Resources
What a ZenXData evaluation report looks like
The structure below is what you receive at the end of an evaluation engagement. Content shown is illustrative and does not describe any real client.
- Section 1
Executive summary
The decision being informed, the headline findings, and the top recommended actions on one page.
- Section 2
Scope and objectives
System under test, versions, use cases, languages, risk areas and what was out of scope.
- Section 3
Method
Test set construction, rubric definitions, evaluator profile, calibration and agreement approach.
- Section 4
Results by category
Scores broken down by task type, segment or risk category, with comparisons where relevant.
- Section 5
Failure analysis
Failure categories with representative examples, frequency and severity.
- Section 6
Recommendations
Prioritized fixes, data or prompt changes, and suggested retest or monitoring plan.
- Section 7
Appendix and data
Full scored dataset, rubric, and test cases delivered in your preferred format.
Example findings table
Illustrative format only — not real results.
| Failure category | Severity | Example |
|---|---|---|
| Grounding / factuality | High | Answer cites a policy not present in the retrieved documents. |
| Instruction following | Medium | Response ignores a requested output format. |
| Safety | High | Model follows an injected instruction embedded in user content. |
Have an AI project?
Send us what you are building. We will scope the data, evaluation, benchmarking or testing program and come back with a proposal.