June 4, 2026 · Reference post · Web surface
The 131+ DST factors that make or break a website's agentic readiness
Every audit Appkitekt AI runs on a website rolls up to one score on one frame: Discovery · Selection · Trust. Below is a preview of the full list — only a subset of the 131 factors we score across metadata, per-agent crawl access, RAG retrievability, entity presence and trust signals. Run an audit to unlock every detail.
Why DST for websites
A score is only useful if it tells you where you're losing. DST gives every website factor a home in one of three places an agent's decision can break:
- Discovery (30%) — can the agent crawl, parse, and retrieve your site at all? Schema markup, llms.txt, per-agent robots rules, sitemap hygiene, entity presence in training data and grounding surfaces.
- Selection (45%) — once you're in the consideration set, do you get cited? API and tool clarity, decision-grade data, performance, intent alignment, brand preference in prompt testing.
- Trust (25%) — will the agent stake its answer on you? Authoritativeness, safety and compliance, sentiment, semantic consistency between site and API.
Web · 131 factors
Web surface — DST factor preview
What agents read on your marketing site, docs, schema, llms.txt and grounding surfaces across ChatGPT, Claude, Gemini, Copilot, Perplexity, Grok and Meta AI.
Discovery · weight 30% · 64 factors
Is the app visible to the model's training data and real-time retrieval (RAG)?
Technical Metadata (SEO for Agents) · 16
- Schema.org implementation (SoftwareApplication, Product, Service)
- JSON-LD presence and syntax validity
- OpenGraph (OG) tags for visual model previews
- Twitter Card metadata (twitter:app:id)
- Robots.txt allow status for GPTBot, CCBot, PerplexityBot
- /.well-known/ai-plugin.json or apple-app-site-association
- XML sitemap reachable at /sitemap.xml, lists all indexable routes, references API docs
+ 9 factors hidden
LLM Training Presence · 8
- Mention frequency in Common Crawl datasets
- Brand presence in Wikipedia (entity establishment)
- Backlinks from high-authority .edu and .gov sites
- Discussion density on Reddit
+ 4 factors hidden
Contextual & RAG Discovery · 20
- Semantic density of intent keywords
- Long-tail problem-solution phrasing in FAQs
- Regional availability data in metadata
- Pricing transparency in text (not images)
- Multi-language metadata
- Zero-shot brand recognition
- Few-shot brand recognition
- Presence in 'Best of [Category]' curated lists
+ 12 factors hidden
Per-Agent Crawl & Retrieval Access · 12
- Robots.txt explicitly allows GPTBot (ChatGPT training crawler)
- Robots.txt explicitly allows OAI-SearchBot (ChatGPT live search retrieval)
- Robots.txt explicitly allows ClaudeBot and Claude-Web (Anthropic crawlers)
- Robots.txt explicitly allows Google-Extended (Gemini training opt-in)
- Robots.txt allows Googlebot for Gemini grounded answers (separate from Google-Extended)
+ 7 factors hidden
Per-Agent Social & Grounding Surfaces · 8
- Verified X / Twitter handle with consistent bio — primary grounding surface for Grok
- Facebook Page + Instagram Business account + WhatsApp Business — Meta AI grounding surfaces
- Threads handle linked to the Instagram Business account (Meta AI cross-surface)
- LinkedIn Company Page complete with verified domain — primary B2B surface for Copilot
+ 4 factors hidden
36 factors in this section are only visible in a full ARO audit.
Run audit to unlockSelection · weight 45% · 37 factors
When an agent finds 5 options, why does it pick yours?
API & Tool Clarity · 8
- OpenAPI / Swagger documentation completeness
- Semantic clarity of function names
- Precision of parameter descriptions
- Use of required vs. optional parameters
+ 4 factors hidden
Decision-Making Data · 8
- Real-time availability indicators
- Dynamic pricing access via API
- Shipping speed/estimates as raw data
- Discount/coupon code accessibility
+ 4 factors hidden
Performance & Efficiency · 9
- API latency (Time to First Byte)
- Token efficiency (minimalist JSON payloads)
- Support for batch requests
- Rate limit transparency
+ 5 factors hidden
User Intent Alignment · 12
- Brand preference in prompt testing
- Cost-effectiveness perception (value vs premium)
- Niche specialization (generalist vs expert)
- Response relevance to the problem statement
- Historical selection rate (winner's circle)
+ 7 factors hidden
20 factors in this section are only visible in a full ARO audit.
Run audit to unlockTrust · weight 25% · 30 factors
Does the agent believe the info is accurate, safe, and authoritative?
Authoritativeness & Verification · 8
- Verified badges on social/business directories
- WHOIS data transparency
- Physical address verification
- Customer support response time metrics
+ 4 factors hidden
Safety & Compliance · 7
- Privacy policy clarity (GDPR/CCPA)
- OWASP Top 10 compliance for LLM apps
- Prompt injection guardrails
+ 4 factors hidden
Reputation & Sentiment · 7
- Sentiment score on Reddit/forums
- Ratio of verified-purchase reviews
- Public security bug bounty program
+ 4 factors hidden
Semantic Consistency · 8
- Information parity (website vs API)
- Uniformity of brand voice across LLMs
- Fact-check pass rate
- Lack of conflicting metadata
+ 4 factors hidden
16 factors in this section are only visible in a full ARO audit.
Run audit to unlockNext step
Unlock the complete factor list.
Run an ARO audit to see every signal scored for your website.