Procurement software companies often use the same words for different kinds of work. A product might find a contract clause, recommend a response, prepare an approval or make a change in another system. Calling all four activities AI tells you very little about what the product actually does.
WOP's vendor coverage starts with that distinction. The purpose is to help practitioners understand the products, the evidence behind their claims and the practical implications of adopting them.
This is the assessment standard we are introducing for new coverage. Older landscape articles will be reviewed individually. Their original review dates remain visible until that work is complete.
The same questions, every time
Every member assessment addresses the same eight criteria:
Work: the specific procurement job and the people doing it.
Role: what the product records, controls, interprets or executes, and what remains in other systems.
Capability: what is available today and what is a preview, roadmap claim or unknown.
Authority: what the software may recommend, prepare, submit or execute, and who approves consequential actions.
Dependencies: the data, integrations, configuration and operating ownership needed.
Experience: what is documented or observed about requester, operator, administrator and supplier journeys.
Outcomes: whose results are reported, how they were measured and what they establish.
Fit and limits: where the evidence supports the approach and what could change the assessment.
These questions prevent a product description from becoming a polished restatement of a vendor's website. They also give readers a consistent way to move between articles. The answers can differ considerably, but the questions remain stable.
Category coverage adds more specific questions. Contract intelligence needs scrutiny of extracted clauses, amendments and obligation ownership. Supplier risk needs scrutiny of entity matching, risk coverage and the route from alert to action. Orchestration needs scrutiny of handoffs, permissions and the systems that remain authoritative.
Classify the work
A supplier onboarding product and a supplier risk data service can both discuss supplier management. They may still solve different problems for different users. We classify the product by its primary job, then explain meaningful overlaps.
We also identify its role. It may maintain an authoritative record, route work and approvals, generate intelligence, or execute an action. Some products combine these roles. The article should explain which role is central and which capabilities support it.
AI is considered within that operating context. An agent that prepares a purchase requisition has a different authority boundary from one that submits it. A product's use of the word autonomous does not settle that question.
Show the basis of the assessment
We distinguish who made a claim from how it was examined. A vendor-authored case study remains vendor-authored even when it names a customer. Several articles repeating the same announcement do not provide several independent confirmations.
Our evidence descriptions are:
Stated: a source asserts the capability, but gives insufficient detail to establish its operation.
Documented: the reviewed material explains the specific behaviour, scope or controls.
Demonstrated: a demonstration was reviewed, with its scenario and environment recorded.
Tested by WOP: an actual WOP test was completed and recorded.
Customer-reported: an attributable report describes a deployment or outcome. We identify its publisher and context.
Not verified: the available evidence is insufficient, inaccessible or contradictory.
These descriptions are not ratings. A documented feature can still be difficult to deploy. A successful customer implementation does not establish that every capability works in every environment. A demonstration can explain a workflow without proving its behaviour at production scale.
Desk research is labelled as desk research. We do not present it as hands-on testing. Suggested scenarios are labelled as examples to investigate, rather than experiences we have had with the product.
Keep unknowns visible
There is a meaningful difference between saying a feature is missing and saying we could not verify it. Public documentation may be incomplete. A capability may exist in a particular configuration, edition or release. A statement may refer to a roadmap rather than a generally available product.
When that distinction is unresolved, we say so. An unknown should lead to a better question, rather than a confident negative conclusion.
Material claims carry a source and research date in our working evidence record. Articles link the evidence supporting their main conclusions. Where sources disagree, we preserve the disagreement until it is resolved.
Explain judgement
WOP's interpretation is part of the value of the coverage. Readers should still be able to distinguish it from a product fact.
If an approach appears relevant to a complex enterprise environment, we explain why. We also identify what could change that view. Integration capacity, internal process ownership, data quality and the scope of the actual product can all matter more than the vendor's preferred customer label.
We do not use an overall vendor score. Product capability, strength of evidence, market presence and suitability for an organisation are different questions. Combining them into a single number can conceal the differences that practitioners need to understand.
Relationships and corrections
Relevant employment, sponsorship, consulting, affiliate and product-access relationships should be disclosed with the coverage. A relationship does not determine an article's conclusion. Readers should have enough information to judge the context themselves.
Vendors can offer factual corrections supported by inspectable evidence. We assess the evidence and record material corrections. That process does not give a vendor editorial control over the assessment.
Free coverage and membership
Each vendor article starts with a free vendor-at-a-glance reference: its job, users, category, advertised capabilities, customer evidence, source links and research basis. Members receive a developed assessment against all eight criteria, with vendor-specific findings, interpretation and limits. The paid section assesses the vendor; it is not a generic buying checklist.
The evidence standard is the same in both. Methodology, material limitations, relevant disclosures and the sources supporting the central assessment should remain accessible to free readers.
Tracked sources are checked monthly, with substantive category reviews planned quarterly and earlier reviews when a material change affects an article. A check is not automatically a new assessment. Publication, source-check and substantive review dates mean different things.
WOP methodology 2.0, introduced 11 October 2026.