Salience

Agent analytics

AI agent analytics for your website

See which AI agents and crawlers request your pages, how often they return and how your website responds. Salience turns your own server and CDN data into analytics for search, content and development teams, so you can follow activity from ChatGPT, Claude and other AI systems alongside Google Analytics and Google Search Console.

Free plan available. No credit card required. No browser tracking script.

Explore the free demo

Example Log Explorer rows for demo-site.example: ChatGPT-User fetching a guide after a user's question, OAI-SearchBot and PerplexityBot requesting product pages, GPTBot requesting a blog post, all verified against their providers' published ranges; ClaudeBot receiving a 404 on a retired product URL; and a request claiming to be GPTBot from an address outside OpenAI's ranges, marked unverified. A warning alert reports ClaudeBot receiving 404s on product URLs.

Log Explorer

demo-site.example · requests in real time

Example
  • ChatGPT-UserVerifiedUser-triggered/guides/returns-policy200
  • OAI-SearchBotVerifiedAI search/products/trail-running-shoes200
  • GPTBotVerifiedTraining/blog/how-to-choose-a-tent200
  • ClaudeBotVerifiedTraining/products/camp-stove-mk2404
  • PerplexityBotVerifiedAI search/products/camping-stove200
  • GPTBotUnverified/blog/packing-list200
WarningClaudeBot receiving 404s on product URLs

38 requests to 6 /products/ URLs returned 404 in the last hour, all from addresses inside Anthropic's published ClaudeBot ranges.

Recommended: check the 6 URLs and the redirects from the retired product pages.

AI systems recognised and classified, among 200+ crawler identities
  • GPTBot
  • ClaudeBot
  • PerplexityBot
  • Amazonbot
  • Applebot
  • Meta AI

What is agent analytics?

For a website, agent analytics means measuring requests from AI agents and crawlers. Salience makes this activity easier to investigate using your own server and CDN records, and keeps machine requests separate from the people AI assistants refer to your website.

  • Which AI systems reach your site
  • Which URLs they request
  • When they return
  • How your infrastructure responds

See what AI systems request and where access goes wrong

Get from a traffic total to a specific agent, page and response.

Which AI systems are reaching your website?

Identify recognised agents and crawlers, then inspect verification evidence where it is available. See the system behind the request, with more context than a name in a user-agent string.

Which pages are they requesting?

Explore requests to your products, articles, documentation and other important pages. See where activity is concentrated and how often identified systems return.

What does your website send back?

Inspect successful responses, redirects, missing pages and server errors. Find the affected URLs and give your team a clear starting point for investigation.

What changes over time?

Compare recorded activity before and after a release, content update or infrastructure change. Check whether the requests and response patterns you expected are appearing.

Explore the agents, pages and response codes in the demo

Understand the type of AI activity

AI systems request websites for different reasons. Salience groups recognised crawlers using the roles their providers describe. A recorded request does not prove that a page was trained on, indexed, cited or included in an answer.

User-triggered retrieval

Some systems fetch pages following a person's question or action. Examples include ChatGPT-User and Claude-User.

AI search crawling

Search crawlers collect pages to support discovery and search results. Examples include OAI-SearchBot and PerplexityBot.

Model development and training crawlers

Other crawlers collect content that may contribute to model development or training. Examples include GPTBot and ClaudeBot.

Looking for Google Analytics for AI agents?

Google Analytics helps you understand measured visitors and what they do on your website. Google Search Console reports on Google's search performance, including generative AI impressions, and how Google crawls your site. Salience adds the website request data: which identified AI systems reach your pages, what they request and how your infrastructure responds.

What do visitors do after arriving?

Google Analytics
Sessions, engagement and key events
Google Search Console
Google Search performance before the visit
Salience
Request activity; a GA4 connection adds visitor metrics

Are AI assistants sending visitors?

Google Analytics
AI Assistant acquisition channel
Google Search Console
Google Search traffic, including Google's AI search features
Salience
AI referral reporting from available source signals

Which AI agents request specific URLs?

Google Analytics
Not a complete website request log
Google Search Console
Google's crawling, not a cross-provider agent log
Salience
Identified systems, requested URLs and recorded activity

What responses do crawlers encounter?

Google Analytics
Not a native crawler response log
Google Search Console
Google's crawl response and availability reporting
Salience
Response codes from connected server and CDN records

Use Salience alongside your existing analytics to connect website access with the rest of your search measurement.

