# How long does Google take to crawl a new page? On this site, a median of 3 days, and the waits came in batches
> On this site, Google's crawler first requested 13 of 15 new pages a median of 3 days after launch, in batches, and none on the day a page went live.
Canonical: https://brandonlazovic.dev/articles/new-pages-wait-days-for-googlebots-first-visit/
Author: Brandon Lazovic
Published: 2026-10-22
## The short version
- On this site, 13 of 15 recent articles got a first request identifying as Googlebot, Google's crawler, a median of three days after going live. The shortest wait was one day and the longest was nine.
- The waits came in batches. Six articles published across nine days were all first requested on September 30, so the wait tracked when Googlebot next came through, not when the page went up.
- Other crawlers reached 14 of the 15 articles on the day each went live, so a page can be read by some crawlers on day one and still wait days for Googlebot.
- A crawl is not an index entry. This measures requests only, on one site over 22 days, with a sitemap entry, an RSS (Really Simple Syndication) feed item and internal links all in place, so it cannot say which signal Googlebot followed or what other sites should expect.
A new page on this site waited a median of three days for its first visit from Googlebot, Google's crawler, the program that visits web pages to read them. Each of those visits is a crawl, one request for a page.
Of 15 recent articles, 13 drew a request that identified as Googlebot. The shortest wait was one day and the longest was nine, and none was requested on the day it went live. [1] Going live and being read by Google were two events, days apart.
> On this site, 13 of 15 recent articles got a first Googlebot request a median of 3 days after going live, and none was requested on its publish day.
## How long did Googlebot take to crawl new pages on this site?
On brandonlazovic.dev, Googlebot, Google's crawler, first requested 13 of 15 recent articles a median of three days after they went live. To crawl a page means to request it so its content can be read. The shortest wait was one day, the longest was nine, and no article was requested on its publish day. [1]
Two articles had no Googlebot request by October 11. One had been live for 16 days and the other for two. The log is sampled. Cloudflare, which records it, keeps only a share of requests, so a missing request does not prove a missing visit. Those two are reported as "none yet", which means not requested or not sampled. It does not mean never crawled.
## How did I measure the wait for Googlebot's first visit?
I joined this site's crawler log to the publish date of every article dated inside the log's range. The log is a daily record of requests from bots, and each request carries a user agent, the label software sends to say what it is. For each article I took the first day a request for its address carried a Googlebot label. [1]
The log runs from September 20 to October 11, 2026, and 15 articles fall inside it, dated September 21 to October 9. A match is a request for the article's address, with or without the trailing slash. Requests for the Markdown version of an article, and for its main image, do not count.
Googlebot here means a user agent containing the version tag Googlebot/2.1, in either its phone or its desktop form. Three other Google crawlers, Googlebot-Image, GoogleOther and Google-Safety, are counted separately and are not included. The labels are claims, because the free Cloudflare plan this site runs on cannot check which network sent a request. A program pretending to be Googlebot would be counted as Googlebot. I write "requests identifying as Googlebot" for that reason.
The wait is the first matching log date minus the publish date, in whole days. A scheduled job makes each article public at 05:00 UTC (Coordinated Universal Time) on its date, and the log's dates are also UTC, so a wait of zero would mean the same day. [2]
I read every Google row for two of the articles by hand. For one article, the first request fell on September 30. Its publish date was September 23, so the wait is seven days, which matches the table below.
## How long did each of the 15 articles wait for Googlebot?
Across 15 articles on this site, the wait for the first request from Googlebot, Google's crawler, ran from one day to nine, and two articles had none by October 11. Six were first requested on September 30 and three on October 5. The count covers requests only and says nothing about whether Google stored any of the pages. [1]
The table lists each article's publish date, the date of its first request identifying as Googlebot, and the wait in whole days. Each article is named by the last part of its web address. Where Googlebot has not yet asked, the last column shows how many days the article has been live in place of a wait.
