AI Prompts for Supplier Evaluation

Supplier evaluation goes wrong when the criteria are decided after the proposals arrive, when scores are impressions rather than evidence, and when the cheapest bid is quietly weighted higher than the model says. A disciplined evaluation fixes the criteria and weights first, requires evidence for every score, and separates the commercial comparison from the capability assessment so neither contaminates the other.

The prompts here build the model, apply it to candidates with an evidence table, and prepare the verification step — the site visit or due diligence that tests what the proposal claimed. They are for selecting new suppliers; ongoing performance of existing suppliers is a different exercise covered by the supplier scorecards page.

Before you use these

Have these ready to replace the highlighted [variables]:

• What the supplier must deliver and the failure modes that would hurt most
• Candidate suppliers and their proposals or information
• Your weighting priorities (cost, quality, delivery, capability, financial stability, ESG, risk)
• Who scores and how disagreements are resolved

The prompts

• 1. Build the weighted evaluation model
• 2. Score candidates with an evidence table
• 3. Prepare the site visit and due diligence checklist

1. Build the weighted evaluation model

Act as a procurement lead designing a supplier evaluation model for [category].

Requirements: [technical, quality, delivery, service, compliance]
Priorities and rough weighting: [e.g. quality and continuity over price]
Non-negotiables: [certifications, financial thresholds, geographic or legal constraints]
Evaluators: [roles]

1. Define pass/fail gates first: requirements a supplier must meet to be scored at all. Keep this list short and objective.
2. Build a criteria tree with 4–6 top-level dimensions (e.g. technical capability, quality system, delivery and capacity, commercial, financial stability, risk and compliance, sustainability). Break each into 2–4 sub-criteria.
3. Assign weights top-down (dimensions sum to 100%) and within each dimension. Justify each weight against the priorities and the cost of failure on that dimension.
4. For every sub-criterion write scoring anchors on a 1–5 scale: what evidence earns a 1, a 3 and a 5. Anchors must be observable, not adjectives.
5. Specify the evidence source for each sub-criterion (proposal section, reference check, audit, financial statements, sample test).
6. Define the scoring process: independent scoring, then calibration meeting, then consensus — and the rule for handling a spread of more than 2 points.

Check the model for a hidden price bias: if commercial weight plus any cost-related sub-criteria exceeds the stated priority, say so.

2. Score candidates with an evidence table

You are scoring suppliers against an agreed evaluation model.

Model: [criteria, weights, scoring anchors]
Supplier information: [per supplier: proposal content, references, certifications, financials, audit notes]

For each supplier:
1. Check pass/fail gates. Stop and report if any fail.
2. Score each sub-criterion 1–5 using the anchors. For each score record the evidence (quote or reference the specific claim or document) and its quality: verified / claimed / absent. Where evidence is absent, score no higher than 2 and mark 'unverified'.
3. Compute weighted dimension scores and the total.

Then across suppliers:
- Ranking with totals and the gap between first and second.
- Sensitivity: does the ranking change if commercial weight moves ±10 points, or if all unverified scores are set to 1?
- The specific claims that most affect the ranking and must be verified before award.
- Shortlist recommendation and the questions to put to each shortlisted supplier.

Do not let a strong commercial offer lift capability scores. Do not infer capability from company size or brand.

3. Prepare the site visit and due diligence checklist

Act as a supplier quality and risk assessor preparing a due diligence visit to [supplier] for [category].

Claims that influenced their score: [list with the criterion each supports]
Highest-cost failure modes for us: [e.g. late delivery in peak, quality escapes, capacity shortfall, financial failure]
Visit: [duration, our participants, their participants]

Build the checklist:
1. For each claim: what to observe on site, what to ask, what document or record to collect, and what would contradict the claim. Prioritize by the claim's weight in scoring and the cost of it being false.
2. Capacity and delivery: how to verify demonstrated capacity, current utilization, order book, and what share of their capacity we would represent.
3. Quality system: records to sample (nonconformance, corrective actions, calibration, training), and the difference between a certificate and a working system.
4. Financial and continuity: signals to look for (working capital stress, key-person dependence, sub-tier supplier concentration), and questions for the finance conversation.
5. Culture and transparency: how the supplier handles questions about problems — note evasiveness as a finding.
6. Red flags that should stop the award regardless of score.

Output as a checklist with owner per item and space for findings. Keep it to what fits in the visit time; mark items that can be done by document request instead.

What scoring anchors look like

The evaluation model only works if a 3 means the same thing to every evaluator. Two sub-criteria from a typical model, written the way the first prompt asks for them:

Related prompts

• RFP and RFQ Drafting
• Supplier Scorecards and SLAs
• Supplier Risk Assessment
• Supplier Negotiation
• Sourcing Strategy
• Procurement Strategy

Logical next step

After this, most operations teams move on to RFP and RFQ Drafting.

All Operations & Supply Chain prompts · Search the full library