Skip to main content



Coupa Success Portal

Smart Extract


Coupa Smart Extract performs a routine where the system will attempt to extract data from contracts. The data points extracted are fixed and based on the Coupa Universal Contract Model. Currently the extraction rules cover 30 fields, but will be expanded over time. This capability is currently offered as part of an Early Access Program. More information is available at Smart Extract EAP.

Design Options

Access to this feature is controlled through membership to specified user groups. As such, the main decision is around which users are permitted to use the feature.


  1. It is not possible to run the extraction process across multiple documents in bulk. The process needs to be initiated at the document level and the results reviewed at the document level.
  2. Scanning quality of documents uploaded will heavily affect the result accuracy.
  3. Even with 100% legible scanned documents, the extraction is done programmatically and it is therefore not possible to provide 100% accurate results every time. To ensure 100% accuracy, a review process must be done against the results.
  4. The intended model future model will be to run the extract in bulk on the contracts for a limited set of fields. As the set of fields increase, the exercise can be re-run to expand the set of metadata stored against the contract, to give you ever greater insight into your contracts. With this in mind, any customers planning to use this feature need not migrate a full/rich metadata set when contracts are migrated onto the system.

Things to avoid

Smart extract is not a silver bullet. It can demo well (on contracts it has been trained against) and can lead to the assumption that it will achieve the same high accuracy levels on customer contracts. Do not make this assumption.
Although Coupa is aiming for an 80% accuracy score, the structure, scanning quality and content of the documents can have a dramatic impact on the accuracy levels.

  • Was this article helpful?