| Article | Published | First Googlebot request | Wait in days |
|---|---|---|---|
| ten-barcodes-one-forwarded | 2026-09-21 | 2026-09-30 | 9 |
| llms-txt-drifts-by-construction | 2026-09-22 | 2026-09-30 | 8 |
| catalog-shared-by-default | 2026-09-23 | 2026-09-30 | 7 |
| gate-the-irreversible-step | 2026-09-24 | 2026-09-30 | 6 |
| video-transcripts-nobody-can-read | 2026-09-25 | none yet | 16 days live |
| lighthouse-ai-catalog-audit | 2026-09-28 | 2026-09-30 | 2 |
| googlebot-image-meta-tag-ignored | 2026-09-29 | 2026-09-30 | 1 |
| agentic-checkout-idempotency | 2026-09-30 | 2026-10-05 | 5 |
| mobile-first-indexing-two-dates | 2026-10-01 | 2026-10-05 | 4 |
| google-ai-contribution-pilot-payout | 2026-10-02 | 2026-10-05 | 3 |
| ai-coding-spend-is-rereading | 2026-10-05 | 2026-10-06 | 1 |
| videoobject-required-recommended-fields | 2026-10-06 | 2026-10-08 | 2 |
| agents-disable-each-others-guardrails | 2026-10-07 | 2026-10-09 | 2 |
| deploy-failed-site-looked-fine | 2026-10-08 | 2026-10-09 | 1 |
| name-the-counting-method | 2026-10-09 | none yet | 2 days live |
Each line joins one article's publish date on the top row to its first request identifying as Googlebot on the bottom row, from the table above. The amber lines are the two batches, six articles first requested on September 30 and three on October 5. A hollow circle is an article with no request yet. [1]
Every first request used the smartphone Googlebot label, the version of Googlebot that identifies as a phone.
## Why did the waits come in batches?
Googlebot's first requests clustered on a few days instead of following each publish time. Six articles published across nine days, from September 21 to September 29, were all first requested on September 30. Three published from September 30 to October 2 were all first requested on October 5. The wait tracked when Googlebot, Google's crawler, next came through the site. [1]
The middle column of the table shows it. My read: the wait on this site was set by the crawler's schedule and not by the publish time, because articles published seven days apart got their first visit on the same day.
From October 5 onward the waits fell to one or two days. Why the rhythm changed is unknown. Fifteen articles on one site is directional evidence.
## Did other crawlers get there first?
On 14 of the 15 recent articles on this site, the earliest request from any crawler in the log fell on the publish day itself, most often from a crawler identifying as Cloudflare AI Search, a crawler run by Cloudflare. Googlebot, Google's search crawler, came later, so a page can be read by some crawlers on day one and still wait days for Googlebot. [1]
The fifteenth article, ai-coding-spend-is-rereading, is the exception. Its log shows a request from curl, a command-line tool for fetching a web address, six days before the article went live. That was a tool asking for a page that did not exist yet, and not a crawler.
Other Google crawlers also came first on the two articles I checked by hand. On those two, a user agent identifying as Google-Safety requested the page on its publish day, and GoogleOther requested it days before Googlebot did. I checked only those two, so I make no claim about the other 13.
The same log shows how other crawlers treat scripts and stylesheets, which is the subject of the article on [whether AI crawlers request your JavaScript](/articles/ai-crawlers-javascript-requests/).
## What does a Googlebot crawl-wait measurement not tell you?
A measurement of Googlebot's first request cannot say which signal brought Googlebot, Google's crawler, to a page, whether the page was indexed, or what other sites should expect. Every article here launches with an entry in its sitemap, a file that lists the pages a site wants crawled. It also launches with a feed item and with internal links. All three arrive at the same moment, with no manual submission. [2]
The feed is an RSS (Really Simple Syndication) feed, a file that lists recent posts in a form programs can read. So this measures one discovery setup, and I cannot credit the sitemap, the feed or the links.
A crawl is not an index entry. An index is the database a search engine builds from the pages it has read, and I did not check Search Console, Google's free reporting tool for site owners, so I cannot say whether any of the 13 pages is in it. The figures describe 15 articles on one site over 22 days, and I make no claim about a typical wait elsewhere.
Two earlier articles cover neighboring questions. One covers what the [sitemap "Couldn't fetch" status](/articles/sitemap-couldnt-fetch-quality/) can mean in Search Console. The other covers [pages that no internal link points to](/articles/orphan-pages-still-crawled/). That piece counts requests to older articles over time. This one measures the first request to articles that launched inside the log window.
## How do you measure the wait on your own site?
To measure your own wait, subtract each page's publish date from the first date your log shows a request identifying as Googlebot, in whole days, and read the result across a batch of pages instead of one. The figures in this article belong to one site, so they show what to measure and not what to expect.
Look at the dates of first requests as well as the waits. If several pages published on different days share one first-request date, as six did here, the schedule of the crawler's visits is setting your wait. And treat a page with no request yet as unanswered, because a sampled log cannot show that Googlebot never came.
## Sources
1. First-hand: brandonlazovic.dev bot request counts from Cloudflare's analytics API, analytics/ai-crawler-ua-paths.csv (2026-09-20 to 2026-10-11, 22 dates, 7,888 rows), measured 2026-10-12
2. First-hand: brandonlazovic.dev article publish dates (pubDate in src/content/articles/) for the 15 articles dated 2026-09-21 to 2026-10-09, and the scheduled job that makes each one public at 05:00 UTC, read 2026-10-12