Skip to main content

Your first dive: Test of Details

This 15-20 minute walkthrough will get you running your first Test of Details from start to finish.

Video tutorials

Before you start

Make sure you have a project with documents in your Workspace. If you haven't set this up yet, see Set up your first project.

Step 1: Prepare your documents

Make sure your Workspace contains the documents you'll be testing against (invoices, contracts, etc.). If you synced from SharePoint, verify the files have finished processing.

Step 2: Upload your test sample

Prepare an Excel file containing the transactions you want to test. This is your test sample — the list of items Moby will verify against supporting documents.

Step 3: Create and run the TOD

  1. Go to Tests of Details in the sidebar
  2. Click New Test
  3. Select your test sample and supporting documents
  4. Configure the matching criteria
  5. Click Run and wait 10-15 minutes for Moby to process

Step 4: Review and finalize

Once Moby completes the analysis, you'll review the results, validate what’s correct, resolve anything flagged for review, and export.

Use the full review workflow here: Run a Test of Details → Reviewing AI Results.

Quick checklist:

  • Confirm matches — unmatch/rematch if needed (see Matching Documents)
  • Verify values with grounding — click any value to see its source in the PDF
  • Validate — mark cells or rows validated once confirmed
  • Handle exceptions — correct values or document your conclusion

Step 5: Export to Excel

Once your review is complete, click Export to generate your audit-ready Excel file.

For a more integrated experience, use the Moby Excel Add-in — it brings your results and grounding directly into Excel, letting you view source documents side-by-side with your data.

Learn more: Excel Add-in

Troubleshooting

If your test didn't go as planned:

  • No matches found? Check that your test sample has clear headers and at least one matching field (Invoice Number, Amount).
  • Wrong documents matched? Review your matching instructions.
  • Extraction looks wrong? You can edit the model to improve field descriptions.

For common issues, see Troubleshooting.