Client

The Call Tracking company helps medium sized businesses track the daily phone calls they receive, record them and provide deeper insights on phone call patterns.

Business Challenge

The Client had hundreds of customers they served and hence the call volume for each businesses added up quickly. The Client had on-prem data servers that

• Tracked all the calls driven to businesses
• Extracted the caller name, caller ID, time and duration of every call
• Converted the recording of every call from .wav to .mp3 format
• Converted the audio recording to text
• Performed sentimental analysis on call text to identify and mark the intent of call and categorize as new lead, appointment booking, existing client etc
• Emailed businesses instantly after the call on details of the call

These processing needed lots of CPU power simultaneously especially at peak times and the client was finding it difficult to scale up the data servers reliably. The disk storage was also filling up at a rapid pace. During peak times, the client found that the servers were over-utilized and pushed beyond their performance limits, resulting in lot of latency in sending the email to businesses and occasional loss of data. The client found it wasteful to overprovision to account for unpredictable peak times. Also, the disk storage holding all the audio files kept hitting the limits, having to constantly upgrade to higher capacity.

Solution

After careful evaluation, Agile GCP Labs, a Google Cloud Partner, suggested the move from on-prem to Google Cloud to help provision the VMs on demand for processing and storage. Agile GCP Labs architected and implemented the solution as below.

• Deploy Cloud Pub/Sub, a fully managed real-time messaging service to initiate the processing once the call data is available.
• Deploy a Cloud Function that receives and records the voice calls, convert the .wav to .mp3 and store them to Cloud Storage.
• Deploy a Cloud Function to process incoming call data and store the caller name, caller ID, time and duration of every call into BigQuery.
• Deploy a Cloud Function to transcribe the voice recordings, perform sentimental analysis and store them in Google BigQuery.
• Build a visually appealing dashboard using Data Studio to provide insights on total calls, minutes spoken, average call duration, geographical details, call volume patterns across days of the week and times during a day and more.

Results

The processing speed of huge call volumes improved dramatically on the migration to Google Cloud Platform.

	Before GCP migration	After GCP migration
Processing 1000 calls simultaneously with average call duration of 3 minutes	3-4 minutes	4-5 seconds
Latency in emailing businesses on call details during peak times	4-5 minutes	5-6 seconds

The client was able to scale seamlessly with increased workloads during peak times, with no observable latency in emailing businesses on the call details. The cost of storage was reduced by more than 50% resulting in significant savings. The client was able to grow their customer base with the confidence on the reliable scalable system in place built on Google Cloud Platform.

Related Case Studies

Agile GCP Labs helps Universal Music Group improve Analytics performance by 60%

Agile GCP Labs built Marketing Analytics for Kids Atelier on Google Cloud Platform

Agile GCP Labs helps Call Tracking company scale the VMs with Google Cloud Platform

Agile GCP Labs helps SocialHi5 with migration of Digital Marketing Analytics to Google Cloud

Agile GCP Labs helps Digital Marketing Firm with migration of 150 websites to Google Cloud Platform

Agile GCP Labs quickens Pharma sales insights for AMGEN by building a AI-powered Enterprise BI Bot

Agile GCP Labs enables CI-CD DevOps for Global M&E Conglomerate