Human-verified 100% detection accuracy allows contact centers to confidently utilize Network Level Binary Matching (NLBM) technology without worrying about generating Silent Calls and breaching regulations.
Traditional AMD relies on listening to the line, and using analog comparisons like signal to noise ratio, and ‘gaps’ in speech (hello, hello) to determine whether the call is picked up by a human or machine. This can never be 100% accurate, and so sometimes silent calls are created when the system mistakes a human for an answer machine, leading to a potential regulatory breach.
ContactNow’s NLBM differs from traditional AMD technology in one key way. It does not use any analog measurements to determine whether or not a call is connected to an Answer-phone. Instead, it compares the digital data stream coming from the telephone network with pre-stored patterns, and only determines if the call is an Answer Machine when there is an exact match.
Each 1,280 byte signature file is sufficiently accurate to ensure there is no measurable possibility of creating a false positive.
To demonstrate this, ContactNow tested the NLBM system in 2010 by passing over 50,000 call recordings, which yielded 5,000 exact matches. These matches were then listened to manually to check if they were valid. No false positives were found.
The system is now maintained on an ongoing basis by automatically generating new signature patterns and verifying validity. New signature patterns are created due to people changing recorded messages and network equipment being replaced and added into the telephone network, changing digital patterns.
The detection rate continues to improve as signatures are being constantly generated by the system, and more and more people migrate from analog phone lines to digital phone lines. The detection rate is now approaching that of traditional analog AMD systems which can claim 80%+ detection rates.
Patented Solution (GB2487734)
Verified Zero 'False Positive' rate
Detection accuracy of 100%
Detection rate of 60-80%
Cost per detection is fractions of a penny