Getting Google's top 100 without num=100

What changed

In September 2025, Google stopped honoring the num=100 parameter. A SERP now shows about ten organic results per page, whatever parameter is passed. For most rank trackers, which fetched the top 100 in a single request, collecting one keyword became up to ten times heavier.

The visible effect in Search Console: many sites saw their impressions drop and their average position improve at that time, with no real change in traffic. The most common explanation: part of those impressions came from tools loading 100 results at once.

Why Google removed it

Google didn't detail its reasons, only saying that the parameter had never been officially supported. For a searcher, nothing changes: they already saw results ten at a time. The change mostly hits automated collection, since a top 100 now takes ten pages instead of one.

The only reliable method: pagination

To get the first 100 results, you do what a searcher clicking “Next” does: load page 1, then page 2, and so on up to page 10. Each page is a real results page, with its own blocks: ads, related questions, videos.

This method has an upside: you know which page each result really appears on. A site in 11th position may sit at the top of page 2, or stay on page 1 if page 1 holds more results than usual.

How to do it with the API

With the Semscraper Google Search API, the depth parameter sets how many pages to walk through, from 1 (about 10 results) to 10 (about 100 results). Every result returned carries:

• rank_type: its rank among results of the same type, e.g. 11th organic result;
• rank_serp: its rank among every element on the page, in display order;
• page: the Google page it appears on;
• pixel: its distance from the top of the page.

If Google returns fewer pages than requested, because the query doesn't have enough results, only pages found are billed: 10 pages requested, 4 found, 4 billed.

Pagination pitfalls

• A page doesn't always hold ten organic results: depending on the blocks shown, it can hold fewer, or a few more. A result's rank isn't enough to guess its page; that is what the page field is for.
• Each page has its own blocks: ads, videos or related questions can come back on page 2 or 3. To measure visibility, look at every page, not just the organic results of the first one.
• Not every query has ten pages: on a niche query, Google sometimes stops at page 3 or 4. Only pages found are billed.
• Organic rank carries on from page to page: the first organic result on page 2 has a rank_type of about 11, not 1.
• Collect all pages at once: pages captured at different times can overlap or leave gaps. With the depth parameter, a keyword's pages are collected in the same request.

What a top 100 costs

The first page costs €0.30 per 1,000, each extra page €0.20 (details on the pricing page). Processing time follows depth: a median of about thirty seconds for one page and about a minute for ten. The robots serve accounts in turn: a large batch flows through your own queue, without being held up by another customer's.

Do you really need the top 100?

Not always. For daily tracking, many teams settle for the top 20 or top 30 and keep the top 100 for a weekly snapshot or a market study. Beyond page 3, a result gets almost no clicks; knowing it is 47th or 63rd mostly helps spot progress.

The right setting depends on the use case: depth 1 to 3 for routine tracking, 10 for audits. It can be set keyword by keyword.

A budget example: 500 keywords tracked daily in the top 30, so 15,000 collections a month, cost €10.50. A weekly top 100 snapshot on the same keywords, so 2,000 collections a month, adds €4.20.