[ vendor-rfp:// ] experimental
Describe what you need → get an RFP outline: evaluation criteria, must-haves, nice-to-haves, must-NOT-haves, scoring rubric.
// system prompt
You generate RFP outlines for vendor selection. User describes what they're buying. Output: # RFP: <Product Category> ## Background (for vendors) <2-3 sentences. Who we are. What we're solving. Why we're buying now.> ## Scope - <scope item> - <scope item> ## Evaluation Criteria (with weights) CRITERIA WEIGHT NOTES -------- ------ ----- <each criterion + % weight + one-line note on what excellent looks like> ## Must-Haves (table-stakes) - <hard requirement> - <hard requirement> ## Nice-to-Haves (differentiators) - <nice-to-have> - <nice-to-have> ## Must-NOT-Haves (dealbreakers) - <if vendor X is true, eliminate> - <…> ## Questions for Vendors ### Architecture / Capability 1. <question that surfaces real capability vs marketing> 2. <…> ### Operational 1. <…> ### Commercial 1. <…> ### Risk / Exit 1. <data-portability / lock-in question> 2. <…> ## Scoring Rubric Each criterion scored 1-5 by 3 evaluators independently. Aggregate = avg of scores × weight. Top 3 by aggregate score → reference calls. ## Reference-call Questions (for finalists) 1. <question that surfaces real-life experience> 2. <…> Rules: - Weights sum to 100%. - Must-Haves and Must-NOT-Haves are filters BEFORE scoring — failing one disqualifies. - Questions for Vendors must distinguish vendors. "Tell us about your product" doesn't. "Show us your data export format and a sample exported file" does. - "Must-NOT-Haves" are critical for vendor selection — most RFPs omit them. Include: must not require multi-year lock-in, must not store data outside named regions, etc. - Reference-call questions go beyond happy-path; ask about a specific incident or churn.
⚡ powered by Cloudflare Workers AI · quota deducted on success
// output
// sample output
# RFP: Customer Data Platform (CDP) ## Background (for vendors) We are a mid-market B2B SaaS (~50 internal users) selecting a CDP to consolidate ~250k customer records across HubSpot, Salesforce, and Snowflake. We are currently running fragmented per-tool segmentation; we need a unified customer profile for marketing campaigns and account-team enablement. We are budgeting approximately $120k/year and want to be live in 90 days. ## Scope - 250k customer records, growing ~15% YoY - 50 internal CDP users (marketing, account managers, ops) - Bi-directional sync: HubSpot (CRM) + Salesforce (sales) + Snowflake (warehouse) - Real-time segmentation for marketing campaigns - GDPR-compliant data handling (EU customer base) - US-business-hours support; on-call coverage for production issues ## Evaluation Criteria (with weights) CRITERIA WEIGHT NOTES -------- ------ ----- Time-to-First-Value 25% Excellent = 30-day starter integration with first segment live by day 60. Integration depth (HubSpot/SF/Snowflake) 20% Excellent = pre-built connectors with bi-directional sync; not just CSV export/import. Data portability / exit 15% Excellent = full schema + data export on demand, no fee, no lock-in clauses. GDPR / compliance posture 15% Excellent = ISO 27001 + GDPR DPA + EU data-residency option. Total Cost of Ownership (3yr) 10% Includes platform fee + implementation + integration maintenance. Support quality 10% Excellent = named CSM + < 4h response SLA in business hours. Product roadmap fit 5% Excellent = published roadmap; alignment with our needs (real-time, AI-segment). ## Must-Haves (table-stakes — disqualifies if missing) - GDPR-compliant with DPA available - EU data residency option (or all-EU hosting) - Native HubSpot + Salesforce integrations (bi-directional) - Snowflake reverse-ETL or native connector - SOC 2 Type II or equivalent (current within 12 months) - Multi-user with role-based access control ## Nice-to-Haves (differentiators) - AI-assisted segment building or audience suggestion - In-product collaboration / commenting on segments - Self-serve identity resolution rules - Built-in lifecycle / journey orchestration - Public API with SDKs in at least Python and JavaScript ## Must-NOT-Haves (dealbreakers) - Multi-year minimum commit (we will not sign 3+ years) - Data-export fees or "data egress" charges - Contractual lock-in on schema (must be able to take our data with us in standard formats) - Sub-processor in country/region restricted by our customers - Headquartered in / hosted in jurisdiction we cannot legally process EU data through ## Questions for Vendors ### Architecture / Capability 1. What is your identity-resolution model — deterministic, probabilistic, or hybrid? Provide a worked example for two records merging. 2. How do you handle CRM record-level access controls? If our Salesforce has field-level security, does that propagate into your platform? 3. What's your real-time segmentation latency from event-in to segment-membership-out (p50 and p95)? 4. Provide one concrete example of how you handle a "deleted in source" cascade (GDPR right-to-be-forgotten). 5. Send us a 100-record sample data export in your standard format. ### Operational 1. What's your in-business-hours support SLA for P1 issues? Provide the SLA contract page. 2. Who would be our named CSM if we sign? What's their tenure at your company? 3. What's your average implementation timeline for a mid-market customer with our 3-tool integration shape? 4. Show us the dashboard a customer admin sees on day 1, day 30, day 90. ### Commercial 1. Standard contract term — annual? With what notice period for non-renewal? 2. Pricing model — per-record, per-user, per-event, flat? Show us pricing at our scale. 3. What's included in implementation cost, what's extra? 4. What's the renewal-price escalator clause typically? ### Risk / Exit 1. If we terminate, what's the data return mechanism? Format, timeline, fees. 2. Have you been acquired in the past 5 years, or had a leadership change? How did existing-customer relationships change? 3. What's your customer churn rate? (If they won't share, ask the reference calls.) 4. Where are your sub-processors located? (Name them, not just regions.) ## Scoring Rubric Each criterion scored 1-5 by 3 evaluators (PM, Engineering lead, Data lead) independently. Aggregate = average × weight. Top 3 by aggregate score → reference calls. After reference calls, final vendor selection by PM + VP of Operations. ## Reference-call Questions (for finalists — 30 min each) 1. Describe the most painful incident you've had with this vendor and how they handled it. 2. What's the gap between what the sales process said and what implementation delivered? 3. Who from the vendor side has been most helpful, and have they stayed (or have they churned)? 4. If you were starting the procurement again, what would you ask that you didn't? 5. What does the vendor do badly that they'd never admit?
// powered by cloudflare workers ai · quota deducted on success ← back to catalog