Contact Center SoftwareSourced

Case Study: CallTrackingMetrics reaches >90% transcription accuracy with Deepgram

CallTrackingMetrics Case StudySourced & dated by Case Study Desk

Key facts · TL;DR

Company

CallTrackingMetrics

Industry

Contact Center Software

Challenge

40% of transcriptions too inaccurate for reliable analytics

Headline result

CallTrackingMetrics replaced unreliable speech recognition with Deepgram, reaching a greater-than-90% accuracy rate after 40% of prior transcriptions were too inaccurate to use.

Key results

>90%

Transcription accuracy

Deepgram's stated accuracy rate

40%

Transcriptions previously too inaccurate

before Deepgram

The challenge

CallTrackingMetrics needed a highly accurate, cost-effective speech recognition solution for both its call-tracking analytics and contact-center products. Roughly 40% of its call transcriptions were too inaccurate to provide reliable analytics and insight.

The solution

Serving 100,000 users in over 90 countries, CallTrackingMetrics deployed Deepgram's AI speech platform within its AWS VPC to power transcription across both its call-tracking and contact-center offerings.

“Deepgram is by far the most accurate speech vendor we evaluated and we've seen a tremendous improvement since deploying Deepgram, both on the call tracking analytics and contact center side.”
TF
Todd Fisher
Founder, CallTrackingMetrics

The results, in context

With Deepgram's greater-than-90% accuracy rate, CallTrackingMetrics significantly improved the quality of its customer conversation analytics. It continues to run Deepgram in its AWS VPC, citing improved accuracy, speed, lower latency, and stronger security.

Products used

Deepgram Deepgram AI Speech PlatformDeepgram Speech-to-Text API