Monitoring extractions
In the Sensible dashboard, you can view the following metrics about your extractions in real time:
Extraction coverage
Extraction coverage measures how fully an extraction captures your target data from the document. For example, a coverage score of 70% for an extraction with no validation errors means that 30% of output fields are null. A low percentage can indicate a poor-quality extraction. Or, it can indicate that your documents contain sparse data. For example, if you define many target fields to extract from a supplemental insurance form, but applicants leave most of the questions blank, then the form’s expected average extraction coverage is much less than 100%.
Since coverage success criteria can vary by document type, you can set your own ranges for each document type in the Sensible app. To set ranges for coverage for past extractions, click Dashboard and scroll to the Extraction coverage section. In the section, you can filter by:
- extraction coverage range
- extraction ID
- date range
- document type
- extraction batch
In the preceding screenshot, 0 out of 16 extractions that used the wells_fargo_savings configuration in the bank_statements document type in the past 7 days scored in the coverage range of 60% -100%.
To determine your own coverage criteria, examine your past extractions. For example, if home inspectors typically report about 60 out of your 100 target data points, you can set a range of 60%-100% for your home_inspection_report document type to assess extraction success. In contrast, if you find that drivers licenses typically contain all your target data points, you can set a range of 95%-100% or even 100%-100% to assess extraction success for the drivers_license document type.
To view an individual extraction’s coverage score, click Dashboard and scroll to the Recent section:
In the preceding screenshot, the Sept 25, 2023 7:30 PM extraction has a coverage score of 61.1%.
For more information about how Sensible calculates coverage scores, see Coverage.
Most-used configurations
To view which configurations were used for past extractions, click Dashboard and scroll to the Most used section:
In the preceding screenshot, the 1040_2018 configuration was used for 54.7% of extractions in the last 30 days in the tax_form document type. If you adjust the filter to include both bank statements and tax forms, you see that it was used for 33% of all extractions across the selected document types:
Documents extracted
To view the number of past extractions you’ve run by document type, click Dashboard and scroll to the Document extracted section. You can filter by date range, document types, and extraction coverage: