# AI Crawler Requests to Toolradar Tool Pages by Category

By [Toolradar Editorial Team](https://toolradar.com/editorial-policy) Updated 5 November 2026 11 min read

Productivity had the most identified AI crawler requests to Toolradar tool pages in the six-day window from 30 October to 4 November 2026: 5,252 requests across 1,547 tools. Finance had the fewest of the ten categories shown, with 1,920 requests across 584 tools. A request is one page fetch by a crawler the table identifies as an AI crawler, not a citation, a recommendation or a human visit.

Toolradar runs this directory and sells placements, including sponsored boxes and paid positions in buyer guides. Toolradar is part of Dupple, which sells newsletter sponsorships. Toolradar data: the figures come from Toolradar's own crawler table, cover six UTC days from 30 October to 4 November 2026, and have not been independently audited. Read [how we make money](https://toolradar.com/how-we-make-money) and the [editorial policy](https://toolradar.com/editorial-policy) before you use them for a buying or marketing decision.

## The ten category rows

Identified AI crawler requests, top categories

1. Productivity5,252 Requests to tool pages from identified AI crawlers, 30 October to 4 November 2026. Overlapping categories are not additive
2. AI Assistants3,947 Requests to tool pages from identified AI crawlers, 30 October to 4 November 2026. Overlapping categories are not additive
3. Developer Tools3,678 Requests to tool pages from identified AI crawlers, 30 October to 4 November 2026. Overlapping categories are not additive
4. Analytics3,255 Requests to tool pages from identified AI crawlers, 30 October to 4 November 2026. Overlapping categories are not additive
5. AI & Automation2,585 Requests to tool pages from identified AI crawlers, 30 October to 4 November 2026. Overlapping categories are not additive
6. Workflow Automation2,516 Requests to tool pages from identified AI crawlers, 30 October to 4 November 2026. Overlapping categories are not additive
7. AI Agents2,371 Requests to tool pages from identified AI crawlers, 30 October to 4 November 2026. Overlapping categories are not additive
8. Marketing2,334 Requests to tool pages from identified AI crawlers, 30 October to 4 November 2026. Overlapping categories are not additive
9. Security2,001 Requests to tool pages from identified AI crawlers, 30 October to 4 November 2026. Overlapping categories are not additive
10. Finance1,920 Requests to tool pages from identified AI crawlers, 30 October to 4 November 2026. Overlapping categories are not additive

Requests to tool pages from identified AI crawlers, 30 October to 4 November 2026. Overlapping categories are not additive

The [CSV](https://toolradar.com/data/ai-crawler-tool-page-categories-2026-11-05.csv) lists the ten categories with the largest total in this extract, ordered by requests, then category slug for a tie. The query includes currently published tool records with at least one logged request to their individual tool page during the window. Each displayed category contains at least eight such tools. These are category aggregates, not per-page advertising forecasts.

Table: Requests to tool pages by category, 30 October to 4 November 2026

| Category | Tools with recorded requests | Identified requests | Cloudflare-verified subset | Observed UTC days |
| --- | --- | --- | --- | --- |
| Productivity | 1,547 | 5,252 | 1,472 | 6 |
| AI Assistants | 1,296 | 3,947 | 964 | 6 |
| Developer Tools | 1,420 | 3,678 | 887 | 6 |
| Analytics | 1,110 | 3,255 | 719 | 6 |
| AI & Automation | 927 | 2,585 | 695 | 6 |
| Workflow Automation | 897 | 2,516 | 607 | 6 |
| AI Agents | 832 | 2,371 | 568 | 6 |
| Marketing | 705 | 2,334 | 528 | 6 |
| Security | 618 | 2,001 | 651 | 6 |
| Finance | 584 | 1,920 | 453 | 6 |

A tool can belong to more than one category. Its requests therefore contribute to each category to which it was assigned at extraction. Do not add these rows to estimate unique requests across Toolradar.

> Identified requests are page fetches by user agents that identify as AI crawlers. They are not citations, recommendations or visits, and the rows overlap, so never add them up or read a higher row as a better product.

## What the ranking shows

Productivity sits at the top with 5,252 identified requests, ahead of AI Assistants (3,947) and Developer Tools (3,678). Its 1,547 tools with recorded requests are also the most of any row. At the bottom, Marketing (2,334), Security (2,001) and Finance (1,920) close the table. Finance has 584 tools with recorded requests, the fewest of the ten rows.

The spread is modest. The top row has about 2.7 times the requests of the bottom row (5,252 divided by 1,920), and about 2.6 times as many tools with recorded requests (1,547 divided by 584). Eight of the ten rows fall between 2,001 and 3,947 requests. The ranking mostly follows how many tool pages in a category recorded requests. It does not show which category buyers prefer, which one an assistant recommends or which one earns the most citations. All ten rows cover the same six observed days, so the spread is not explained by one row having more days.

