Start with the evidence you actually have

Northstar receives M-12’s return: 60 people reached, marked as an estimate, and 140 attendance entries. A supporting document covers the previous quarter. A comment says that scheduling made participation difficult. These are different kinds of evidence, not a single answer waiting to be summarized.

Begin with completeness and consistency. Is the period clear? Are people and visits defined? Does the attached document support this return? Then ask the substantive question: what does the evidence suggest about participation and support needs?

On-arrival analysis means you have planned these questions and checks in advance. It does not mean every finding is automatically approved or that the system must issue real-time notifications. Analysis timing, routing and human review depend on the configured workflow.

Separate a flag from a conclusion

The two counts may both be valid: one person can attend more than once. The document-period mismatch is a concrete clarification need. Neither observation proves misconduct or poor performance. Record what is known, what is inferred and what remains unanswered.

A useful review record contains the source reference, proposed finding, relevant definition, uncertainty, reviewer and status. If the contributor supplies a current document, update the review while preserving the original evidence and reason for the change.

For comments, begin with a small codebook relevant to the question. A scheduling-barrier category needs a definition and an example. Allow multiple themes where appropriate, and retain an unclassified or uncertain state. Do not force every comment into a category merely to produce a neat chart.

Keep meaning through languages and formats

Northstar may receive feedback in several languages, translated notes or document extracts. Preserve the original wording, language and source. A translation or extraction is a working representation; it should not silently replace the original.

Use a competent reviewer to check consequential or ambiguous interpretations and a varied sample of routine ones. Record what was reviewed and what remains provisional. When review capacity is limited, report that limitation rather than treating unreviewed responses as having no issue.

A shared codebook helps organize analysis but does not make groups comparable by itself. Invitation coverage, wording, collection conditions and interpretation quality also matter. Counts of themes describe the material reviewed, not everyone in the network.

Use the assistant to prepare evidence for judgment

Instead of asking for a broad summary, specify the authorized records, question, definitions and evidence requirements. Request source references and explicit gaps. Uploaded content supplies evidence; it does not authorize new access or external actions.

For this exercise, ask for participation-related themes in the approved returns, with a source passage for each and a list of unresolved period or unit issues. The reviewer can then check the result rather than reconstruct the analysis from a polished paragraph.

The workflow differs from repeatedly downloading comments and reading attachments afresh because the criteria, sources and review history continue with the records. AI preparation still needs verification. A fluent response is not proof that the interpretation or calculation is correct.

Work through the Northstar example

This is a fictional teaching example. Adapt the fields and rules to the question your own network needs to answer.

Add this part to your plan

Fill workbook page 3 with two review records: the document-period issue and the scheduling theme. Include the source, interpretation, uncertainty, owner and next action.

Self-check: Can another reviewer distinguish evidence from interpretation and reproduce the basis of your finding?

Watch: why qualitative evidence gets overlooked

A 2 minute 34 second companion overview. The lesson and exercise above stand on their own.

Watch on YouTube if the embedded player is unavailable

Questions you may have

Does analysis on arrival mean no human review?

No. It means the analysis questions and criteria are ready when information arrives. Human review remains necessary where interpretation, clarification or a consequential decision requires it.

Can an untranslated response be counted as no issue?

No. Record it as awaiting interpretation. State the analysis base and pending responses so incomplete review does not become a misleading result.

Apply the method to your work

Use the five-part plan to assess the sources, analysis and permissions your team needs. The solution page shows where a connected platform can support that workflow.