Discover how machine learning is used to extract 100% of revenue statement data — and take a sneak peek at the technology that makes it possible.
Oil and gas revenue statements are how royalty and override owners and non-op working interest partners see the details of their revenue from a well or set of wells.
The Problem: An interest owner will receive anywhere from one to twelve statements per year per operator they are working with.
If an owner only owns one or two wells, all the data might fit on one page. But usually these statements are dozens (or even over a hundred) pages long, detailing all the gross and net proceeds and deductions by property and by month.
Getting this information into an accounting system would allow companies to better track their income and deductions (which helps when filing taxes) and look for lease violations (such as Cost Free Royalty). Even having this data in a spreadsheet is better than just PDFs.
Revenue Statement Processing - Part One
The Approach: The first part of this webinar is about how RevCHX.com is approaching Revenue Statements differently. We aren’t outsourcing hand-keying of the data overseas, we aren’t limited by what oil companies want to give to us.
We use Machine Learning to extract data from revenue statements supplied by the owners themselves and return that data to the owners as OCRed PDFs, spreadsheets, and CDex files. This allows us to capture 100% of the details of 100% of the revenue. We are also able to go back and process historical revenue records.
Revenue Statement Processing - Part Two
The Solution: The second part of the video shows how Grooper Models that Intellese created for RevCHX.com performs these extractions. Revenue checks can look simple, but from a technical perspective they are full of challenges.
Every oilfield data management system has it’s own basic layout, which can have variations, and every state has different tax codes/labels. This is in addition to the “normal” challenges of working with scanned documents (mainly OCR quality).
