The confidence score represents the degree of certainty for the accuracy of the credit insights outputs. It is a float bounded between 0 and 1.

How is the Confidence Score calculated?

There are a number of factors which affect the quality of the credit analysis outputs:

• The quality of the submitted documents. If the documents have a low image quality, it is more difficult to accurately parse information from them.
• The familiarity of the submitted documents. Our parsing models perform better on more common document formats.
• The quality of the transaction descriptions. Our transaction categorisation engine performs best on transactions with obvious descriptions, as opposed to cryptic ones.

The above features are combined and scaled to create the Credit Analysis Confidence Score.

How should the Confidence Score be used?

In order to get high quality outputs, we recommend gating on the Confidence Score. That is, Inscribe’s Credit Insights should be consumed where the Confidence Score is above some threshold. A confidence score of 0.5 or above is considered high, while a confidence score of above 0.85 is very high. After doing a Credit Insights demo with Inscribe, our engineers can recommend a confidence score for your appetite for automation rate vs. accuracy.