Direct Answer
Until around September 2025, adding &num=100 to a Google search URL returned up to 100 results in a single request. Google removed it, so rank trackers now paginate roughly ten requests to cover the same top 100 — about 10x the request volume for deep tracking. That raises IP-quality and pacing pressure, so pace politely, spread requests, and track google.de from a real German mobile vantage point.
What &num=100 Was, and What Changed
For years, &num=100 was a quiet workhorse of rank tracking. Appending it to a Google search URL asked Google to return up to 100 organic results on one page instead of the default ten. One request, one hundred positions — that is how most trackers read the full top 100 for a keyword in a single call.
Around September 2025, Google stopped honouring the parameter. Add &num=100 today and you get the standard ten results back, the same as if you had never set it. There was no announcement and no replacement; the lever simply stopped working.
The immediate consequence is arithmetic. To still see positions 1 through 100, a tracker has to walk the pages: page one for results 1–10, page two for 11–20, and so on to page ten. One request became roughly ten for the same depth of coverage. If you track deep — full top-100 history across a keyword set — your request volume went up by about an order of magnitude overnight.
The Search Console Side Effect
The change had a second, noisier symptom. Around the same time, a large number of sites saw their desktop impressions drop in Google Search Console, often sharply, without any matching drop in clicks.
The widely shared explanation is that a lot of those impressions were never human. Automated tools loading deep result pages — page after page, out to position 100 — were registering as impressions for every listing they pulled into view. Once &num=100 stopped working, those deep automated loads fell away, and the inflated impression counts fell with them. Average-position numbers shifted too, because the long tail of rarely-seen deep rankings stopped being counted.
The practical takeaway: if your Search Console impressions dipped in autumn 2025 while clicks held steady, that is most likely a measurement artefact, not lost visibility. Treat pre- and post-change impression data as two different baselines rather than a trend.
We say "widely shared explanation" deliberately — Google did not publish a full account, so the mechanism above is the industry's best reconstruction, not a confirmed statement.
What This Means for Your Tracking Setup
More requests for the same result is the whole story, and it lands on three things at once.
- Volume. Deep top-100 tracking now costs roughly 10x the requests it used to. A keyword set that was comfortable at one request each is now ten times heavier on whatever collects it.
- IP quality. Ten paginated requests per keyword, repeated across a keyword list on a schedule, is a much stronger pattern than one request was. Weak or shared datacenter addresses that used to slip through get rate-limited or blocked faster — see why datacenter proxies get blocked and what is IP reputation.
- Pacing. Firing ten pages back-to-back per keyword, as fast as the code allows, is exactly the shape that triggers 429 Too Many Requests. The old one-shot method hid this problem; pagination exposes it.
None of this is a reason to panic, and none of it is fixed by trying to "get around" anything. The honest response is to send fewer, better-paced requests from IPs that look like real users, and to accept that deep tracking is now a heavier job than it was.
How to Adapt: Pace, Spread, and Use a Real German Vantage Point
Pace requests politely
The single most effective change is to slow down. Add a short, slightly randomised delay between paginated requests instead of hammering pages one after another. A rank check does not need to finish in milliseconds; a keyword that takes a few extra seconds to walk to page ten is fine. Query at a human-reasonable rate and you avoid most of the blocks that the extra volume would otherwise cause.
Spread the load
Do not collect an entire keyword set in one tight burst. Spread checks across a wider window — hourly batches, or a daily run that trickles through the list rather than sprinting. Hold one IP for a single keyword's ten-page walk using a sticky session, then rotate between keywords so no single address carries the whole run. The country stays German throughout; rotation changes the address, not the location.
Track google.de from a real German mobile IP
Deep tracking is heavier now, so the IP you send it from matters more than it used to. A real German mobile IP, trusted via CGNAT because thousands of real subscribers share it, is far less likely to be rate-limited than a datacenter address running ten pages per keyword. It also gives you the correct vantage point: tracking google.de SERPs from Germany shows you what German users actually see, region-true, which is the only valid measurement for a German market.
If you run tracking through a managed rank-tracker API, the same principles apply on the provider's side — the pagination cost and IP-quality pressure did not disappear, they moved.
Stay a polite, legitimate client
Rank tracking observes public, ranked results. It is measurement, not manipulation. Google's own robots.txt disallows automated access to /search and its terms do not invite scraping, so treat automated tracking as a deliberate decision rather than blanket permission: collect public ranking data only, never personal data, in line with the GDPR, and keep your request rate reasonable. Google Search Console still reports your own site's rankings first-party and remains the cleanest source for your own domain.
Takeaway
Removing &num=100 did not break rank tracking — it made deep tracking about ten times heavier and pushed IP quality and pacing to the front of the problem. The fix is not a trick; it is discipline. Pace requests, spread the load across time, hold and rotate sessions sensibly, and track google.de from a real German mobile vantage point so the results reflect what German users actually see.
FAQ
What did the &num=100 Google parameter do?
Adding &num=100 to a Google search URL asked Google to return up to 100 organic results on a single page instead of the default ten. Rank trackers relied on it to read the whole top 100 in one request. Around September 2025 Google stopped honouring the parameter, so it no longer works.
Why does removing &num=100 mean about 10x more requests?
With the parameter gone, one request returns roughly ten results instead of a hundred. To cover the same top 100 positions a tracker now paginates through about ten pages, so a deep daily rank check that used one request per keyword needs roughly ten. That is why deep tracking sees about ten times the request volume.
Did removing &num=100 cause Search Console impression drops?
Many sites saw desktop impressions fall in Search Console around the same time. A widely shared explanation is that automated tools reading deep result pages had been inflating impression counts, and once &num=100 stopped working those deep loads dropped away. It reflects how the data was collected, not a real loss of visibility.
Related guides
- Track Google AI Overviews in Germany (AI Overview Tracking)
- How to Track Google.de SERPs with a German Mobile Proxy
- Rank Tracker API + Mobile Proxy: Google.de SERP Pipeline
- SERP Monitoring for Client Reporting: German Mobile IPs for SEO Agencies
CTA
If deep rank tracking now costs you ten times the requests, the IP you send them from decides whether they get through. Get a German mobile proxy → — real German 5G IPs, sticky sessions, rotation on demand, fair-use 20 GB/day.
Get a German mobile proxy: Real 4G/5G IPs from Germany, fair-use 20 GB/day — see German 4G/5G mobile proxy options or get a dedicated proxy on the Blue network.