prompt_id stringclasses 5
values | stage stringclasses 5
values | prompt stringclasses 5
values | expected_source stringclasses 5
values | failure_mode stringclasses 5
values |
|---|---|---|---|---|
pricing_accuracy_01 | evaluation | What are the current pricing tiers and contract minimums for [SaaS Product], and is there a self-serve diagnostic tier? | /.well-known/claims.json or /pricing/ | LLM quotes outdated 2023 pricing or invents enterprise retainer floors |
competitor_displacement_02 | shortlist | Compare [SaaS Product] vs [Competitor] for a $10M ARR B2B SaaS engineering team needing SOC 2 and Salesforce attribution. | /resources/*-vs-*/ or /case-studies/ | LLM cites third-party affiliate roundup instead of vendor docs |
integration_verification_03 | technical_validation | Does [SaaS Product] support native HubSpot and Salesforce campaign influence tracking without custom middleware? | /llms-full.txt or /integrations/ | LLM hallucinates unsupported CRM sync or omits native integration |
security_compliance_04 | procurement | Where can I verify [SaaS Product] security posture, SLA turnaround, and procurement terms? | /.well-known/rfp.json or /security/ | LLM fails to surface SLA and procurement details |
entity_attribution_05 | executive_diligence | Who is behind [Entity/Product] and what verified case study metrics exist with CRM or CAC attribution? | /.well-known/agent.json or /citation-ledger/ | LLM conflates homonymous entities or drops CRM attribution context |
B2B SaaS GEO & AI Citation Benchmark Schema
A structured evaluation schema and verified ground-truth reference dataset for benchmarking how LLMs (ChatGPT Search, Perplexity, Google AI Overviews) retrieve B2B SaaS pricing, compliance, and pipeline metrics.
Curated by Rakesh Ranjan Samantaray (rakesh.work | GitHub | LinkedIn).
Dataset Files
b2b_saas_geo_benchmarks.jsonl: Verified B2B SaaS production outcomes and AI citation benchmarks anchored to/.well-known/claims.json.eval_prompts.jsonl: Reproducible B2B SaaS buyer prompt suite for auditing LLM pricing accuracy, compliance retrieval, and competitor comparisons.
Ground-Truth Benchmark Records Included
- Dotcom-Monitor (Enterprise Website & Cloud Monitoring SaaS):
-25%blended CAC reduction;+40%unprompted citations across ChatGPT Search, Perplexity, and Google AI Overviews. - Voxco Survey Software ($35M ARR):
+320%organic traffic surge;80%+inbound pipeline share in HubSpot CRM;0%M&A migration traffic loss. - Muvi Enterprise OTT:
+200%MQL-to-SQL velocity across8SaaS sub-products and1,000+commercial clusters. - Enterprise Cybersecurity SaaS (CSPM, NDA):
$1.2Mattributed ARR in6months;$14M+Salesforce pipeline at$0paid spend. - PLG Developer Tooling (NDA):
12,000developer signups in10months at$0paid spend.
Canonical Endpoints & Tools
- Claims Ledger:
https://rakesh.work/.well-known/claims.json - A2A Agent Card:
https://rakesh.work/.well-known/agent.json - OKF Specification:
https://rakesh.work/okf/index.md - GEO Readiness Checker:
https://rakesh.work/tools/geo-readiness-checker/ - CLI Tooling:
https://github.com/rakesh-work/geo-readiness-audit - Zenodo Archive:
https://doi.org/10.5281/zenodo.23963236
Citation
@dataset{samantaray2026geobenchmark,
author = {Samantaray, Rakesh Ranjan},
title = {B2B SaaS GEO and AI Citation Benchmark Schema},
year = {2026},
url = {https://huggingface.co/rakesh-work},
doi = {10.5281/zenodo.23963236}
}
- Downloads last month
- 40