From an error response to a useful next step

Scenariosearch and development team

An AI crawler is getting 404s on our product URLs. Which pages, and what do we fix?

An error response traced to the pages behind it

  1. 1Spot the issueAn identified AI crawler is receiving 404s on product URLs.
  2. 2Inspect the affected pagesReview the recorded paths and responses, then ask your team to check the pages and redirects.
  3. 3Check what happens after the fixWhen the crawler returns, review the subsequent requests to see how your website responds.

The recorded paths and responses give the team the URLs to check, and the crawler's later requests show how the site responded after the fix. Salience does not control when a crawler returns.

Scenariocontent team

Visits from ChatGPT and Perplexity are rising, but we can't tell which of our pages the AI systems are actually fetching.

AI referrals set beside the pages AI systems fetch

  1. 1MondayThe AI referral report shows 126 visits from chatgpt.com, perplexity.ai and claude.ai in the last week, landing on 34 pages.
  2. 2Same viewThe request record shows OAI-SearchBot, PerplexityBot and ChatGPT-User fetched 1,045 pages in the same week, the 34 landing pages among them.
  3. 3Tuesday11 guides were fetched more than 50 times each with no referred visit. The team reviews those guides first.

The fetched pages and the referred visits sit side by side, so the team works from recorded requests rather than guessing which pages AI systems read.

Use the infrastructure your website already runs on

Connect a supported server, CDN or hosting source. Salience organises the available request data without adding a browser analytics tag to your pages. Platform requirements, coverage and update frequency vary by integration.

Choose your source

Find the integration that matches your website and review its requirements.

Connect your request data

Follow the setup guide for your platform, or share it with the developer who manages your infrastructure.

Explore agent activity

Once records arrive, review identified AI systems, requested pages and website responses.

View supported integrations Read the setup guides
Connect a supported source

Start with your own website, for free

Connect one website on Salience's free plan, or explore the demo before connecting your data.

  • One website
  • Up to 500,000 requests a month
  • 30 days of history
  • No credit card required
Start free View pricing

Agent analytics questions

Can Google Analytics track AI agents?

Google Analytics can measure visits referred by AI assistants and events collected by a website or app. It also filters known bot traffic. Its standard reports are not a complete record of AI crawler and agent requests. Salience adds that request-level view from connected server and CDN data.

Is Salience like Google Search Console for AI agents?

Salience gives website teams a way to investigate requests from supported AI crawlers and agents, including the pages requested and responses recorded. Search Console remains the source for Google's search performance and generative AI impression reports. The two tools provide different evidence about your website.

What is the difference between agent requests and AI referral traffic?

An agent request is a machine requesting a URL. An AI referral visit is a person arriving through a link from an AI service, where the source can be identified. They measure different activity and should not be added together as if they were the same kind of visit.

Does a crawler request mean my page appeared in an AI answer?

No. A request records access to a URL. It does not by itself prove answer inclusion, a citation, an impression or model training. Salience helps you investigate the website access behind that activity. Downstream visibility needs separate evidence.

How is this different from AI prompt tracking?

Prompt tracking examines the answers produced for selected prompts. Agent analytics examines recorded requests to your website. Salience provides first-party access data that you can use alongside research into mentions and citations.

Can Salience verify every AI agent?

Verification depends on the system and the evidence its provider makes available. Salience checks supported identities and shows their verification status. An unknown or unverifiable identity is not automatically malicious, and a claimed user-agent name is not proof of identity.

Will Salience show every request to my website?

Coverage depends on the connected source and its configuration. For example, a request blocked before a collection Worker runs may never reach that source, and some platforms can supply sampled logs. Review your integration's coverage when interpreting totals or missing activity.

Is this for monitoring AI agents I build?

This page covers agents and crawlers accessing your website. Monitoring the internal execution, prompts and costs of an agent application is a different use case.

Can I explore Salience before connecting my site?

Yes. Open the free demo to explore the product, or create a free account when you are ready to connect a supported source.

Related

Trust & data protection

Your data stays yours.

You stay in control

Salience processes your data only to provide the service. You remain the controller.

Stored in the UK

Core data is hosted in AWS London and encrypted at rest.

No browser tracking

Collection happens server‑side. There’s no client‑side pixel or tracking script.

Never sold or pooled

Your logs are not sold, used for advertising, or shared between customers.

Privacy and data protection

See how AI systems access your website

Free plan available. No credit card required. No browser tracking script.

Explore the free demo