A second reading uses the verified column. The verified share is verified requests divided by identified requests, rounded to one decimal place. It is computed for this page from the CSV and is not a column in the file. Security has the highest share at 32.5 percent (651 of 2,001), although it ranks ninth by identified requests. Productivity follows at 28.0 percent (1,472 verified). Analytics has the lowest share at 22.1 percent (719 of 3,255). Every row stays below one third, so most requests in every category were not marked verified. A high share means more of a group's requests were verified. It does not mean that group's pages were read more.

## What a request counts

Toolradar's AiBotFetchDaily table groups Cloudflare AI crawler request counts by UTC date, tool, page type and identified bot. This report's query selects the tool page type, the individual /tools/<slug> page, and sums the request counts. It includes requests with 2xx and 304 responses. It excludes comparison pages, alternatives pages, category pages and other page types, including tool subpages. The query has no bot condition, so every bot the table identifies is included, and the extract does not show which bot made each request. The table identifies 17 bots by user agent: ChatGPT-User, OAI-SearchBot, GPTBot, Claude-User, Claude-SearchBot, ClaudeBot, Perplexity-User, PerplexityBot, Google-Other, Amazonbot, Meta-ExternalFetcher, Meta-ExternalAgent, MistralAI-User, DuckAssistBot, YouBot, CCBot and Diffbot. Toolradar leaves out Exa, Bytespider and Applebot on purpose, because they read the whole catalog every day, so their requests are in none of the figures here.

A request is a fetch of a page. It is not a unique bot, session or end user, and the same bot can request a page repeatedly. The count alone does not show that a model used the page for training, incorporated it into an answer, quoted it or recommended the product. It also does not show which question a person might have asked an assistant.

## How categories are assigned

The [read-only method](https://toolradar.com/data/ai-crawler-tool-page-categories-method-2026-11-05.txt) joins the crawler table to Toolradar's published tools and their current category memberships. A tool with a recorded request is counted once in the distinct-tool column of each category it belongs to, and its logged requests are summed within each such category. Categories with fewer than eight tools with requests are excluded before the top ten are selected.

Membership and published status are evaluated at extraction, not reconstructed for each historical day. If a tool changed category or was unpublished after the window, a later rerun can differ even with the same source dates. The six observed days show that a category has at least one record on each of those dates. They do not prove that every tool or every crawler was measured on every day.

Only the top ten qualifying categories are shown, so a category left out of the file may still have crawler activity. The totals favor groups with more tool pages or more frequent fetches, and they are not normalized by catalog size. Dividing requests by the displayed tool count gives a descriptive ratio among tools with recorded requests. It is not a probability that a tool will be cited.

## Verification and measurement limits

The verified column is the subset of identified requests that Cloudflare marked as verified. [Cloudflare's verified bot documentation](https://developers.cloudflare.com/bots/concepts/bot/verified-bots/), read 5 November 2026, describes validation by Web Bot Auth or IP validation. Verification coverage differs by bot, and user-agent identification is not proof of the operator. In this table, PerplexityBot requests are never counted as verified. A request outside the verified subset is not automatically fraudulent, and the verified subset is not a census of every legitimate AI request.

The window contains six complete UTC days, 30 October through 4 November 2026. That is too short to show a trend. A crawler may revisit a category for reasons this aggregate does not capture. Higher request counts do not establish higher demand from software buyers or better product quality.

## What a software vendor can do with the data

The table answers one question: how much AI crawler traffic reached tool pages in each category during this window. It cannot tell a vendor that its product was read, that an assistant used the page, or that an assistant would recommend it. A vendor can still use it in five steps.

1. Find your categories. Open your public tool page at /tools/<your-slug> and read the category names shown there. Outcome: you know which rows apply, because the table counts a tool once in each of its categories. If none of your categories is among the ten rows, this table says nothing about your group.
2. Compare with the neighbours. Look at the rows just above and below yours, and set identified requests against tools with recorded requests. Outcome: you see whether your category is large or small in this window. That describes the group, not your product.
3. Read the verified share for your row, as computed above. Outcome: a low share, such as Analytics at 22.1 percent, means most of that group's requests were not marked verified. Treat it as a caveat about measurement, not a judgment of your product.
4. Check the facts on the page itself. Make sure the name, tagline, description and pricing link on your tool page are accurate and current, because that page is what each request fetched. Outcome: any reader or assistant that reads the page sees the current facts, and a wrong fact is one you can fix.
5. Compare the same row in a later extract only if the window, the page type and the category rules are the same. Outcome: a change you can attribute to the data rather than to a change in method.

Nothing in this table predicts a citation or a recommendation, and a listing does not buy a ranking. The ranking score and Toolradar Data take no advertising or placement input, as the [vendor FAQ](https://toolradar.com/vendors) and the [method on how we rate](https://toolradar.com/how-we-rate#toolradar-data) explain. The [agent information page](https://toolradar.com/for-agents) describes machine-readable access, and the [statistics hub](https://toolradar.com/statistics) holds catalog data with its own methods.

## Requests are not advertising impressions

Toolradar sells [sponsored boxes](https://toolradar.com/advertise/formats/sponsored-box). A box appears on /best/<category> and /alternatives/<slug> pages where one is sold, not on /tools/<slug> pages, and this report counts only tool pages. A crawler request in this table is therefore not a view of a sponsored box. The box carries a visible Sponsored label, and its link uses rel="sponsored nofollow noopener". Requests for a box are made at [/company/sponsored](https://toolradar.com/company/sponsored).

