Skip to content
  • There are no suggestions because the search field is empty.

Ella Proof of Concept: Your Data Checklist

Running an Ella proof of concept (POC) shows you how Ella reads your real purchase orders and turns them into clean, priced, ERP-ready orders. To get started, you share a few file exports and some sample POs, no ERP access or credentials needed. Use the checklist below to gather everything before your POC begins.


Table of Contents


Before You Begin

  • Sign the mutual NDA before sharing any data
  • Gather a set of sample purchase orders as listed in
  • Get the secure upload link or shared folder from your WizCommerce contact

A. Must-Have vs Good-to-Have Data 

To run this POC, we would require two sets of data: must-have and good-to-have. The must-have sets are needed to run the POC. The good-to-have sets are optional, but they give Ella the same context your team uses every day, so the results look closer to your real orders. A simple rule: if it is in your system, include it. For a data or IT analyst, this is usually a 60–90 minute job.

 
Tier Dataset What it Contains If Missing
Must-have Customer master Customer name, Customer ID, bill-to address, ship-to address(es) Ella reads customer and address details straight from your POs
Must-have Product master SKU IDs, SKU names, descriptions, base price Ella captures every line item from your POs
Must-have Sample POs Real purchase orders (see Section C) These are what Ella reads, so a few are essential
Good to have Customer-specific pricing Price levels/tiers per SKU, and/or contract pricing per customer Runs with list prices only, so PO-price mismatches may surface
Good to have Part-number mapping Your customer's part number ↔ your SKU, per customer Without a mapped SKU, Ella matches on description or SKU text instead. 

B. Data Dictionary

Field-level detail for each dataset. "Must-have" follows the tiers above.

» Customer master (one row per customer / branch)
Field Needed Notes
Customer ID Must-have Unique identifier in your system
Customer name Must-have Legal or trading name
Bill-to address Must-have Billing address on file; multiple is fine
Ship-to address(es) Must-have All ship-to locations; not a blocker if you keep only one
Price level / tier code Good to have The field that decides which price list the customer buys at
Contract / buying-group code Good to have Links the customer to contract pricing; often a different field from Customer ID
» Product master (one row per SKU)
Field Needed Notes
SKU ID Must-have Your item code
Description Must-have Standard item description
Price Must-have List/base price
Unit of measure Good to have Each / box / case; needed to compare PO prices correctly
Pack / carton quantity Good to have Units per pack; avoids false price mismatches on case or bag orders
» Pricing by price level
Field Needed Notes
SKU ID Good to have Same ID as the product master
Price level code Good to have Matches the customer's price-level field
Price Good to have Price of that SKU at that level
» Contract pricing (customer-specific prices)
Field Needed Notes
Customer / contract-group key Good to have Whichever field ties a customer to their contract prices
SKU ID Good to have The item under contract
Contract price Good to have Overrides level and list price (contract > price level > list)
Start / expiry dates Good to have So only active contracts are applied
» Part-number mapping (your customer's SKU ↔ your SKU)
Field Needed Notes
Customer ID / key Good to have Which customer the mapping belongs to
Customer part number Good to have The code as it appears on that customer's POs
Your SKU ID Good to have The internal SKU it maps to

Note: Export straight from your ERP, there's no need to clean or deduplicate first. Ella works with the data as it is, and handling real, imperfect data is part of what the POC demonstrates.


C. Choose Your Sample Purchase Orders

The POs are the core of the POC, Ella runs your real POs and shows the resulting orders, so which ones you pick matters more than how many.

What to send:

  • Start with the hard ones. The POs your team spends the most time keying in today: long multi-line orders, part-number-based orders, messy layouts.
  • A mix across templates and customers. POs from different end-customers, each with their own layout.
  • A mix of file types. PDFs, scans, faxes, email-body orders, Excel sheets.
  • POs that line up with the data you share. At least some should be from customers in your customer master, with items that exist in your product data.

There's no hard minimum, but 20–30 POs covering this variety is a good working target.

Note: For best results, include a few POs whose customers and items appear in your master files, so Ella can match every line to your data.


D. Send Your Data 

  1. Confirm the mutual NDA is signed before sending anything.
  2. Export each dataset as Excel (.xlsx) or CSV, one file per dataset (customer master, product master, pricing, part-number mapping).
  3. Keep sample POs in their original format, PDF, scan, image, email export, or spreadsheet. No need to convert or re-key.
  4. Package everything into a single folder or ZIP with clearly named files.
  5. Upload through your secure link or shared folder. Don't email sensitive data as plain attachments.

Once it arrives, the team reviews it and follows up quickly if anything is needed before scheduling the demo.


E. What the POC Demonstrates

During the POC, Ella works through your real POs and shows the result for each type of data you provide. Use this table to see what to look for in the demo, based on the files you send.

Data You Provide What Ella Demonstrates
Customer master + POs Identifies the ordering customer and picks the correct bill-to / ship-to
Product master + POs Extracts line items and matches them to SKUs
Price levels / contract pricing Applies the right customer-specific price (contract > price level > list) and flags PO-price mismatches
Part-number mapping Resolves the customer's part numbers to internal SKUs automatically
Customer descriptions / colors Matches lines using the customer's own naming

FAQs

1. Can we run a POC with only customer names, products, and a few POs?

A. Yes, that's the minimum set. Missing good-to-have items just narrows what can be shown. With no contract pricing, for example, the POC runs on list prices only.

2. Do you need access to our ERP?

A. No. Excel/CSV exports and PO files are all that's needed. ERP access only becomes relevant at implementation, after a successful POC.

3. Our data is messy or has duplicates, should we clean it first?

A. No, export it as-is. Handling real, imperfect data is part of what the POC demonstrates, and anything that genuinely blocks the run gets flagged during review.