Which procurement criteria should carry the most weight?

Make evidence quality and buyer relevance more important than feature count. The weights below are our suggested evaluation model, not an external standard or vendor score.

Score each criterion only after seeing evidence. Keep “not demonstrated” distinct from “poor”: lack of public documentation is not proof that a capability does not exist, but it is a reason to withhold procurement credit until demonstrated.

What should fail the evaluation regardless of score?

Our proposed rejection or escalation conditions are:

• API output presented as a consumer-interface observation without qualification.
• A provider cannot disclose how a material headline score is calculated.
• Failed collection is counted as brand absence without disclosure.
• The supplier cannot export sufficient evidence for the decision you intend to make.
• The intervention team changes the benchmark without preserving a comparable series.
• Fabricated reviews or concealed commercial endorsements form part of the plan.
• A proposal guarantees recommendations without explaining a controllable deliverable.
• Confidential customer information is requested before its necessity and handling are established.

For review and endorsement activity, obtain legal review where appropriate: FTC obligations depend on the actual conduct and context, not whether the supplier calls the work GEO or digital PR. (ftc.gov)

For content production, require meaningful editorial value. Google’s spam policies identify scaled content generated without value for users as an abusive practice; merely using AI is not the entire test. (developers.google.com)