The standard-tool problem
73/ 100
An opaque score with no repair path
A proprietary readiness score can look precise while hiding the response, crawler rule, route, or data block that actually needs correction.
For the technical reviewer
Built on deterministic code checks, not black-box AI guesses.
Growthract’s core verification layer does not ask an AI model to decide whether your infrastructure passed. We inspect observable network responses, raw HTML, crawler rules, structured data, and metadata against explicit technical checks.
Request Technical ReviewObservable inputsBinary findingsDocumented remediation
deterministic-checks.log
$ growthract inspect example.com
GET /robots.txt 200 PASS
GET /llms.txt 200 PASS
initial HTML found
JSON-LD parse valid
canonical metadata found
5 checks · 5 observed resultsEvidence before interpretation
Deterministic vs. heuristic
A score is not a technical finding.
Narrative summaries can help explain a result. They should never replace the response, rule, route, or syntax failure that produced it.
The Growthract approach
200 / PASS404 / FAILJSON / INVALID
Verifiable input, result, and remediation
An endpoint responds or fails. A JSON block parses or does not. Required content is present in the initial HTML or absent. Every finding points back to observable evidence.
Three inspection pipelines
Each infrastructure layer produces separate evidence.
Protocol access, first-response content, and entity data are inspected independently so a passing result in one layer cannot conceal a failure in another.
APipeline A
Protocol & retrieval
robots.txt & /llms.txt
Output
Status · headers · rule match · Markdown structure
What it executes
- Fetches /llms.txt and records its live HTTP status, content type, headers, and Markdown structure.
- Fetches robots.txt and parses applicable user-agent groups, Allow rules, and Disallow rules.
- Evaluates automated search crawlers including OAI-SearchBot, PerplexityBot, and Claude-SearchBot.
- Reports user-triggered agents such as Claude-User separately because provider access semantics differ.
Why it matters
This establishes the response and crawler rules visible at scan time. It does not claim that a third-party platform has already crawled, indexed, or used the content.
BPipeline B
DOM & hydration
SSR / ISR inspection
Output
Initial HTML · content presence · metadata · hydration risk
What it executes
- Fetches the initial HTML response without waiting for browser-side hydration.
- Searches the raw response for agreed product positioning, pricing, feature, and conversion content.
- Identifies content that appears to depend entirely on client-side rendering or delayed browser API calls.
- Inspects titles, descriptions, canonical URLs, and Open Graph output present in the response.
Why it matters
This verifies whether critical content is present on the first network response. It does not assume that every crawler executes—or ignores—the same JavaScript lifecycle.
CPipeline C
Entity & schema
JSON-LD & schema-dts alignment
Output
JSON parse · entity types · properties · @id relationships
What it executes
- Extracts application/ld+json blocks from the initial HTML document.
- Parses each block as JSON and reports malformed or inaccessible structured data.
- Checks agreed entity types such as Organization, WebSite, SoftwareApplication, Product, Service, and Offer.
- During implementation, uses TypeScript and schema-dts to type-check the agreed entity components inside the repository.
Why it matters
This confirms JSON syntax, expected properties, and the agreed entity structure. It does not guarantee how an external model will interpret, rank, or cite the entity.
Pricing + scopeEngagement blueprint
Know the scope before you commit.
See how Growthract structures diagnostics, implementation and technical search engagements, then choose the level of support that fits your business.
Clear scopeDefined deliverablesPractical engagement paths
Designed for engineering trust
Observable scans. Controlled implementation. Reviewable output.
Infrastructure
Server-side scan execution
External audit tasks run server-side and return results after the requested routes respond and parsing completes. Timing depends on network conditions, target availability, and route count.
Infrastructure
Read-only external access
The diagnostic performs safe retrieval requests and response analysis. It does not write to your website, modify production code, or require production credentials.
Infrastructure
Reviewable implementation
Sprint changes are delivered through isolated, documented pull requests or reviewable patches. Your engineering team controls review, merge, and deployment.
Technical field guide
The Entity Science Guide
A deeper technical guide to entity identity, structured relationships, retrieval infrastructure, and the signals that help machines build a more coherent representation of a company.
- Stable entity identifiers
- Organization and product relationships
- Route-level authority
- Machine-readable context
entity-graph.json
Organization
├── owns → WebSite
├── offers → Product
├── serves → Audience
└── describedBy → WebPage
Product
├── category
├── featureList
├── offers → Offer
└── provider → OrganizationThe guide explains how these relationships support clearer entity interpretation without claiming guaranteed ranking or citation outcomes.
Verification boundary
We verify the observed response—not future platform behavior.
What the engine verifies
- Observed HTTP status, headers, and content type
- Applicable robots.txt rules for named user agents
- Raw first-response HTML content presence
- JSON parse validity and agreed entity properties
- Metadata and canonical output on inspected routes
What controls the rest
OAI-SearchBot, PerplexityBot, and Claude-SearchBot are automated search crawlers with provider-controlled schedules. User-triggered agents such as ChatGPT-User, Perplexity-User, and Claude-User may retrieve pages under different access rules.
Corrected infrastructure is available after deployment, but Growthract does not guarantee when an external platform will crawl, index, retrieve, rank, or cite it.
Observed bot activity
“Allowed to crawl” and “did crawl” are different facts.
robots.txt can show whether a crawler is permitted to request a route. Server logs can show whether a recognized bot actually requested the site during the log window being examined.
Crawler Logs provides that second layer of evidence without uploading the supplied log data.
Provider references