Let's be honest, in the digital world, every top tech company is basically giving us free public transport while fiercely guarding the fuel that powers it: data. For years, AI developers — including the teams training models like ChatGPT — were running a quiet little jugaad right under Google’s nose. SEO tools also benefited, but more as passengers than drivers. The whole operation revolved around a simple, unofficial Google URL parameter: &num=100.
When you search on Google, you rarely move past the first page. Because, realistically, there’s no need.
But models like ChatGPT don’t just need answers; they need volume. They need to consume massive spans of text, examples, and context to learn how to respond in natural language. Ten search results at a time doesn’t cut it.
The undocumented trick was simple:
By adding &num=100 to the end of a Google search URL, the search engine would load 100 results on one page, instead of the usual 10. This allowed AI crawlers to collect large, structured, chronological data much faster and at lower cost. It was like using Google as a library where you could borrow 100 books at once instead of one shelf at a time.
SEO rank trackers used this as well, but their dependency was mostly practical. They just needed to check keyword rankings efficiently. The real heavy lifting was happening in AI training pipelines, where this shortcut directly helped speed, scale, and cost.
Around mid-September, without any press release or warning, Google quietly disabled support for &num=100. Everything reverted to 10 results per page. No glitch. No accident. Just a silent change.
The reasoning is straightforward: Google cannot allow the data it organizes to be used to strengthen a rival AI ecosystem. Especially now that ChatGPT is not just a novelty but a mainstream product operating in direct competition with Google’s Gemini.
This was not about user experience. This was Google protecting its competitive moat.
On the SEO side, chaos hit first. Rank tracking dashboards suddenly showed fewer visible keywords. Agencies assumed rankings had tanked. Clients panicked. But it wasn’t a ranking drop — it was simply harder to measure the rankings because the bulk view disappeared.
For ChatGPT-type training systems, the impact was more structural. The pipeline for efficiently collecting large search-indexed data became slower and more expensive. One of the simplest, highest-yield data taps was effectively sealed.
The internet may feel open, but the infrastructure underneath it is not neutral. Data is currency. Data is leverage. Data is power. Google simply reminded everyone that while the web might look free, the indexing, sorting, and accessibility layer belongs to them — and they will not casually subsidize a competitor’s growth.
AI companies are now direct opponents, not curious researchers. SEO, in this case, just got caught in the crossfire. There are no shortcuts in a data empire. And Google still holds the keys.
