Ventana Research, in their article XBRL Filing Errors Point to Need for Automation through Technology, points out the need for automation in the process used to create SEC XBRL filings. I agree with Ventana and would go on to say that if one does not have automation then one needs hefty validation in order to detect and correct errors caused by using humans in the process.
The Ventana Research article points to an SEC Staff Observationwhich communicates the results of a review the Commission's Division of Risk, Strategy and Financial Innovation performed. In that review, the SEC staff points out common classes of errors found in SEC XBRL filings.
When one thinks about it, the ultimate test of an SEC XBRL filing is using that filing within some analysis process. Simply comparing two filings from different SEC filers will do the trick.
I will point out two resources which can help you figure out the types of errors you will have to deal with. The first is a model SEC XBRL filing. In particular, this PDFwhich explains the details of the simple filing, helps you see the information integrity issues you need to be sure to watch over. See the notes in that PDF for information.
The second resource is what I call the comparison example. This has three fairly basic sample financial statements. The statements have both positive and negative examples. One thing this shows is the very negative impact of filers creating their own concepts where a concept in the GAAP taxonomy should exist.
The issues SEC filers are bumping into are not all of their own making. Many of the issues come from issues with the US GAAP Taxonomy itself and lack of clarity in the SEC XBRL filing rules.
Like I said above, trying to analyze company information in those SEC XBRL filings is the ultimate test. This is not to say that everything is, or even should be, comparable. This will start the conversation where those in the financial reporting supply chain will discuss these sorts of issues.