Cheap and well-lovedisn't the same as worth renewing.

Each AI/SaaS vendor on two axes from your own data — reliability (uptime/SLA, incident load, support responsiveness) and value (adoption, value vs cost, switching alternatives) — into a 0–100 spend-justification score, plus a verdict: RENEW, RENEGOTIATE, or DO NOT RENEW. You also flag whether the vendor is business-critical, which arms the reliability gate. It grades the vendor relationship, never a person, and applies no industry multiplier — every input is your evidence (status pages, incident logs, invoices, usage).

Because reliability is a veto, not just a weighted input. The reliability-floor gate is dispositive: if a vendor is business-critical and fails on any reliability dimension (uptime, incident load, or support), it's DO NOT RENEW no matter how high the justification score — strong adoption and value can't buy reliability back. In the worked example, the Core LLM API justifies at 75 on great economics but it's a critical dependency with failing uptime, so it gates to DO NOT RENEW. A weighted average would quietly average that failure away; this won't.

You decide — it's a flag you set for any vendor your operation genuinely depends on, where an outage hurts customers or stops work. The gate only applies to those: a critical vendor failing the reliability floor can't be renewed on economics alone. A non-critical vendor with the same reliability scores just loses points on the reliability axis; it isn't vetoed. That's deliberate — you can tolerate a flaky nice-to-have, but not a flaky core dependency.

Different object. A SaaS/subscription auditor grades utilization — paid vs used seats, right-sizing — and would happily keep a cheap, well-adopted tool. This grades reliability and renewal justification, and it won't renew that same tool if it's a critical dependency that keeps going down. The reliability veto is the part no seat auditor has. They're complementary: right-size seats with the auditor, decide renewals with this. It's also distinct from the Vendor Data-Flow Register, which grades data-flow governance (DPAs, cross-border), a privacy object.

No — it's a renewal compass, not a finance model or a contract valuation. It scores whether the spend is justified relative to reliability and value from the inputs you provide, and names the one thing to fix first (renegotiate price, demand an SLA, scope an alternative). It doesn't price contracts or model ROI in dollars; confirm renewal, contract, and budget decisions with the relevant owners. It's a planning aid, not financial advice.

One .xlsx Vendor Scorecard (the two-axis score, the reliability-floor gate, the per-vendor verdict and fix-first step), a Facilitator Playbook, a Vendor-Review Runbook, and a worked six-vendor example. It's deterministic and offline — opens in Excel, Google Sheets, or Numbers, nothing uploaded. Run it as a repeatable vendor-review cadence ahead of renewals.