What would be an acceptable % success rate (scanned and no corrections need to be made) of preprinted corporate type checks run through an RDC scanner.  We have been running test batches and have had up 37.5% and 45% failure to read check dollar amounts.  We are using our bank's internet RDC solution, since our audit department vetoed the software version.  We had a much better rate with the software solution, utilizing the same scanner, Digital TS220E.  In our most recent test, we had an 85% success rate.

An 85% success rate is very good. Based upon personal experience, and discussions with dozens of solution providers and end-users, a general range of success for the types of checks you outline is generally around 75-80%. We've even heard of instances of 90%+.

The success rate depends primarily upon the quality of 1) the source document, 2) the CAR / LAR / OCR / ICR software being used, 3) the actual scanner, and 4) the degree to which the equipment is (or is not) maintained and cleaned.

Are you using the same scanner with the browser-based version?

We are using the same scanner, the Digital TS 220E.  It is new, having only been used to run a few test batches, possibly 200 checks in total.  The biggest problem checks seemed to be the ones with the multiple 'Void' (covering the whole face of the checks) watermarks/security features. 

If 85% is good for printed corporate checks, what can we expect from hand written consumer checks?

If you're using the same scanner, but are having vastly different results on CAR / LAR reco rates on a thin client vs. thick client operational model, the most likely cause is the Reco software being used in the thin environment is not as complicated / sophisticated as the version you had locally installed. This makes logical sense as the more horsepower the reco software has / needs equates to the need for more horsepower centrally, more data from the scanner, and more time to transmit and process the data.

In conversations we've had with hosted solutions providers, there is a delicate balance between processing time and robustness of the CAR/LAR and IQA/IQU software. Most solution providers have a goal of performing the reco process in the background without any impact to the end user. Based upon the experiences you've outlined above, maybe this isn't the right goal?

Similar to all the various versions of Microsoft Office and other competing software packages, Reco Software comes in many degrees of competency, from basic to editions lebelled as "platinum", and results are further complicated by source documents, the scanner being used, the condition of the scanner and more.

Most of the reco rates I am familiar with are part of a thick-client installation where the reco software may very well be better than the capabilities currently available in a thin-client environment. In a thick-client environment, we have heard from the industry to expect 80% success on printed checks and 60% success would be good on hand-written checks, on average. Please keep in mind, though, any success rate results in new processing efficiencies for the end user.

Hopefully, a couple of solution providers might be able to post some additional insights on this issue.

I have been doing some research on the various blogs and "informational" sites on the Patent Reform Act of 2007.  It is considered by many that this will not pass the votes needed to make into law.  In fact, people are saying that this is going to loose alot of steam.  Sure, it made it out of committee, but that does not necessary make it into law.  People are trying to get this voted on before the August break if it has any chance of maintaining the needed steam. 

My opinion, for what it is worth, is that this bill will not make it into law.  What do other people think?  Remember, this has to be a joint bill which both the Senate and House pass in the same form.  Also, remember that Sen. Sessions amendment dealing with check imaging is only on one of the bills.  So, it could be axed as part of the process. 

CopaKid