Also offered: data audits
Checking extracted credit data against the source
Spreads, rent rolls, covenant terms and loan tapes are increasingly extracted by software. We tie that output back to the source documents in code and tell you exactly where it differs.
Two ways to use it
Audit a sample of your own data
Send us a sample of extracted output with the documents it came from: financial spreads, rent rolls, covenant terms, or a loan tape with the matching servicer report. We recompute every field, reconcile the totals and send an exception report listing each difference, its size and the line it comes from. Fixed fee, quoted per scope.
Test an extraction tool
If you build or sell an extraction tool, we run it through our data audit suite three times and send a private scorecard.
What the suite covers
The data audit suite is small and new: 9 tasks, of which 3 are clean cases where the extraction is correct and the right answer is to say so.
- Financial spreads: a figure taken from the wrong fiscal year, or a statement in thousands spread as whole dollars.
- Rent rolls: a unit dropped at a page break, and a rent misread by a few dollars.
- Covenant terms: levels extracted from the original agreement when an amendment has since changed them, and an amendment that has been signed but is not yet in effect.
- Loan tapes: a tape that does not reconcile to the servicer report because of a duplicated loan, and delinquency buckets that do not match days past due.
Two examples (fictional). A tape of ten loans has eleven rows, because one loan appears twice, and it totals $518,250 more than the servicer report.
In another tape, a loan 97 days past due is labelled 60–89, so $196,400 of seriously delinquent balance is missing from the 90+ bucket.
Every answer key is computed in code and checked by a second, independently written program before a task goes into the suite.
Our tasks are synthetic. A sample audit is limited to the documents you send and describes that sample only; it is not an audit opinion or a warranty.
Contact
Talk to us about a data audit
Tell us what kind of data you extract and roughly how much. We will reply with a proposed scope.