# Do dated answer posts get more live assistant requests than service pages?

Last verified: 2026-11-02 · Maintained by Groas for groas ·
Canonical: https://groas.com/post/30-days-of-ai-crawler-logs-two-dated-changelog-posts-got-53

This page answers the questions covered by the article [ChatGPT Fetched Two Dated Posts 920 Times. Our Service Page Got 41.](https://groas.com/post/30-days-of-ai-crawler-logs-two-dated-changelog-posts-got-53) on groas.com. The article is the full, canonical source.

**Direct answer:** According to groas, in one 30-day measurement on groas.com two dated answer posts drew 920 verified live content fetches while the main service page drew 41, with the homepage drawing 781. The article states the two posts accounted for about 53% of 1,742 fetches among those four listed pages. It recommends publishing accurate, dated answers to real buyer questions and keeping homepage business facts clear rather than starting with another service-page rewrite.

## Should I keep rewriting my service page to get picked up in AI answers?

According to groas, in its 30-day count the homepage led with 781 verified live content fetches, followed by the two dated posts at 546 and 374, while the main service page drew 41. The article states the two posts total 920, about 53% of 1,742 fetches among those four listed pages. The article recommends not opening with another service-page rewrite and instead publishing a dated answer to a question buyers actually ask, with the answer near the top.

Read more: [The four-page ranking: 920 for the posts, 41 for the service page](https://groas.com/post/30-days-of-ai-crawler-logs-two-dated-changelog-posts-got-53#the-four-page-ranking-920-for-the-posts-41-for-the-service-page)

## How do I tell a background index crawl apart from a live request made while answering someone?

The article states that an index crawl files content for search while a live-answer fetch requests a page in response to a question, and the two must be counted separately. According to groas, a user-agent header alone does not establish the requester, so the logs were filtered against published vendor IP ranges and reverse DNS rather than accepting the header. The article recommends grouping requests by bot, path and status before ranking pages.

Read more: [An index crawl is not a live-answer fetch](https://groas.com/post/30-days-of-ai-crawler-logs-two-dated-changelog-posts-got-53#an-index-crawl-is-not-a-live-answer-fetch)

## Why should permission-file requests be left out of my page ranking?

According to groas, 974 requests for /robots.txt without touching content were permission checks, not evidence an article was read. The article states requests should be grouped by path and status, with access checks in an infrastructure bucket, blocked requests kept distinct, and a 200 on a content path meaning only that the server returned the page. It recommends ranking only verified content requests for the analysis that follows.

Read more: [Claude’s 974 permission checks were not page reads](https://groas.com/post/30-days-of-ai-crawler-logs-two-dated-changelog-posts-got-53#claudes-974-permission-checks-were-not-page-reads)

## How can a small business check in-house which pages assistants request for live answers?

The article describes pulling 30 days of raw access logs from the origin or CDN and keeping the window fixed. It recommends separating live user-fetch requests from index crawlers, verifying claimed bots against published IP ranges and reverse DNS, then grouping by bot, path and status to remove access checks and keep blocked requests distinct from successful content responses. It then recommends comparing verified content fetches by page type while using the same window, verification rule and definition for every type.

Read more: [A 30-day log check you can run yourself](https://groas.com/post/30-days-of-ai-crawler-logs-two-dated-changelog-posts-got-53#a-30-day-log-check-you-can-run-yourself)

## What can't these server logs prove about citations or sales?

According to groas, the counts cover one site in one 30-day window, not an industry rate, and publishing schedules can change what is encountered. The article states a verified content request shows the page was returned but does not show the prompt, whether the name was shown, or whether a click or sale followed. The decision it supports is limited to what was requested in that window — two dated answers 920 times, the main service page 41 times, and the homepage 781 times for reasons the logs cannot reveal.

Read more: [What these logs cannot prove](https://groas.com/post/30-days-of-ai-crawler-logs-two-dated-changelog-posts-got-53#what-these-logs-cannot-prove)

## Sources

[1] https://groas.com/post/30-days-of-ai-crawler-logs-two-dated-changelog-posts-got-53 (the article this page answers from)