Find the Pages That Drive a Site's Search Traffic

Knowing a site’s total search traffic is useful. Knowing exactly which pages earn it is actionable. This article walks through a page-level audit: take a site with real search volume, strip away the noise, and read which pages carry the traffic — and what that tells you about the demand underneath.

The worked case is brainly.com, an online-education site where students look up homework answers — think of it as a homework-help community. The original notes about 14.5 million monthly visits, 87.88% of it from the United States, and 86.32% of all traffic arriving from search engines. (Figures in this article are as mentioned in the original source and have not been independently verified.)

1. Start from one country, and strip the noise

Open the domain in Semrush and look at the keywords for its biggest market first — here, the United States. Before analyzing anything, remove the brand terms and the typo terms: every misspelling of the brand that users type by accident. The original lists 18 of them for this site (brainly plus 17 variants like baianly, brianly, braily, and brain ly).

After exclusion, what remains is about 1.7 million keywords carrying roughly 960K search visits. That is the real number: demand the site earns without its own name doing the work. Brand traffic is loyalty; everything else is capturable.

2. Sort by traffic, then map keywords to pages

Sort the cleaned list by traffic, high to low, and look at which pages those keywords point to. The pattern here is striking: most of the top keywords are full question sentences, and you can see it in the URLs — brainly.com generated one standalone page per question for search engines to index.

This is the site’s whole strategy, readable from the data: one page per question, thousands upon thousands of them, each capturing the exact sentence a student types. The audit reveals not just the keywords, but the page architecture that harvests them.

3. Read individual pages like a P&L

Zoom into single pages and read them like a profit-and-loss statement. One question page — “How are viruses different from bacteria? APEX,” a textbook homework question many students search verbatim — carries 22.2K total search volume, of which brainly.com captures only 2.9K, about 13%. The reason: a featured snippet answers the question right on the results page, diverting most clicks. Even the site holding the featured snippet only takes 5.5K, about 25% — the snippet splits the traffic for everyone.

Another page — “which sentence should be revised to reduce wordiness” — has 4.4K total volume and brainly.com takes 1.1K, about 25%, helped by holding both the Q&A block’s top spot and the first organic result. And across the remaining long tail, keyword difficulty is mostly low.

The lesson: featured snippets cap what any single ranking page can take, so you must discount keyword volume by the SERP layout, not just the ranking. A keyword is only as good as the clicks its results page actually lets through.

4. The compounding insight: many small pages, one big total

No single question page carries much traffic on its own. But with enough indexed questions and steadily growing domain authority, the total becomes enormous. That is the compounding insight: the unit of this business is not one great page, it is ten thousand adequate pages.

The original draws the forward-looking conclusion: GPT-4 is very good at answering questions, and the early movers have already acted. Google’s stated stance is that it judges content by quality, not by how it was generated — the standard being whether the visitor’s question gets a satisfying answer and whether they keep clicking to other pages. If that is the bar, then “making GPT’s answers more accurate” is the core competence of this kind of site.

5. Turn the audit into your own checklist

Run the same five moves on any search-driven site you want to learn from:

1. Pick its biggest traffic country and strip brand plus typo terms.
2. Sort the remaining keywords by traffic, not by volume.
3. Map the top keywords to their pages and name the page pattern (one page per question, one tool per job, one template per query).
4. Discount each keyword by the SERP layout — featured snippets, Q&A blocks, video carousels.
5. Ask whether the page pattern is reproducible at your scale, and what the moat would be (data, accuracy, community, speed).

Pages are where demand meets product. Total traffic tells you a site won; its pages tell you how — and that is the part you can copy honestly, in your own build, in your own words.

Series: Build Overseas Websites with GeFei

This article is based on GeFei’s shares, reorganized in our own words. Source: https://github.com/qiayue/opengefei