This data is not monolithic and most real-world data easily consists of over 100 columns. Maintaining the quality can be challenging, given the variety of sources feeding into just a single feed. Even the most simple quality checks can snowball into a daunting task. Everything from tickers, sedols, cusips, products, sub-products, issuers, and issuing countries can further complicate the problem. Identifying anomaly values earlier in the data ingestion process can significantly reduce downstream complexity. Furthermore, finding improbable patterns before they're used for making decisions can save costly remediation efforts.