Most Optical Character Recognition (OCR) tools do a great job of translating PDF documents to American Standard Code for Information Interchange (ASCII) text with a typical accuracy of 95%-98%, but they can struggle to accurately convert tables back into rows and columns, which is essential when conducting document analytics or trying to apply AI to a workflow with unstructured data in PDFs. This is especially true when there is limited industry standardization among common types of business documents.

Baker Tilly’s Automated Document Analytics helps identify the relevant details from tables of data, regardless of their location on the document. Data points from common documents such as professional service invoices and bank statements are easily identified and loaded into databases, based on pre-built data models. This results in a significant reduction in implementation time, and little to no need for custom coding.

Use cases include:

• Invoices for professional services

• Profit and loss statements

• Corporate telecommunication invoices

• Pick lists, packing slips, and shipping bills

• Real estate contracts

• Re-insurance contracts

• Benefit contribution statements

• Fraud detection/forensic accounting

Example: A company is making an insurance claim with the aim of identifying potentially suspicious transactions and is looking for these specific metrics:

• Round dollar amounts ending in ‘000 and ‘0000

• Transactions over a designated amount

• Transactions in and out of the same account in the same day or within a few days

• Transactions with other companies in neighboring warehouses

• Cash and check transactions

• Transactions on public holidays and weekends

• Offshore payments

• Payments to individuals rather than companies