This explanation fits the timing and the pattern well. Most experienced SEOs have settled on it, and I agree with them. But Google has not confirmed that Search Console was counting impressions from automated 100-result pages. Its help pages also don’t say how bot traffic is handled in the Performance report.
So here is the honest position. The drop lines up with the num=100 change, and the cause makes sense. Clicks and real traffic don’t show a matching fall. That is strong evidence of a reporting change, not a ranking loss.
What Changed in Each Search Console Metric
Each number in the Performance report reacted differently. Here is what to expect if your site was affected.
| Metric | Typical change after mid-September 2025 | Why |
|---|---|---|
| Impressions | Down, mostly on desktop | Fewer automated 100-result page loads showing deep-ranking results |
| Clicks | Broadly steady | Bots loading result pages were not clicking; human behaviour did not change |
| Average position | Improves (number gets smaller) | Deep positions like 40 or 80 drop out of the average |
| CTR | Rises | Same clicks divided by fewer impressions |
| Number of queries | Often down | Queries where you only ranked deep may no longer record any impression |
Why Average Position “Improved”
Average position only counts queries where your site got an impression. It uses the highest position your site held. Before the change, a tracker’s 100-result page could log an impression for you at position 70. That pulled your average down.
Once those deep impressions go away, the average comes mostly from positions real people see, usually on the first page or two. The number looks better, but your rankings didn’t move. Treat a sudden jump in average position from mid-September as a side effect of the data, not a win.
Why Fewer Queries Appear
Say your site ranked only at position 50 for a long-tail query (a longer, more specific search). A human searcher would rarely go that far. The only “views” were automated ones. Without them, that query may no longer record any impressions. So it drops out of your queries table completely.
This can make it look like you lost keyword coverage. In most cases, you only lost report visibility for rankings that never sent visitors anyway.
How to Read Search Console Data Now
I’d argue the data after mid-September is closer to what humans actually see. The problem is comparing it with the period before. Here is how to handle it.
Treat mid-September 2025 as a New Baseline
Comparisons that cross 10 September 2025 are not like-for-like for impressions, average position, CTR or query counts. Month-over-month reports for September versus August will show a decline that didn’t really happen that way. So will year-over-year reports for the next twelve months.
For trend work on those numbers, start your comparison window after the change. Or compare against a matching period after the change once you have one.
Lean on Clicks for Performance
Clicks were the least affected number, because bots were not clicking. If clicks are steady or rising, your organic performance is very likely fine, whatever impressions say. Clicks are the first number I check in a case like this.
Back this up with your analytics tool. Organic landing page sessions, leads and sales give you a separate check. That check doesn’t depend on how Search Console counts impressions.
Segment by Device
Use the Devices tab or a device filter in the Performance report. Did desktop impressions fall sharply around 10 to 12 September while mobile impressions and desktop clicks held steady? Then you are looking at the num=100 pattern, not a ranking problem.
If mobile impressions and clicks also fell at the same time, look deeper. Something else may be going on, like an algorithm update, a technical issue or seasonal demand. Google’s Search Status Dashboard lists confirmed updates and incidents, so I check it for the same dates.
Watch Out for False Alarms in the Other Direction
The change can also hide real problems. If clicks dropped at the same time as impressions, don’t wave it away as “just the num=100 thing”. Check indexing, recent site changes, and the pages that lost clicks before you rule out a real decline. I see this mistake often: one well-known cause gets blamed for everything.
If you are newer to the Performance report and its filters, our beginner’s guide to the Search Console reports that matter walks through each one.
How to Tell a Reporting Change From a Real Ranking Drop
Here is a quick check you can run in about ten minutes.
- Open the Performance report and set the date range to cover about four weeks before and after 10 September 2025.
- Turn on all four numbers: clicks, impressions, CTR and average position.
- Look at the shape of the lines. The num=100 pattern is a step down in impressions around 10 to 12 September, with flat clicks, rising CTR and a better average position.
- Filter by device. Check that the drop in impressions is mostly on desktop.
- Check the Queries tab with a before-and-after comparison. The queries that vanished should mostly be ones where your position was deep, often past the first two pages.
- Cross-check organic sessions and conversions in your analytics tool for the same dates.
