> ## Documentation Index
> Fetch the complete documentation index at: https://knowledge.goautonomous.io/llms.txt
> Use this file to discover all available pages before exploring further.

# Reference data

> Upload and manage reference data sets that enhance extraction accuracy and product matching.

## Overview

Reference data is supplementary information that Go Autonomous uses during processing to improve extraction accuracy and automate data lookups. It covers any structured dataset your organisation needs — delivery addresses, vendor details (name and inbox address for vendor quotations), customer lists, account mappings, pricing tables, carrier codes, and more.

## Reference data and areas

Reference data can be uploaded per [area](/knowledge-base/administration/areas), allowing different teams or regions to use different data sets. When Go Autonomous processes a request, it uses the reference data assigned to the request's area. If no reference data exists for that area, it falls back to the **Default** area's data automatically.

### Default area requirement

Every reference data grouping (e.g., "Delivery Addresses") **must have a Default area file**. This file acts as the fallback for any area that doesn't have its own upload. If there is no Default file for a grouping, requests processed against areas without their own data will have nothing to fall back on.

### Column structure must match across areas

All files uploaded for the same reference data grouping — across all areas — must have **identical column structures**. The column names, order, and count must be a 1:1 match between the Default file and any area-specific file. If they differ, the functionality will break and lookups may fail or return incorrect results.

<Warning>
  Before uploading an area-specific file, always verify that its columns exactly match the Default file for that grouping. Mismatched column structures will cause lookup failures.
</Warning>

## The Reference Data page

Navigate to **Administration > Company Settings > Reference Data** to manage your reference data sets. The page shows a grouped list:

* Each **reference data type** (e.g., "Delivery Addresses" or "Carrier Codes") is an expandable section.
* Inside each section, you see one row per **area** where that data type is configured.
* Use **Expand All** / **Collapse All** to manage the view.

## Upload reference data

<Steps>
  <Step title="Click Upload Reference">
    Click the **Upload Reference** button in the top toolbar.
  </Step>

  <Step title="Configure the upload">
    Set the upload parameters:

    * **File** — select a CSV file containing your reference data.
    * **Encoding** — choose the character encoding (e.g., UTF-8).
    * **Delimiter** — select the column separator (comma, semicolon, tab, etc.).
    * **Area** — choose which area this reference data applies to.
    * **Name** — give the reference data type a descriptive name.
  </Step>

  <Step title="Review and upload">
    Review the preview of your data to ensure columns are parsed correctly, then complete the upload.
  </Step>
</Steps>

## Review and edit reference data

Click any reference data row to open a review drawer. The drawer shows a paginated, searchable table of all entries where you can:

* **Search** for specific values across all columns.
* **Add** new rows of data manually.
* **Edit** existing values inline.
* **Delete** rows that are no longer needed.

## Download reference data

Select one or more reference data sets using the checkboxes and click **Download** to export them as CSV files. Each download is named by type and area for easy identification.

## Delete reference data

* **Single delete** — select a single reference data row and delete it.
* **Bulk delete** — select multiple items across types and areas, then click **Delete Selected**.

<Warning>
  Deleting reference data may affect extraction accuracy for requests that rely on those lookups. New requests will no longer have access to the deleted data for matching.
</Warning>

## What's next

* [Configure fields](/knowledge-base/administration/fields-manager) to control how reference data is used during extraction.