CUSTOMER DATA PROCESSING POWERED BY MICROTASKING AND CROWDSOURCING
No matter how much digitization or automation is going on, paper is still used in businesses all over the world. Large companies and organization still struggle with unorganized paper and digital documents clogging their workflows. Time and money are constantly spent on integrating automated solutions which, in the end, still require internal employees to participate in the processing, lowering overall work efficiency and multiplying processing costs. In the end, companies need to compromise and give up on cost effectiveness, speed, accuracy or data confidentiality. Furthermore, most of these solutions are inflexible and require further investments or even full replacement anytime something changes in the usual workflow.
Now meet MPS IntelliVector – a solution that instead of trying to improve one aspect of the processing like accuracy or speed, completely transforms the traditional approach to extracting and processing customer data from paper or digital sources.
MPS IntelliVector combines automated processing (OCR and ICR) with microtasking and enterprise crowdsourcing to yield benefits of both – automated and human processing. On one hand – maximizing the level of automation, including straight-through-processing, reduces the need for manual processing (both data entry and data validation) to the lowest possible, yet still guarantee exceptionally high data accuracy. On the other hand microtasking optimizes manual processing – when data validation can’t be avoided – and allows to securely outsource or crowdsource the occasional manual data validation significantly reducing the processing costs, yet still guarantee 100% data confidentiality.
SOLUTIONS BASED ON MPS INTELLIVECTOR
How it works?
MPS IntelliVector breaks down the incoming documents or forms into small individual microtasks, like part of a name or an address, or a few digits of a number. These microtasks are automatically processed by a OCR and ICR recognition engines and at the same time, sent out to users for manual data entry.
These users receive microtasks containing information out of its original context, meaningless on its own, this way the confidentiality of the initial data is preserved. This means companies can free their own, expensive internal resources and outsource the data entry process to any part of the world or utilize even cheaper crowdsourced workforce, without risking the confidentiality of the processed sensitive data.
The results of the automated recognition and the manual data entry are automatically crosschecked for result validation. In case of mismatching results the microtask is sent to another data entry user and his result is also matched against the first two. Only if none of the results match the microtask is sent for quality control.
This approach not only guarantees high data accuracy, but also reduces the amount of the usual double data entry to approximately one fifth of all cases, but also minimizes quality control to 1-2% of cases – a fraction of the usual.