Brantial Platform / Agent Analytics
Beta · Contact us about access and scope
Understand how SI agents access your site.
Inspect the bot and agent identifiers observed in your records, the pages requested and the responses returned. Evaluate access findings alongside your SI visibility and content work.
Agent Analytics analyses the server and CDN records you connect to make agent access visible. Our agentic commerce direction aims to extend this observation to the outcomes of product selection and test-cart tasks.
- Request and page analysis
- Access responses
- Measurement that complements SI visibility
| Claimed identity | Page | HTTP response | Identity status | Review |
|---|---|---|---|---|
| OAI-SearchBot | /products/trail-shoe/ | 200 | Verified in sample | Inspect the response record |
| GPTBot | /guides/size-chart/ | 200 | Verified in sample | Inspect the response record |
| PerplexityBot | /products/trail-shoe/ | 403 | Unverified | Review identity and the access reason |
| ClaudeBot | /guides/delivery/ | 404 | Unverified | Review the address and redirect |
| Unrecognised client | /categories/shoes/ | 301 | Unknown | Do not assume an SI agent |
These rows are a fictional illustration, not real traffic, provider performance data or evidence of a finished bot-verification feature. Each record is an access observation; a citation in an answer or a shopping outcome cannot be inferred from it alone.
Current focus · Log analysis
Which pages are requested? Where do problems appear?
A request for a product page, a citation in an answer and the use of that product in a shopping task are different events. Start by understanding the requests reaching your site and the responses it returns.
Inspect bot and agent identifiers.
Evaluate requests using recognised identity signals. Distinguish the bot name a request claims from whether that identity can be verified.
See which pages are requested.
Explore requests for product pages, categories, guides and other content. Identify where repeated access appears.
Review access responses.
Use status codes to investigate redirects, missing pages and access issues. Identify pages that need technical review.
Follow changes over time.
Review request frequency within a selected period. Compare records before and after content or access changes, accounting for the available data scope.
Read the findings alongside visibility data.
Evaluate crawl records, citations in monitored answers and measurable SI referrals together. Observing them in the same period does not, by itself, establish causation.
How it works
From connection to findings in four steps
- 01 Identify your data source.
Review the log scope of supported connections such as Cloudflare and Vercel. Discuss transfer requirements for other server or CDN setups.
- 02 Define the data scope.
Review which site, period and fields will be transferred. Analysis is limited to the records delivered to Brantial.
- 03 Inspect the requests.
Evaluate bot and agent identifiers, requested pages, status codes and repeated access.
- 04 Define the next action.
Route access issues to your technical team and content findings to the relevant work plan. Review subsequent records after a change is applied.
Measurement scope
Each measurement answers a different question.
| Measurement | The question it answers |
|---|---|
| Bot and agent requests | Which requests appeared in connected records, and which responses were returned? |
| SI visibility and citations | How did a brand or page appear in monitored SI answers? |
| SI referrals | Which visits can be identified as SI referrals in analytics data? |
| Shopping task outcomes Coming soon | Did a defined test select the right product and variant, and did the test cart match the expected result? |
GA4 and Search Console continue to serve their own measurement purposes. Agent Analytics adds a request-level area of investigation alongside them.
Scope note: A bot request alone does not prove that content was used to train a model, cited in a particular answer or responsible for a sale.
Coming soon · Planned capabilities
From agent access to shopping task outcomes
Our next goal is to add a second question to "Could the agent reach the product page?": "Did the shopping step meet the specified requirements?"
Product and commercial accuracy
We aim to compare product attributes, variants, price, stock and delivery information with task requirements and an authoritative data source.
Task-step observation
We plan to evaluate product discovery, variant selection and add-to-test-cart steps in supported, authorised test environments.
Understanding where a problem occurs
We aim to distinguish missing product information, outdated commercial data, access issues and incorrect action outcomes.
Checking again after an improvement
We plan a workflow that presents source-backed changes for authorised approval, applies them within supported scope, reads the destination system again and retests the task.
An agent is looking for waterproof shoes in size 42 within a defined budget.
Opening the page is an access observation. Adding the right variant at the current price to a test cart is a separate task outcome that requires verification.
If waterproofing is not documented in the source, it should be flagged as missing information rather than assumed.
| Stage | Evidence to see |
|---|---|
| Define the requirements | Budget, size, attribute and authorisation scope |
| Evaluate the product and variant | The product/variant record in the authoritative source and its check time |
| Check commercial conditions | Price, stock and, where relevant, delivery conditions for that scenario |
| Inspect the test cart | The test environment response and the product, variant, quantity and amount read back from the destination system |
| Evaluate the outcome | Expected versus observed; an explanation of missing information or errors |
This section describes our development direction. Supported systems, access and plan conditions will be announced separately. The initial scope is product selection and test carts, rather than payments or real order execution.
Explore our agentic commerce approachUse it alongside
Use it alongside your existing workflows.
FAQ
Frequently asked questions
What data does Agent Analytics analyse?
It analyses request records delivered to Brantial through supported connections. Available fields, historical coverage and freshness depend on the source and connection scope.
Can I see historical agent traffic by entering my domain?
Historical request analysis requires access to the relevant records. A public website check cannot replace server or CDN logs.
Can bot identity be established with certainty?
A claimed user-agent name alone is not conclusive identity evidence. Provider-published verification information and available record fields need to be evaluated together. Unverified requests should not be treated as confirmed identities.
Is Google-Extended a separate visiting bot?
Google-Extended has no separate HTTP user-agent identity. It is a control defined by Google for managing usage preferences through robots.txt and should not be counted as a separate visitor.
Does this replace GA4?
It provides a different measurement layer alongside GA4 and Search Console. Google Analytics automatically excludes known bot traffic; request records and analytics sessions should not be expected to produce identical totals.
Does a crawl record prove the source of a citation or sale?
Not on its own. Crawls, citations, referrals and commercial outcomes require different evidence. The same page and similar timing are insufficient to establish a definitive conversion chain.
Which plan includes it?
Agent Analytics is in Beta. Discuss account access, supported sources and usage conditions with our team.
Can logs contain personal or sensitive information?
Request records can contain IP addresses, URLs and header fields. Setup should include a review of the fields transferred, removal of unnecessary information and retention conditions.
Are agentic commerce tests available today?
The task and test-cart capabilities labelled "Coming soon · planned" describe our development direction. They do not share the availability status of current log analysis.
Understand agent access to your site.
Tell us about your infrastructure and the question you want to answer. We can review the data and access scope for Agent Analytics together.