How is the weighted score calculated?

For each vendor the tool multiplies each criterion score (1–5) by that criterion's weight (1–5), sums the products, and normalises to a percentage of the maximum possible. Higher-weighted criteria move the result more, so a vendor that wins on what you care about ranks first.

Why use weights instead of a simple average?

A simple average treats "nice-to-have UI polish" the same as "must-have SOC 2 compliance". Weighting forces you to declare priorities up front, which makes the decision defensible and prevents a strong vendor on trivia from beating a strong vendor on essentials.

Can I add my own criteria and vendors?

Yes. Add or remove vendors and criteria freely. Common criteria are pre-loaded — security, data residency, uptime SLA, context window, price, support, and ecosystem — but you can rename them to match your RFP.

Is my evaluation data saved anywhere?

No. All scoring and ranking happen in your browser. Nothing about your vendors, weights, or scores is uploaded or stored on a server.

AI Vendor Comparison Scorecard

AI vendor comparison scorecard

Choosing an AI vendor on vibes is how organisations end up locked into a contract that fails the security review six months later. This scorecard makes the choice explicit: you declare what matters and how much, score each vendor on those criteria, and the tool ranks them by a weighted total. The output is a defensible artifact you can drop straight into a procurement deck.

How it works

You start with common enterprise criteria — security & compliance (SOC 2 / ISO 27001), data residency, uptime SLA, context window, price, support, and ecosystem maturity — each with an importance weight from 1 to 5. Then you score each vendor 1–5 on every criterion. For each vendor the tool computes a weighted sum (score × weight, summed across criteria) and normalises it to a percentage of the maximum achievable, then ranks vendors highest-first. Add or remove rows to match your real shortlist and RFP.

Tips and notes

Set the weights before you score the vendors — deciding priorities while staring at the scores invites bias toward a favourite. Reserve a weight of 5 for true dealbreakers (a missing SOC 2 report should sink a vendor regardless of how good the demo was). Score from evidence, not marketing: trial the models, read the DPA, and ask for the actual SLA credit terms. When two vendors finish within a few points, the ranking is a tie — break it on a hands-on pilot or contract flexibility, not by nudging a score. Everything stays in your browser, so you can iterate freely.