Wednesday, December 15, 2021

Customer support: How Azure Metrics Advisor can help improve customer satisfaction

Customer service & support is crucial to every single organization on the planet. Customer experience enables an organization to develop a competitive advantage that is harder to replicate. According to BCG (The Boston Consulting Group), companies with the highest customer satisfaction scores have generated twice as much shareholder value over the last ten years relative to the average score. (Source:  BCG) 

Developing a robust support organization means – you are serving customers 24*7, anticipating their needs, offering lower wait times and faster resolutions. Some of the core support operational challenges are  

  • The Pandemic: The Pandemic has really shifted the dynamics of support organizations globally 
  • Acts of Nature such as natural disasters affecting your support sites/ suppliers 
  • Unexpected surges in support volumes  

 

Anomaly Detection in Support: 

 

To provide better support 24*7, one must constantly monitor the support KPIs (Key Performance Indicators) such as Support Volume, Support HeadCount, and Customer Wait time, detect anomalies across geographies and products/services, etc. An anomaly can be defined as a meaningful change in the expected value of variability. Anomaly detection is an instance where AI (ARTIFICIAL INTELLIGENCE) can help by detecting the anomaly, notifying both the business process and planning team, providing the root cause, and either automatically fixing the issue or enabling the support operations managers to make better-informed decisions. 

 

User stories: 

 

There are countless user stories. However, some common user stories amongst most support organizations are 

 

Scenario 1: The Surge of the Pandemic 

 

The Pandemic drastically affected every support and professional organization in the world. – Pushing many to Work from Home models while trying to manage an array of new operational and logical challenges. Customer Support Business leaders were trying to manage this impact, while Support volumes continued to rise. Our customers needed more assistance than ever. Business leaders and crisis managers knew that the customers needed support but there were significantly fewer resources available to provide it. Determining which businesses, partners, and geographies were impacted and by how much each day was a slow and laborious task. There was no way to quickly analyze the headcount information, the volume of customer support requests, and alert the crisis managers to businesses that were the most impacted.  

 

Scenario 2: Acts of Nature 

 

Many businesses are geo-diversify and outsource operations support to improve business resiliency, reduce cost, and scale globally. While operations teams, suppliers, and partners often have robust business continuity processes – none can break free from the unpredictable acts of nature such as flooding, earthquakes, or other human-made disasters. A massive typhoon crashed into a major country, putting lives at risk, damaging homes and businesses, disrupting critical support operations. The significant reduction in available support staff triggered massive backlogs of support requests, long wait times, and negative customer experiences. Business leaders needed insights into how these volumes could be rerouted all over the world. Lacking a 24*7 alerting mechanism to identify the most impacted support queues meant that the business leaders and crisis managers had to rely on manual investigations and wasted precious hours during critical moments that matter.  

 

Scenario 3: Unexpected Volume Surge 

 

Planned and unplanned Outages for products and services often generate a huge surge in support case requests. While engineering teams focus on resolving the issue, support requests from all modalities flood in, causing SLAs (SERVICE LEVEL AGREEMENTS) to be missed as engineers’ backlogs became unmanageable. Identifying the Root Cause Analysis and the impact of one metric over another such as Volume on SLAs are some of the hardest challenges in support organizations. 

Value of metrics advisor in Support: 

 

While the support metrics data can be analyzed in any Data Visualization platform like PowerBI – The Visualization tools fall short of analyzing multi-dimensional metrics, integrated alerting mechanisms, and RCAs (Root Cause Analysis). 

  1. Determining which businesses, partners, and geographies were impacted and by how much each day was a slow and laborious task: Metrics Advisor can analyze multi-dimensional metrics (Eg: Support Volume can be analyzed by multiple dimensions such as Geography, Products/Services, and Support Modalities such as Phone, Email, etc.) and generate multiple series for the same metric.
    • Eg: To monitor “Support Volume” metric across many geographies and many products/Services, Metrics Advisor will generate many time series’ for “Support Volume” metric across all the permutations of Geographies and Products/Services 
  2. Lacking a 24*7 alerting mechanism to identify the most impacted support queues meant that the crisis managers rely on manual investigations and waste precious hours during moments that matter: Metrics Advisor has a built-in sophisticated alerting system that can configure a multi-level alerting hierarchy based on the severity of the alerts and the multiple dimensions of a metric. Simple integration into Teams and Emails makes it super easy. Metrics Advisor also can capture action items associated with each Alerts. 
  3. Identifying the Root Cause Analysis and the impact of one metric over another such as Volume on SLAs are some of the hardest challenges in the support organizations: Metrics Advisor can perform Root Cause Analysis across metrics and the impact of one metric over another in an easy to consumable graphic.  

