Today, what portion of the checks or other review steps performed to make sure the quality of a financial report is what it needs to be are performed using automated processes? The answer is ZERO. Well...I guess you could count using spell check in Word an automated task.
Why is that? Why can't certain quality control tasks and other review steps be performed using automated machine-based processes.
The reason all review tasks have to be performed using human-based processes is because financial reports of the past have been unstructured documents which are impossible for a machine to read reliably. Well sure, a word processor can "read" a financial report; but word processors know nothing about financial reports. Word processors understand documents and they can tell you if you messed up the document structure (i.e. word, paragraph, section, etc.). If this is unclear to you, consider watching the video How XBRL Works.
XBRL-based financial reports created an opportunity to change the review process used to make sure financial reports are created correctly. XBRL alone cannot change this. But, if you put three things together a transformation can occur. To achieve this change you need three things:
Because of XBRL-based financial reports, machine-readable rules, and software that can effectively understands the rules; it is no longer the case that ZERO percent of the process of reviewing a financial report to make sure it has been created correctly.
Today, it is possible to automate some percentage of those quality checks and other review tasks. What percent? While the actual percentage of what can be reviewed using automated machine-based processes is unclear; that percentage will be significant. The percentage will be determined by the volume and quality of the machine-readable rules and the problem solving logic available from the software.
When will this software be available? Well, the software is available today. Here are two software applications that perform automated review tasks:
You can fiddle with the reporting checklist here if you are interested; experience the details for yourself. This ZIP archive contains an Excel based application that lets you extract and see the machine-readable rules used by both of those software applications. There are THREE Excel spreadsheets: one for US GAAP, another for IFRS, and another for what I call the XASB reporting scheme which is an imaginary scheme I created to figure some things out. This document explains the details. This document explains the techniques used.
Currently, XBRL Cloud and Pesseract are both flawed in my view. The flaw is that they are both batch processes run AFTER a report is created. Why is that a flaw? It is a flaw because the software does not help you actually create the report.
What I am doing is creating software that is used DURING the process of creating a financial report. Basically, what I refer to as an expert system for creating financial reports. The idea of expert systems is not new. Expert systems is one of the most successful areas of artificial intelligence. When I talk about human and machine collaboration or machines augmenting humans; what I am talking about is an expert system for creating financial reports.
Ten years ago I could not explain the difference between "syntax" and "semantics". Literally. When I created that video, How XBRL Works, on May 1, 2008; I was still struggling with understanding the difference between those two terms.
A lot has changed since then. I have accumulated a lot of information and step-by-step figured out how to create the expert system I believed could be created. I figured it out with the help of a bunch of other people. There are still a few details but the problem of how to do this has been solved. That is the purpose of Pesseract, to figure this out.
Financial reporting...get ready to be disrupted.