Your first dive

Four exercises, from the simplest table to the most complete, to see how the Extract targeted data workflow reads an Excel table and fills it from your documents. Allow about twenty minutes. You work in a ready-made workbook from the Moby library, with sample documents.

Before you start

1. Open Moby in Excel and sign in. See Install and sign in.
2. In the Workflows tab, open Workpaper library (or type /workpaper in the chat). On the Moby tab, choose Getting started with the extraction workflow and add its tabs to your workbook.
3. Download the sample documents, one zip per exercise, and unzip them on your computer.

Exercise	What it adds	Sample documents
1. Invoices	A row of titles is enough	exercise-1-invoices.zip, 5 PDFs
2. Contracts and amendments	A description row, and one row per file	exercise-2-contracts.zip, 4 PDFs
3. Matching	Your rows first: Moby finds the document of each row	exercise-3-matching.zip, 6 PDFs
4. Invoices and delivery notes	Two document types, and two documents for one row	exercise-4-invoices-and-delivery-notes.zip, 9 PDFs

The workbook opens on a Read me tab and ends with an Expected results tab: compare your result with it after each exercise.

The same steps in every exercise

1. In the Workflows tab, click Extract targeted data.
2. Step 1/2, the documents. Click Upload (or Change), drop the PDFs of the exercise, then Submit. Keep only the documents of the exercise ticked, click Apply, then Next.
3. Step 2/2, the table. Select in the sheet the range given by the exercise, then click Launch.
4. Moby writes the values into the table and adds a Document column on the right: the file each row comes from.
5. Click a value: the Preview tab shows the page of the document with the value highlighted.

Moby picks its mode from the table you select. There is no setting.

	Extraction (exercises 1 and 2)	Matching (exercises 3 and 4)
For	Building a table from a batch of documents	Testing rows you already have: finding the document of each row
Your table	Column titles, nothing underneath	Your rows: filled columns (the keys) and empty columns to fill
You select	The header rows only	The whole table: header rows and data rows
Moby writes	One new row per document	The empty cells of each row, never a cell that is already filled

Exercise 1: extract values from a batch of invoices

The simplest case: you have no list yet, only documents.

Tab	1. Invoices
Table	Six titles: Invoice no., Invoice date, Customer, Amount excl. VAT (€), VAT (€), Amount incl. VAT (€)
Select	The title row only, C21:H21
Result	One row per invoice under the titles, five rows

Key points:

• Precise titles are enough: Moby creates one row per document.
• One title, one value. "Amount excl. VAT (€)" reads one way only. "Amount" alone would read three ways.
• Keep the rows under the titles free: Moby writes its rows there without inserting any.

Exercise 2: contracts and amendments

The same steps, with a description row under the titles. It tells Moby what to read, in which format, and what to answer when the document says nothing.

Tab	2. Contracts and amendments
Table	Ten titles, each with a description: document type, reference, borrower, dates, amount, rate, prepayment fee, financial covenant, security
Select	Both header rows, titles and descriptions, C22:L23
Result	One row per file: two agreements, two amendments

Key points:

• The description is part of the question. It sets the format (DD/MM/YYYY, number only) and the answer to give when the document is silent ("None", "Unchanged").
• One file, one row. Moby reads each PDF separately: the amendment row holds only what the amendment itself says.
• To read an agreement and its amendment as one document, merge them first with Organize PDFs.
• Select every header row. A description row left out of the selection would be overwritten by the first result row.

Exercise 3: match ledger entries with their invoices

This time you start from a list: six ledger entries. Moby finds the invoice of each row and fills the empty columns next to it.

Tab	3. Matching
Table	A band row naming the blocks, the titles, the descriptions, then six rows. The Accounting block is filled, the Invoice block is empty, the Check column is a formula
Select	The whole table, C24:K32
Result	The invoice columns filled on every row, the difference at 0.00, and one invoice listed on the Unmatched documents sheet

Key points:

• A filled column is a key: Moby uses it to find the document of the row and never writes it.
• An empty column with a title is filled from the document. Each key has an empty twin on the invoice side, so you compare the ledger and the document side by side.
• One document can serve several rows: invoice INV-009, booked on two rows, is matched to both.
• A document with no row never enters the table: it is listed on the Unmatched documents sheet.
• The workflow extracts, Excel computes: the difference is a formula.

Exercise 4: two document types

One block of columns per document type. And an order delivered in two shipments: two delivery notes for the same row.

Tab	4. Invoices and deliveries
Table	Four blocks: Accounting (filled), Invoice, Delivery note, Check
Select	The whole table, C23:O29
Result	Each row filled from its invoice and its delivery note. The Pacific Marine Works row shows two delivery notes, two dates and three files in the Document column

Key points:

• One block of columns per document type, named in the band: Moby knows which document to read for each column.
• The key must be printed on the document. The delivery note does not quote the invoice, but it quotes the order number: that is what links the delivery notes to their row.
• Several documents can match one row: texts are joined ("BL-011; BL-012"), numbers are added up.
• Click the delivery note number of that row: the Preview shows Source 1/2. Check every source, one too many changes the value.

What next

You want to	Go to
Correct a value or change its source	Review and correct
List every value with its document	Document the work
Build your own table	Build a good table
Start from a sheet built for an audit test	Workpaper library