Region OCR accuracy approximately @ 60%

Dear UiPath

we have developed a bot to read specific region in a standard formatted invoice in one of our AP areas

the vendor invoice has a white background,

After processing around 20 document numbers, we encountered that the scraped date was not accurate,
once I changed the scale to the next number , it starts working again.

the next issue we faced was scraping the amount, the bot scale had to be change again in order to scraped the correct amount.

Hence to work around the scaling, we implemented a do while loop which keep lopping until the scraped data matches the excel supporting file

for this instance, luckily we have an digital file to rely on to verify the scraped data was accurate, however, we have a concerns for the invoices that doesn’t have digital files as a support

so far the accuracy for the both engines (Google, and Microsoft) is 60 percent,

@ Support team please help me how to ensure that the accuracy increase

we would prefer to have 100% accuracy. is there something wrong we are doing.
Is there other solution to help to scrape the data accurately?