If all of that lines up, you can be confident the change is in the reporting, not your rankings.
A Practical Example
Imagine a home services business. Its site ranks on page one for a few dozen local terms, and between positions 30 and 90 for hundreds of broader terms. Before September, every rank tracker following those broader terms was loading 100-result pages that included the site.
After the change, desktop impressions fall clearly. The queries table loses many of those broad terms. Average position jumps from somewhere in the 30s to the low teens. Yet calls and form fills from organic search are where they were. Nothing about the business’s real visibility changed.
How to Annotate Your Reports
Notes stop the same question coming up every month. They also protect you from bad comparisons a year from now.
Where to Add a Note
- Dashboards built in Looker Studio or similar tools: add a text note or a reference line at 10 September 2025 on any chart that shows impressions, average position or CTR.
- Analytics: if your analytics tool supports notes, add one there too, so anyone looking at organic trends sees it.
- Spreadsheets and monthly reports: add a footnote to any table that compares periods across the change.
- Your own change log: keep a simple list of dated events (site launches, algorithm updates, tracking changes) that everyone on the team can see. I keep one for every site I work on.
What the Note Should Say
Keep it short and factual. For example:
10 Sep 2025: Google stopped supporting the num=100 results parameter.
Search Console desktop impressions fell and average position rose as
automated 100-result page loads stopped. Clicks unaffected.
Impression, CTR and position comparisons across this date are not like-for-like.Don’t add numbers you can’t back up, like “impressions were inflated by X%”. You don’t know how much of your old impression count came from bots. Nobody outside Google does either.
How to Explain the num=100 Parameter Change to Clients
Clients and managers often see impressions first and panic. A calm, plain explanation usually settles it.
A Simple Way to Put It
Try something like this:
- “Google changed how many results automated tools can load at once. Those tools used to trigger impressions for our site in positions no real person was looking at.”
- “That stopped in mid-September, so the impressions figure dropped. Our clicks, traffic and enquiries did not.”
- “Our average position looks better for the same reason. It is a reporting effect, not a ranking improvement, so we will not claim it as one.”
Being open about the position “improvement” builds trust. It shows you are not picking only the numbers that make you look good. In my experience, this honesty matters more to clients than the numbers themselves.
Change What You Report On
This is a good moment to stop using impressions as a headline number. Lead with clicks, organic sessions, conversions and revenue. Then use impressions as background.
If you are rethinking the full set of numbers you share, our guide on choosing SEO KPIs to measure and report covers how to build a report around results.
Expect Questions About Rank Tracking Too
Third-party rank tracking has also changed. Some tools now check less deeply, update less often, or charge more for the top 100. If a client’s rank tracking report suddenly shows fewer keywords or gaps past the top 20, that is likely a tool change, not lost rankings.
Ask your tool provider what depth and frequency you get now, and tell clients if their reports will look different.
What This Means for SEO Going Forward
The practical lesson is that impressions were always a noisier number than they looked. The num=100 change removed one known source of noise. The data you see now is likely a closer picture of real people seeing your results.
A few sensible changes I’d make:
- Focus on the first two pages. Rankings past position 20 rarely send traffic. If your tools now track less deeply, you lose little of real value.
- Don’t chase lost queries. Long-tail terms that vanished from Search Console were usually deep rankings with no clicks. Only go back to them if they matter to the business.
- Review tool costs and depth. Maybe you pay for tracking you no longer get, or track hundreds of keywords no one acts on. If so, this is a good time to cut back.
- Keep Search Console as your main source for clicks. It is Google’s own data on real clicks from search. That makes it more reliable than any third-party estimate.
Google also hasn’t said whether it will make other changes that affect automated access to results. I’d assume that scraping search results will get harder and more expensive, and plan your reports around your own data. Google already made this harder in January 2025, when Google Search started requiring JavaScript.
Conclusion
The end of the num=100 parameter in September 2025 cut off a large stream of automated page loads. In Search Console, that showed up as lower desktop impressions, a better-looking average position and fewer queries. Google’s only comment was that it does not formally support the parameter. For most sites, clicks and real traffic didn’t change.
Your next step: run the device and date comparison in your own Performance report. Add a note at 10 September 2025. Then update your next client or management report so it leads with clicks and conversions, not impressions.