The report does not associate requests with advertisers, and it does not record box impressions. A vendor comparing a placement with this table should treat these counts as neither campaign reach nor a forecast.

## Glossary

Table: Terms used in this report

| Term | Meaning in this report |
| --- | --- |
| Identified request | One fetch counted for one of the 17 bots the crawler table identifies by user agent (Exa, Bytespider and Applebot are not among them), summed per UTC day, tool, page type and bot. |
| Verified request | An identified request that Cloudflare marked as verified, shown in the verified column. |
| Tool page | The individual /tools/<slug> page. No other page type is counted. |
| Observed day | A UTC date on which the category has at least one row in the window. |
| Tools with recorded requests | Distinct published tools with at least one request in the window that belong to the category. |
| Qualifying category | A category with at least eight tools with recorded requests. Only these can reach the top ten. |
| Verified share | Verified requests divided by identified requests, rounded to one decimal. Computed for this page, not in the CSV. |

## How to reproduce the table from the CSV

Download the [CSV](https://toolradar.com/data/ai-crawler-tool-page-categories-2026-11-05.csv) and run the script below in the same folder. It needs Python 3 and the standard library only. It sorts the rows by requests, then category slug, and prints each row with thousands separators.

# Run in the folder that holds the CSV (Python 3, standard library only). import csv rows = list(csv.DictReader(open("ai-crawler-tool-page-categories-2026-11-05.csv", encoding="utf-8"))) rows.sort(key=lambda r: (-int(r["requests"]), r["category_slug"])) for r in rows: print(f'{r["category_name"]}: {int(r["requests"]):,} identified requests, {int(r["tools_with_reads"]):,} tools, {int(r["verified_requests"]):,} verified')

The printed order matches the table above. The script does not change any figure. It only reads the file, so you can check the table against the published CSV.

## What we could not verify

We could not verify what any request was used for, including training, indexing, an answer or a citation. We did not audit Cloudflare's classification, the bot identification or the logs behind the table. Cloudflare's page describes its validation methods, not the result for each bot, and we did not check any bot's result individually. The six-day window does not show a trend, and a later extract can differ because category membership is evaluated when it runs. We did not check which of the 17 bots made the requests in each row, because the extract has no bot column.

## Frequently asked questions

### Which category had the most identified requests?

Productivity had 5,252 identified requests across 1,547 tools in the window from 30 October to 4 November 2026. That is a category total in this extract. It does not measure AI citations, recommendations or human visits to Productivity tools.

### Are all identified requests verified?

No. The verified column counts only requests that Cloudflare marked as verified. The verified share ranges from 22.1 to 32.5 percent across the ten rows. A request outside that subset is not automatically fraudulent, because verification coverage differs by bot.

### Can the category rows be added together?

No. A tool can belong to more than one category, so the same tool-page requests can appear in several rows. A sum of the rows has no defined meaning across Toolradar.

### Does a crawler request mean an AI assistant recommended the product?

No. A logged page fetch does not show what an assistant indexed, cited, recommended or showed to a person, and it does not show which question prompted the fetch.

### Does this count every AI crawler?

No. The table identifies 17 bots by user agent and leaves out Exa, Bytespider and Applebot on purpose, because they read the whole catalog every day. The totals describe those 17 bots, not every AI crawler, and a bot outside the list adds nothing to any row.

### Why does Security rank ninth but have the highest verified share?

The ranking uses identified requests, and the verified share divides verified requests by identified requests. Security has 2,001 identified requests and 651 verified, a share of 32.5 percent. The two measures answer different questions, and neither shows how much buyers care about a category.

### Does this measure sponsored-box impressions?

No. The report counts requests to individual tool pages. Sponsored boxes appear on best-category and alternatives pages, which this report does not count, and the table does not record box impressions.

Spotted an outdated price or a wrong fact? [Suggest a correction](https://toolradar.com/corrections?url=https%3A%2F%2Ftoolradar.com%2Freports%2Fai-crawler-reads-by-category): we check it against the vendor's own page. Work for one of these vendors? [Claim your listing](https://toolradar.com/claim) to update it yourself.

## Data and method

- [Crawler requests by category (CSV)](https://toolradar.com/data/ai-crawler-tool-page-categories-2026-11-05.csv)
- [How the counts were made](https://toolradar.com/data/ai-crawler-tool-page-categories-method-2026-11-05.txt)

## Related

- [Toolradar Tool-Page Sessions by Software Category](https://toolradar.com/reports/buyer-traffic-by-category)Productivity led Toolradar's 5 November 2026 snapshot of tool-page sessions by category. How to read the…
- [How we rate](https://toolradar.com/how-we-rate)How Toolradar scores tools, and where its own data comes from.

Some offers on this page may be paid placements or contain affiliate links.

---

Source: https://toolradar.com/reports/ai-crawler-reads-by-category