Documentation Customer Journey Analytics Customer Journey Analytics Guide

Dataset reference

Last update: October 13, 2026

Topics:
Integrations

CREATED FOR:

User

Adobe Brand Visibility is a generative AI-first application for Generative Engine Optimization, designed to help brands enhance their visibility, accuracy, and influence in AI-driven search environments. Brand Visibility provides insights into brand presence in AI-generated answers, offers prescriptive content recommendations, and automates optimization fixes.

AI has become a primary discovery channel. Large language model (LLM) agents, such as ChatGPT, Claude, Copilot, and Perplexity, crawl brand content.

PREREQUISITES

You must have a Brand Visibility paid offering. You then configure and enable the Adobe Brand Visibility source in your Experience Platform sandbox.

IMPORTANT

As part of this integration, some temporary processing of Brand Visibility data occurs in the United States. Data is ultimately stored in your designated region as configured in your Customer Journey Analytics contract.

About the dataset

See Overview for more details on the use cases for inbound and outbound integration of Brand Visibility with Customer Journey Analytics.

Brand Visibility reads CDN access logs on the server side and extracts records where the requesting party is a bot or automated agent. Because the data comes from the CDN layer, Brand Visibility captures requests from bots that do not fire any JavaScript tag. Standard web analytics tools miss this traffic entirely.

The dataset uses the CDN Requests Summary field group. Every field is located under a cdn object, so the field names in the tables below take the form cdn.name, for example cdn.url and cdn.botType.

Each record describes one combination of host, URL path, bot type, CDN provider, status code, referrer, forwarded host, and time to first byte for one hour. When the same combination appears more than once hourly, Customer Journey Analytics combines those records into one row and increases the request count. Use the CDN Request Count metric to measure volume. Do not use row count.

Dimensions

The following dimensions are available to use as components in a data view once you have set up a connection that includes a Brand Visibility dataset. The Field column shows the source field in the CDN Requests Summary field group.

Dimension

Field

Description

CDN URL

cdn.url

The normalized full URL for the request, intended as the join key. Brand Visibility combines the host and the requested path into a single URL and normalizes it to match the full-URL form that Customer Journey Analytics stores for web data. Use this dimension to join the Brand Visibility lookup dataset to an event dataset that has an equivalent full-URL field. It includes the host and path, but not the scheme.

CDN URL Path

cdn.path

The raw URL path and query string that the agent requested, as delivered by the CDN. Does not include the scheme or host. Use this when you need the exact requested path rather than the normalized join key.

CDN Host

cdn.host

The hostname that received the request, for example, www.example.com. This host is also part of the CDN URL join key. A dataset can contain multiple hosts when an organization has multiple subdomains on the same CDN account.

CDN Bot Type

cdn.botType

Brand Visibility’s classification of the requesting agent. Values cover classic search crawlers, AI index crawlers, and AI live-fetch agents. See the Bot agent categories below for the full taxonomy.

CDN User Agent

cdn.userAgent

The raw user-agent string from the CDN log. Useful for distinguishing sub-types within a bot classification, or for validating the classification assigned by Brand Visibility.

CDN HTTP Status

cdn.status

The HTTP response status code. Indicates whether the bot received the content it requested. See the Status codes below for interpretation guidance specific to AI traffic.

CDN Provider

cdn.cdnProvider

Which CDN handled the request. Values are akamai, byocdn-akamai, byocdn-fastly, and byocdn-cloudfront. The byocdn- prefix indicates the log collection pathway, not a different CDN vendor. A dataset can contain multiple values when an organization has hosts behind different CDN configurations.

CDN Referrer

cdn.referer

The HTTP Referer header value from the CDN log. Often empty for bot traffic. When present, it can indicate which AI product or domain triggered the fetch. For example, chat.openai.com.

CDN Forwarded Host

cdn.xForwardedHost

The X-Forwarded-Host header value, if present. Relevant when the request passed through a reverse proxy or CDN shield layer before reaching the origin.

CDN Event Date

Derived from the record timestamp

The date part of the hourly batch timestamp for this record.

CDN Event Hour

Derived from the record timestamp

The hour part of the hourly batch timestamp for this record.

Bot agent categories

The CDN Bot Type dimension organizes agents into three categories. Each category answers a different analytical question.

Classic search crawlers index content for traditional search engines. Use this category to measure how visible your content is to traditional search engines.

Bot type value

Vendor

Description

GoogleBot

Google

Google’s main search index crawler. Also serves Google Discover and Google News.

BingBot

Microsoft

Bing’s search index crawler. Also feeds Microsoft Copilot’s web grounding index.

AI index crawlers crawl content to build or update an AI product’s training corpus or search index. These crawlers are preparing a model’s knowledge base, not responding to a live user request. When a URL has high crawler volume, AI vendors consider that content worth indexing. When a URL has low crawler volume but high live-fetch volume, the model draws from cached knowledge rather than fetching fresh content.

Bot type value

Vendor

Description

GPTBot

OpenAI

OpenAI’s primary crawler for model training data and knowledge base construction.

OAI-SearchBot

OpenAI

OpenAI’s crawler for ChatGPT’s web search product. Distinct from GPTBot. This agent builds the real-time search index, not the training corpus.

