Score evidence, not promises. The following weights are a suggested starting point for an established contractor; adjust them before demonstrations, not after seeing the supplier you prefer.
| Criterion | Suggested weight | Evidence required |
|---|---|---|
| Workflow correctness | 25% | Passed scenarios and accurate records |
| Integration and recovery | 20% | Supported actions, controlled retries and demonstrated failure recovery |
| Human escalation and customer handling | 15% | Successful handovers, truthful fallback and agreed coverage |
| Full-year cost | 15% | Written, usage-sensitive quotation with exclusions |
| Data governance and security | 10% | Processing map, contractual evidence and implemented controls |
| Support, ownership and exit | 10% | Named responsibilities, usable exports and handover demonstration |
| Staff usability | 5% | Office-team acceptance and manageable exception workload |
Use unproven, not an assumed pass, where evidence is missing. Hard failures should remain disqualifying even if the weighted total would otherwise look attractive.
Ask references questions that reveal operating reality:
- What did the system get wrong after launch?
- How much correction work remains?
- What happens outside support hours?
- Which charges were not obvious initially?
- Can staff change routine rules themselves?
- Have you tested an outage or an exit?
- What evidence distinguishes additional work from bookings you would have won anyway?
PUT THE RESEARCH TO WORK