We had quite a few issues with DoxCycle’s extraction accuracy last year.
After watching the recent webinar, I’m still not clear on what has changed for the accuracy rate of data extraction since last season. I know improving it was one of the goal of the Taxcycle’s team.
What is the current status on the rate of accuracy of the data extraction?
How does it compare to last year?
What is the goal for the upcoming tax season?
Do you have any new recommendations or improvements that could help improve accuracy for this tax season (knowing we don’t have that much control on the quality of the slips sent by the clients)?
The work to improve the automated classification of documents on import is nearly completed and is expected to ship in the new year ahead of tax season. This will include some additional improvements to the embedded Optical Character Recognition (OCR).
The work to improve data extraction has, unfortunately, been delayed. Based on the work left to do, it is unlikely this improvement will be shipped in time for this upcoming tax season as planned. I expect we will be able to complete this work sometime for a late spring or early summer release 2026.
As for what we can do in the interim, I will check with our communications team to see if we can create some resources for clients to help them produce better quality scans/photos with their mobile phones, which is the method I understand more and more clients are using to share their income tax information with their accountants.
If you are interested in helping with our improvement project and are willing to provide source documents we can use to improve DoxCycle data extraction, please email info@taxcycle.com to express your interest and/or request additional details. The more documents we have to build our model(s) on, the more accurate the results will be.
That is so unfortunate, we were really counting on this improvement for this upcoming tax season. The accuracy was very poor last year, I thought it would have been just an exceptional year.
Yes, the improved quality of scans would be good. But the issue would remain as there are accuracy issues even with the original slips issued in PDF version. I doubt that improving the scan quality would make a big difference. I know in our case, most clients provide a decent quality of scans, but any improvement is welcomed.
Would you have another suggestion to improve the accuracy of the data extraction? Maybe by connecting an additional software?
What percentage of accuracy should we expect this year?