ClaudeBot

Anthropic

Anthropic’s primary crawler for model training data.

Claude-SearchBot

Anthropic

Anthropic’s crawler for Claude’s search and retrieval index. Distinct from ClaudeBot.

PerplexityBot

Perplexity

Perplexity’s index crawler. Perplexity uses this agent to build the corpus for its answer generation.

AI live fetches occur when a real user submits a prompt to an AI assistant and the assistant fetches the page live before responding. Use this category to measure direct user demand arriving through AI assistants.

Bot type value

Vendor

Description

ChatGPT-User

OpenAI

A user asked ChatGPT a question. ChatGPT fetched this URL to read it and form its answer.

ChatGPT Clients

OpenAI

The ChatGPT mobile app (iOS and Android) doing a live fetch. The user-agent string includes the app version and device.

Claude-User

Anthropic

A user or application using Claude live-fetched this URL. The user-agent string may identify the specific Claude product, e.g., claude-code.

Perplexity-User

Perplexity

A user asked Perplexity a question. Perplexity fetched this URL to ground its answer.

Google-NotebookLM

Google

A user opened Google NotebookLM and sourced this domain. NotebookLM fetches every reachable URL within a sourced domain.

Google-ai-mode

Google

Google Search’s AI Overviews feature fetched this URL to include it in an AI-generated answer panel in search results.

Gemini-Deep-Research

Google

A user ran a Gemini Deep Research session. Deep Research makes many sequential fetches across multiple sources to compile a research report.

GoogleAgent-URLContext

Google

A user shared a URL with Gemini and asked questions about that page. Gemini fetched the URL live to answer questions about that specific content.

Amzn-User

Amazon

An Amazon Alexa or Amazon AI agent live-fetched this URL. Typically appears on reference and documentation content.

MistralAI-User

Mistral

A live fetch from a Mistral-powered product or API consumer.

When Brand Visibility cannot match a user-agent to a recognized pattern, it assigns the value Unknown. You can use the CDN User Agent dimension to identify what agent made those requests.

Status codes

HTTP status codes in this dataset indicate whether the AI agent received the content it requested.

Status

Name

Interpretation

200

The bot received the full response. The content was available for the AI to use.

304

Not Modified

The bot confirmed the content has not changed and used its cached version. The content was available.

301

Moved Permanently

The bot was redirected to a new URL. Each redirect adds an extra round-trip. High 301 volume on frequently crawled URLs means the redirect should be resolved at the CDN level.

302

Found (Temporary Redirect)

Same latency penalty as 301. Unlike 301, it does not signal a permanent move, so bots will keep hitting the original URL.

403

Forbidden

The CDN or origin blocked the bot. This can be intentional, e.g., through robots.txt rules or WAF policy, or unintentional, e.g., through overly broad rate limits. When AI fetches are blocked, that content cannot appear in AI answers.

404

Not Found

The URL does not exist. High 404 volume on AI agent types indicates the AI’s index contains stale URLs. Use the 410 status to tell crawlers to remove a URL from their index permanently.

429

Too Many Requests

The CDN rate-limited the bot. Sustained 429 errors on live-fetch agent types mean that users asking AI assistants questions about your content will receive incomplete or missing responses.

504

Gateway Timeout

The CDN stopped waiting for the origin to respond. The content did not reach the AI. When a page times out, the AI cannot access its content and cannot include it in an answer. High 504 volume on live-fetch agent types is a direct AI visibility risk.

Metrics

The following metrics are available to use as components in a data view once you have set up a connection that includes a Brand Visibility dataset. The Field column shows the source field in the CDN Requests Summary field group.

Metric

Field

Description

CDN Request Count

cdn.requests

The total count of CDN requests, summed from the requests field across all rows. Always use this metric to measure volume. Do not use row count.

CDN Error Count

cdn.status, cdn.requests

The count of requests that returned a 4xx or 5xx HTTP status code.

CDN Error Rate

Derived from CDN Error Count

The error count as a percentage of total requests.

CDN Avg Time to First Byte

cdn.timeToFirstByte

The average time in milliseconds from when the CDN received a request to the first byte of the response. CDN-cached responses are typically under 50ms. Responses served from the origin are typically 300ms to 700ms. AI live-fetch agents often show much higher values, which correspond to timed-out or very slow origin responses. High average values on live-fetch agent types are worth investigating as an AI visibility risk.

Boundaries

This dataset captures only bot traffic from CDN access logs. It does not contain the following:

User sessions, conversions, or engagement data. A user who clicks through from an AI answer runs the JavaScript on your page, so that visit is in your existing web data, not in this dataset. You can bring both datasets into Customer Journey Analytics and compare them for the same URL and host.
Any person identifier such as ECID. You cannot perform a person-level join from this dataset. The join operates at the URL and host level.
Sub-second time granularity. The timestamp is hourly. You cannot break down traffic within an hour into minutes or seconds.
Page content or rendered HTML. This dataset records the fact of the fetch and its outcome, not what the AI read from the page.
Conversion data. This dataset does not tell you whether an AI answer led a person to visit your site or convert. It holds aggregate CDN summary data, not person-based event data, so it does not link any request to an individual person or session.

recommendation-more-help

analytics-platform-help-main