Although FormXtract is a very straight-forward API, there are some best practices to consider following when using this Quik! product.
Avoid Sending Duplicate Forms Together
When FormXtract reads over a document and extracts data, the data is aligned to a Quik! FormID (a unique identifier for the form in the Quik! library), a PageID (the unique ID for the page within the entire library) and a Field ID (a unique field ID for that field within the entire library). When FormXtract encounters a duplicate form in the same submitted document, the FormID, PageID and FieldID will be duplicated in the output.
In other words, the resulting data will have two instances of the same field with two different values but with no way to identify which page number within the submitted document the data belongs to.
For Example: Let’s say two accounts are opened together in a household where one account has two transfer forms and one account only has one transfer form, plus two new account forms. If these documents are all submitted together FormXtract will have 2 copies of all the new account fields and 3 copies of all the transfer fields, and it will not be obvious as to which transfer forms go with which accounts.
RECOMMENDATION: Quik! recommends sending each form as a separate API submission to FormXtract.
Do Not Batch or Delay
Although it is common to put work items into batches or delay working on them until a specific time of day when a person can address those items, FormXtract does not require batching or delaying. In fact, doing so will just cause delays for when your documents are done. Plus, FormXtract is running all the time and putting items into a batch to send all at once to Quik! will cause the FormXtract API to experience a spike and high volume, which is less than optimal for both yours and our system.
RECOMMENDATION: Quik! recommends sending documents to FormXtract as they enter your workflow and without delay.
Include Your Document ID
Map To The Quik! Field Definition, Not The Form
Accuracy in FormXtract
While FormXtract enables 99.9% accuracy, no system can be 100% perfect. There can always be some amount of rejected forms and/or fields if the submitted document is of poor quality.
Rejections happen primarily due to illegible handwriting that no human can read, or when the form is not filled out well (e.g. data is in the wrong place, data is missing, pages are missing, etc.).
The output form FormXtract will provide information that will aid in determining if a form or a specific field was unverified and should be handled manually.
How To Achieve More Accurate Results
The best way to ensure 99.9%+ accuracy is to have forms filled out with a computer so the data is typed and not handwritten. In addition to doing as much of the form filling work electronically, here are some best practices:
FormXtract Best Practices
The fastest results come from electronic documents that have digital fields on them (i.e. a PDF with editable fields).
The second fastest results come from electronic documents that were completed entirely online, in a digital format.
Scans, faxes and lower-quality versions of the original document may slow down processing.
Separate your forms into individual documents to be sent to FormXtract, one at a time. Although FormXtract can handle a single file (TIF or PDF) that includes multiple documents, if a document contains more than one copy of the same form, the resulting data can be difficult to match up to the documents.
FormXtract can read handwriting, but handwriting so bad that no human can read it will result in errors or bad results in the extracted data.
FormXtract will attempt to fix poorly scanned documents that are skewed or rotated but such documents may lead to more errors or bad results in the extracted data.