We’ve seen quite a bit of change this year as businesses have had to pivot to accelerating their digital transformation strategy, and placing even more emphasis on leveraging technology as a competitive differentiator. Most have continued to stress the importance of maintaining excellent customer relationships through their contact centers, but the playing field has changed as they now have to tap into data for insights that may have normally been gleaned through an analog approach.
One trend that continues to grow rapidly, is relying on cloud-based technologies and solutions to get ahead. We see this today with Contact Center as a Service (CCaaS) offerings gaining traction because the skill sets needed to maintain complex environments that can operate seamlessly at the speed of business, are becoming increasingly difficult to find. Additionally, upgrading a large contact center environment to take advantage of bug fixes or features in a new release can be a daunting task, but if managed and offered as a service, the results can be a game-changer for customer experience. Those that choose to take a hybrid route to leverage all the goodness of cloud-based services, can get the best of both worlds by maintaining full control over the contact center environment with their own DevOps team, while leveraging the technology to deliver a consistent and exceptional customer experience.
The good news is, no matter the contact center environment, or stage of digital transformation, Splunk can help. For taking advantage of the “best of both worlds scenario,” we’ve introduced the Splunk App for Amazon Connect.
The Splunk App for Amazon Connect uses a variety of data sources to help gain insight into your contact center performance both historically and in real-time. In standard Splunk fashion, data is democratized so from DevOps and AppDev teams, to Network and Service Operations Centers, to Contact Center Operations, to Engineering and Capacity Management, there’s something in the app for everyone. Key features include:
- Call trace record (CTR) correlation
- Agent status
- Agent performance
- Queue statistics
- Instance performance
- Capacity management visibility
- Bidirectional sentiment analysis via Amazon Contact Lens
The application can help with monitoring your contact center and running analytics for Operations teams, with over-seeing web traffic and health of the application for the Services team and with debugging information for DevOps and App Dev teams.
We recommend the following architecture for integrating the Amazon Connect App with Splunk:
Apart from installing the Splunk App for Amazon Connect, the following apps are required to make complete sense of the data collected and visualization of the data itself.
- Splunk Add-on for AWS (for various inputs)
- Splunk Timeline (custom visualization)
- Event Timeline Viz (custom visualization)
Data Sources and Visualization
Your Connect instance generates two types of real-time streaming events - Contact Trace Records (CTR) and Agent Events. CTRs are generated immediately when a contact session ends and the agent events are generated both periodically and at specific intervals when a state change occurs with an agent or contact.
Connect sends these events to a Kinesis Data Stream and Splunk can leverage these events via Kinesis Firehose using the AWS supported Splunk and Firehose integration to determine agent status, contact information and initiation details, and timeline of the entire session among other things. This data can help the operations team monitor and adjust agent schedules, number of contacts and length of sessions. A Splunk user can create alerts based on any of these variables.
The Splunk App for Amazon Connect can also make use of the real time queue metrics generated by your connect instance. In the above architecture diagram, Splunk uses a lambda function that is triggered every 60 seconds to retrieve real-time queue metrics from the application's API endpoints and the records are then coalesced with queue information and then sent to a Kinesis stream. The data is then streamed into Kinesis Data Firehose (KDF) for Splunk to extract using HTTP Event Collector (HEC). Splunk uses these events to determine queue status in real-time.
The app can also interpret the admin-generated, scheduled historic metrics like Queue contact metrics and Agent performance metrics scheduled reports that are in S3. Splunk can retrieve these CSV reports from S3 using the Simple Queue Service (SQS)-based S3 input provided in the Splunk Add-on for AWS.
Splunk uses these reports to display historical agent and queue metrics. Metrics such as Average After Contact Work Time or Average Hold Time for a specific agent in the last one week, or Average Incoming Chat Requests for a specific queue for the last month, can help a supervisor or an analyst to better understand the bottlenecks and improve performance of the contact center.
Connect also generates instance level metrics. These metrics help to determine the overall health of the application. Splunk can retrieve these metrics using the Splunk Add-on for AWS. Metrics such as Voice Call Packet Loss in the last hour can help service or network operations teams to analyze contact center voice quality and and remediate issues in real-time. Call concurrency metrics can be leveraged to plan for growth and better predict capacity requirements.
The Splunk App for Amazon Connect can also make use of contact flow logs. When contact flow logging is enabled in Connect, the logs are sent to Cloudwatch logs. Splunk can pull those logs from specific cloud watch log groups using the Splunk Add-on for AWS. This information can help an amazon connect app developer to debug and figure out exactly which function or module is failing and quickly remedy the contact flows.
Last but not least, Contact Lens for Amazon Connect is a set of machine learning (ML) capabilities integrated into Amazon Connect. When enabled, the service analyzes the voice call and publishes a report to a S3 bucket. Splunk can pull these JSON reports from S3 using the SQS-based-S3 input available in the Splunk AWS TA.
With Contact Lens for Amazon Connect, customer service supervisors can conduct fast, full-text search on call transcripts to quickly troubleshoot customer issues. They can also leverage call-specific analytics, including bidirectional sentiment analysis and silence detection to spot customer experience issues and improve customer service agents’ performance. Reports can be generated from Splunk to identify where additional coaching may be needed, and additional dashboards can be created and leveraged for gamification to improve agent experience.
Probably one of the most exciting features that Amazon will be launching is real-time sentiment analysis. The ability to impact and turn-around a poor customer experience as it occurs, is something that most consumers have wished for at some point in time. Once that feature is launched, Splunk will be ready to help deliver actionable insights to positively influence customer experience in real-time. Supervisors and team leads can automatically be alerted when specific key words are spoken, excessive silence is detected, or sentiment scoring on either the consumer or agent side is deteriorating.
Another popular request from our customers is for visibility into remote agent connectivity to identify when there’s contention for resources or disruptions in “last mile” network connectivity that negatively impact customer experience. We’re planning to incorporate information from CCP logs that should address some of these challenges.
Details on how to configure real-time streaming events, API metrics, historical reports, connect instance CloudWatch metrics, contact flow CloudWatch logs, and Contact Lens reports with Splunk, along with helpful artifacts, can be found here.
Check out the Contact Center Analytics solution guide for more information on how to unlock contact center use cases by taking a data-driven approach and join the Contact Center Data Analytics group on LinkedIn to exchange thoughts and ideas.
Anush Jayaraman is a Senior Solutions Engineer for the Global Strategic Alliances team at Splunk. He joined Splunk in the Summer of 2015 and has worked on a variety of products and got to wear multiple hats within engineering over the years. From building infrastructure and automation, and release engineering to developing micro services for Splunk Cloud Services, Anush is obsessed with learning new technologies, developing tools and building integrations.