Agentic Ready Score

What is the Agentic Ready Score?

The Agentic Ready Score is Delante’s proprietary metric that indicates, on a 0–100 scale, how well a website or online store is prepared to be navigated and utilized by AI agents.

The score is calculated by the Agentic Ready Checker, an in-house tool built from scratch by Delante’s technical team that automatically scans a domain and evaluates the critical factors determining whether an AI agent can discover the site, understand its offering, and take action.

The Agentic Ready Score is not an AI visibility metric. It does not measure how often a brand appears in answers generated by ChatGPT, Gemini, or Perplexity. Instead, it evaluates whether a website technically enables an agent to complete a task: from reading the product offering, through checking price and availability, to proceeding to checkout.

How does the Agentic Ready Checker work?

The Agentic Ready Checker replicates the journey an AI agent takes through a website.

It sequentially checks whether an agent can:

discover the website and access files prepared specifically for AI systems,
understand the offering based on structured data and page content,
utilize a machine-readable interface (such as an API) to verify availability or perform an action,
determine that the website is trustworthy.

The audited elements are not arbitrary. Each corresponds to a specific hurdle an agent might face when attempting an autonomous purchase.

What does the Agentic Ready Checker check?

The Checker analyzes over a dozen elements grouped into categories.

Discovery and access files:

presence of an agents.md file containing instructions for shopping agents,
crawl permissions for AI bots, such as GPTBot, ClaudeBot, and PerplexityBot, in robots.txt,
sitemaps and /api or /agent paths,
.well-known files for agents and MCP,
RSS/Atom feeds signaling updates without requiring full-site crawling,
presence of an llms.txt file and validity of the links it contains (HTTP 200 status code).

Structured data:

JSON-LD markup, including Product, Organization, and FAQPage schemas,
completeness of ContactPoint markup, enabling agents to route inquiries to sales or support,
a canonical company facts page where AI models can source authoritative brand information.

Machine interfaces (APIs):

public OpenAPI/Swagger documentation,
an agent card or manifest located in the /.well-known/ directory.

Content and readability:

server-side rendering (SSR) of core content rather than client-side JavaScript execution,
semantic HTML: heading hierarchy, text density, and image alt attributes,
capability to serve content in Markdown format,
content freshness signals,
a smoke test verifying whether an agent can locate the price, target audience, and checkout path.

Trust:

presence of a security.txt file.

How is the Agentic Ready Score calculated?

Each category receives an individual score on a 0–100 scale.

The Agentic Ready Score is a weighted average of these category sub-scores:

Agentic Ready Score = weighted average of Discovery, Schema, API, Content, and Trust categories

How should the Agentic Ready Score be interpreted?

The raw numerical score provides little insight without context.

It is best analyzed:

broken down by category rather than as a single overall number,
benchmarked against competitors in the same industry,
tracked over time following successive technical implementations.

For instance, in a study of 15 Polish consumer electronics and home appliance stores, the average score was approximately 33/100, ranging from 4/100 to 67/100. In this category, a score of 50/100 represents a distinct advantage over most of the market, though it still leaves considerable room for improvement.

The study also revealed that business scale does not automatically translate into readiness—several smaller e-commerce stores outperformed major retail chains.

The Agentic Ready Score captures the state of the website at the exact time of the scan. Any modification to files, structured data, or site rendering methods can alter the outcome.