We deleted the pages with the most search volume on our site

Every figure below was read in Semrush’s India database between 15 and 17 September 2026 — one head term per page, which is how each page had been targeted in the first place.

Two pages stayed: Ahmedabad, which is where the company is registered and the one place we can describe from the inside, and Gujarat around it.

A note on the number. 41,110 is the sum of one head term per retired page. It is not the total addressable demand for those cities, which is larger, and it is not traffic we had — we ranked for none of it. It is the demand those pages were built to chase.

The pages were not thin, not duplicated and not badly built. Several carried genuinely local material — what the Pune auto-component cluster searches for, why Kolkata’s difficulty sits at 10 while Ahmedabad’s sits at 60. By every ordinary SEO test they were worth keeping.

They were pointed at Indian buyers. The company sells to businesses outside India.

A page aimed at an enquiry you would not take is worse than no page at all, and it is worth being precise about why, because “it might bring in something” is a genuinely tempting argument.

• It spends crawl budget you do not have. This domain has one page indexed out of eighty-three. Every URL competing for Google’s attention is competing with the pages that describe what we actually sell.
• It blurs what the site is about. Fourteen pages saying “SEO company in [Indian city]” against a handful saying “SaaS SEO agency” tells a search engine, reasonably, that this is an Indian local agency.
• It wastes the reader’s time, which is the part that actually matters. Someone in Pune who finds a page written for Pune and then discovers the company does not take Indian work has been misled by us, not by the algorithm.

The test we applied was not “how much volume is this?” It was “if this page worked perfectly, would we want the enquiry?” Fourteen times the answer was no.

Retiring pages goes wrong when people delete them. We did not delete anything. The surviving destinations were the Gujarat page and the locations hub, and each retired URL was pointed at whichever of them actually covered it.

• 301 redirects, chosen by relevance rather than convenience. The four Gujarat cities point to the Gujarat page, which genuinely covers them. The nine metros and the India page point to the locations hub, which is the closest surviving equivalent. Nothing was pointed at the homepage, which is the usual shortcut and the usual way to lose whatever the URL had earned.
• Every internal link removed in the same change. 1,171 of them across 86 files — the footer column, the navigation, the in-body related links and the hub cards. A site that redirects a page but keeps linking to it is telling Google the page still matters.
• The files stayed on the server. The rewrite rules intercept the request before the file is served, so deleting was unnecessary risk for no benefit.
• The sitemap was rebuilt the same hour — 85 URLs down to 71 — and resubmitted.

Google re-read the sitemap the same day and moved its discovered-pages count from 84 to 75, matching the new file exactly. That was the fastest response this property has ever shown to anything, and it is the only part of the outcome we can currently measure.

The honest state of the experiment: we cannot yet say whether it worked.

Search Console’s Page indexing report has not refreshed since 21 September — checked at five separate slots across two days — so the number that would show whether removing fourteen competing URLs helped the rest get indexed is showing data from before any of this happened. Its sitemap report refreshes daily; the indexing one has not. That is Google’s cycle, not ours.

So this article is the decision and the method, not the result. When the report moves we will publish what it says, including if it says nothing changed.

This is a specific decision for a specific situation, and copying it blindly would be worse than not reading about it.

• If you serve that market, keep the pages. The whole argument rests on our not taking Indian work. Nothing here says local pages are bad.
• If the pages convert, keep them — even at low volume. We had no conversions to weigh, which made the decision easy in a way yours may not be.
• If you are established, the cost is different. Pages that already rank and already earn links are not the same asset as pages that have never ranked. We gave up potential; an established site would be giving up performance.
• If the only reason is tidiness, do not. Retiring pages has a real cost in lost equity and broken expectations. It needs a reason as specific as ours.

Search volume tells you how many people are looking. It tells you nothing about whether you want to meet them.

We spent a fortnight building pages for the largest demand we could measure, and then removed them because we had measured the wrong thing about it. The measurement was accurate; the question behind it was wrong.

Before the next keyword sheet, the useful question is not “what has volume?” but “whose enquiry do I want?” The second question is harder, it has no tool that answers it, and it is the one that decides whether the first question is worth asking.