Selection playbook · Data + AI Platform Providers

    How to choose a Data + AI platform

    Six steps from first brief to signed contract, built on 16 vendors evaluated in the PeMa Quadrant (latest edition 2026). Work through it here, or take the kit with you.

    1. 1Needs
    2. 2Shortlist
    3. 3RFP
    4. 4References
    5. 5Score
    6. 6Contract
    1

    Define what you need

    Agree these with your stakeholders before you speak to vendors. Tick them off as you go; your progress stays in this browser.

    0 of 8 agreed

    2

    Build a shortlist

    Start with three to five vendors. The PeMa Quadrant shows who has the market penetration and technology maturity; your needs decide who fits.

    3

    Ask the right questions

    Put the same questions to every vendor, in writing, so the answers can be compared.

    1. Demonstrate our top two use cases on our sample data, not a scripted demo.
    2. What is on the 12-month roadmap, and what has slipped in the last 12 months?
    3. How is pricing calculated at 3× our expected usage?
    4. Which integrations are native, and which need partner or custom work?
    5. How are models, data and prompts governed and audited?
    6. What is your uptime record and support response time in our region?
    7. How do we export our data and configurations if we leave?
    8. Which customers of our size and industry can we speak to?

    Questions our analysts are putting to vendors in this market right now

    • Which of the new Business Data Cloud lakehouse and Databricks features are generally available today, and how is federated query usage priced? Vendor
    • How will Microsoft guarantee Fabric and Foundry capacity in our regions given Azure demand growing 48% year on year? Vendor
    • When will watsonx Orchestrate Discover and Contain reach general availability, and how are they priced relative to existing Orchestrate licences? Vendor
    • When will Snowflake on OCI be generally available, in which OCI regions, and will feature parity match Snowflake on AWS and Azure? Vendor
    • Is Headless generally available and supported under existing licences, and how are actions taken by coding agents logged for audit? Vendor
    • When will AI Workspace and App Composer be generally available, and where does data processed by OpenAI models reside? Vendor
    4

    Check references

    Ask each finalist for two clients of your size and industry, and use the same script with each. Verified buyer reviews on each vendor's AIM profile help here too.

    1. What did you hire them to do, and did they deliver it?
    2. How close were the final cost and timeline to the original plan?
    3. How good was the team they actually put on the work, compared with the pitch team?
    4. How did they handle a problem or a missed deadline?
    5. What would you ask them to do differently?
    6. Would you hire them again for the same work? Why?
    7. Who on your side should we also speak to?
    5

    Score the finalists

    Rate each finalist from 1 to 5 against weighted criteria. The score updates as you go; download the sheet to share it with your team.

    CriterionWeight
    Capability for our use casesProven in our proof of concept30%
    Integration and architectureFits our stack and identity15%
    Security and governanceControls, audit, residency15%
    Vendor strength and roadmapViability, PeMa placement, direction15%
    Total cost over three yearsLicences, usage, services15%
    Support and ecosystemPartners, community, response times10%
    Weighted score–––
    6

    Agree the contract

    Check the contract covers these before you sign.

    • Pricing protection on renewal and for added usage
    • Uptime SLA with service credits
    • Data ownership, export format and deletion on exit
    • Security, privacy and sub-processor commitments
    • Roadmap commitments that matter to you, in writing
    • Support tiers and response times
    • Termination rights if key features are withdrawn
    • Implementation partner responsibilities, if one is involved

    Price benchmarks

    What buyers pay in Data + AI Platform Providers

    Collecting data: 0 of 5 data points. Ranges appear once at least 5 buyers or RFI responses have shared prices for Data + AI Platform Providers. Individual prices are never shown.

    Bought from a vendor in this market? Add what you paid when you review the vendor; it takes a minute and stays anonymous.

    Want an analyst to run this with you?

    Our analysts cover Data + AI Platform Providers all year. Ask a question or book a 30-minute call; it is free for enterprise buyers.