Buyer decision tool
A capability claim and the evidence behind it are rarely the same size. This matrix produces a claim, evidence, dependency and decision record, so you can see whether the proof covers the exact people, role and work you are buying.
Use it: When a supplier says "yes" and you need to know what that answer rests on.
Proof scale
300+ languages 1,000+ brands served since 2015 across AI data, translation, localization, media, and interpretation.When to use this
The point is not to collect more attachments. It is to find out whether the evidence supports the exact people, role and work you are buying.
What it produces
The checklist
Every question below is on the page as text. Use the link on any row to send a colleague straight to that one question.
| Ask this | A useful answer is specific about | What the answer actually tells you |
|---|---|---|
| Which exact role, task and language variety does the supplier’s evidence support? # | The precise work being claimed, including direction or variety where it matters. Cross-check it against the Language Coverage Decision Matrix. | Whether the evidence is relevant to this assignment or only supports a broader claim. |
| Which person, team or process does the evidence actually support? # | A specific person, team or process, linked to the work being assessed. | Whether the proof can be traced to the resource or method you may actually buy. |
| What can you inspect, and is it still current? # | The sample, test or record itself; the instructions it was produced under; when it was reviewed; and who reviewed it. | Whether "submitted" has been mistaken for "examined and accepted". |
| What is already in place, and what still depends on recruiting, training or access? # | Validated resources, confirmed availability, remaining onboarding, required training, and buyer-side decisions. | Whether the team is actually ready or still being assembled. |
| What does "end to end" include? # | Who owns production, project coordination, consent and privacy handling, metadata, quality review, and final acceptance. | Whether the claim covers the whole workflow or hides hand-offs back to the buyer. |
| What capacity is available for this exact work and period? # | People who match the task and language requirement, their confirmed availability, and any dependency that could change it. | Whether the answer is usable capacity or an organisation-wide headcount. |
| What decision will a sample or test support? # | What the sample tests, who judges it, and what result means proceed, revise, or stop. | Whether the exercise can change the buying decision or is only activity. |
| If prior work cannot be shared, what can test the same claim safely? # | A buyer-owned test, controlled sample, redacted process record, or another substitute that both sides are allowed to inspect. | Whether confidentiality can be respected without asking the buyer to accept an unsupported claim. |
How to fill it in
Work down the first column with the supplier in the room. A row is only answered when the middle column is specific — a yes that is not specific about anything is the thing this tool exists to catch. Write the unresolved rows down; those are the ones worth a call.
What counts as evidence
A named owner, a document you can open, a sample produced under the instructions you will actually use, or a test whose result changes the decision. A capability statement, an organisation-wide headcount, or a claim about a different assignment does not answer these rows.
The document
The PDF needs no sign-in and is free to forward or quote. The working document is there if you would rather copy the checklist into your own process. If either will not open for you, tell us and we will fix it the same day.
Where this fits
The guide explains what to look for; this checklist is what you take into the conversation. Both are published in full.
A procurement framework for rubric fit, rater calibration, IAA diagnostics, security, and multilingual pilot evidence.
Read the guideA procurement framework for terminology control, reviewer independence, severity scoring, regulated-content escalation, security, and release-ready LQA reporting.
Read the guideStuck on a row?