MICROEXTRACT DOCUMENTATION

Template Builder, Fields, Tables and Output Schema

A good template describes the business output clearly while leaving document interpretation flexible enough to handle real document variation.

1. What Template Builder means in the current product

Template Builder is the guided MicroExtract workflow for selecting an approved mapping profile, choosing output formats, uploading matching output templates and validating those templates before processing.

It is not a promise that every customer can redesign the underlying extraction schema freely in the browser. New fields or custom schema changes may require an approved onboarding or configuration change.

2. Define business fields deliberately

Use stable business names and clear data types. Avoid creating multiple fields for the same concept simply because labels differ across suppliers.

  • Prefer one target field with approved aliases for source labels.
  • Keep dates, numbers and text fields semantically consistent.
  • Separate value and unit when the downstream system needs them separately.
  • Keep currency separate from amount where practical.

3. Use repeating structures for line items and tables

Variable-length tables should be modelled as repeating rows or arrays rather than fixed fields such as Item 1, Item 2 and Item 3. This reduces the chance that a longer document silently loses rows.

Typical row fields may include description, quantity, unit price, tax, amount, specification, result, unit of measure or other workflow-specific values.

4. Use descriptions, examples and extraction guidance carefully

For custom onboarding, field descriptions and examples should explain what the business value means, not only where it appeared in one sample. Instructions should help distinguish similar concepts and identify accepted source variations.

Do not encode one supplier layout as if it were a universal document rule.

5. Validate the target output template

MicroExtract validates the selected template files before they are used for customer output. The exact validation depends on the output type and approved profile.

A successful template validation confirms the template shape. It does not prove that every future source document will extract correctly.

6. Treat template revisions as controlled versions

Uploaded template revisions are versioned and linked to a tenant and mapping profile. When a template changes, test representative samples again before relying on the new revision.

Do not overwrite a known-good operational template without preserving the ability to identify which revision produced a historical output.

Need help with a real document workflow?

Use representative samples, define the required output and validation rules, then test before production use.

Field Design and Template Acceptance

A reliable mapping describes what each business value means. It should handle supplier variation without treating one sample layout as a universal rule.

Use a field contract

For each required field, record its business definition, type, accepted aliases and missing-value behaviour. Keep amount separate from currency and value separate from unit when the receiving system needs both.

Repeated rows need an array or table model, not Item 1 through Item 10 hardcoded as separate concepts.

Contract itemIllustrative example
Business meaningInvoice issue date, not payment due date
TypeDate with an agreed output representation
AmbiguityFlag an unclear month/day interpretation
Missing valueLeave missing and create a review reason
Table relationshipKeep each quantity with its source line

Validate shape and extraction separately

Template validation checks the intended output structure. It cannot prove that every future document will be interpreted correctly. Test the approved profile with normal and difficult examples after a change.

Keep the template revision associated with the processing job. A new field or schema redesign may require approved onboarding or configuration rather than unrestricted browser editing.

A compact regression set

Include a short document, a long repeating table, a missing mandatory field, an unfamiliar label and a changed date or number format.

Compare the result against reviewed expected values. An improvement on one supplier must not be accepted without checking the existing examples.