 

One of the core challenges in the world of support – is to identify patterns of seasonality. Some simple examples are, 

  • Headcount seasonality by the weekend: Many support organizations have reduced support professionals on weekends or during non-business hours. 
  • Support Volume during release cycles: Support Volumes surges or drops during the release cycles/holidays etc. 

 

Identifying outliers/anomalies in a data stream would be useless if the algorithms cannot account for seasonal fluctuations. Below here is a sample visualization of Daily Support HeadCount over a Period of time. As you can observe, Metrics Advisor does not flag every dip as an anomaly. It can easily identify the anomalies –considering the seasonality.

Qiyang_1-1639551656509.png

Note: Refer Overview of SR-CNN algorithm in Azure Anomaly Detector - Microsoft Tech Community for more details.

With all the above – Metrics Advisor is an ideal tool for the world of support, as we try to build proactive experiences for our customers. With a full set of API (Application Programming Interfaces) capabilities – You can also build your own custom portals/web wrapper over Metrics Advisor. Metrics Advisor puts you in the driver’s seat in creating phenomenal customer experiences. 

 

Data preparation & Onboarding: 

 

It is vital to identify the list of support metrics to track. For a typical global support organization: Some of the core metrics are, 

  1. Support Volume 
  2. Support HeadCount 
  3. Customer Wait Time 
  4. Customer Satisfaction 

 

Some of the core dimensions that are applicable across all the above metrics are 

  • Geography: Country, City 
  • Product/Services: Product/Services Hierarchy followed within your own organization etc. 

 

As you evaluate the data schema - other critical elements to think though are Granularity and Refresh Interval. To achieve real-time or near real-time monitoring – you can decide on a granular time series such as hourly/daily/weekly, etc. 

Here are some sample schemas: 

Timestamp 

Product/Services Hierarchy 

Geography 

Support Volume 

2021-11-01 

Product/Service X 

Country A 

100 

2021-11-01 

Product/Service Y 

Country A 

300 

 

 

 

 

2021-11-02 

Product/Service Z 

Country B 

200 

2021-11-02 

Product/Service X 

Country C 

50 

 

 

 

 

2021-11-03 

Product/Service Z 

Country C 

500 

 

Timestamp 

Product/Services Hierarchy 

Geography 

Support Head Count 

2021-11-01 

Product/Service X 

Country A 

25 

2021-11-01 

Product/Service Y 

Country A 

60 

 

 

 

 

2021-11-02 

Product/Service Z 

Country B 

60 

2021-11-02 

Product/Service X 

Country C 

10 

 

 

 

 

2021-11-03 

Product/Service Z 

Country C 

100 

 

Metrics Advisor can connect to a variety of data sources such as SQL, Kusto, Blob Storages etc. making it easy to get started immediately.

 

Alerting & RCA (Root Cause Analysis):

 

Metrics Advisor offers an integrated alerting system built within the tool. Easy integration with Emails or Teams enables your team to drive operational processes easily. A variety of alerts can be configured to enable the following scenarios: 

  • All alerts should be sent to centralized operations teams. However, severe alerts should be sent to the Customer Support Leaders for visibility 
  • Alerts should be sent to different teams based on Geographies or Products/Services or any number of dimensions (You can leverage “Series Groups” for the same) 
  • Cross metric alerts: send an alert only if “Support Volume” goes above the threshold AND “Support Head Count” goes below the Threshold 
  • And more 

 

Metrics Advisor also enables you to perform RCA on incidents and enables you to also track actions on the same.  

For example: If the support headcount in specific geography dropped due to a natural disaster – Metrics Advisor will flag that as an anomaly and perform the root cause analysis:

  • As a support operations manager, you can capture the Incident summary: The incident, root cause, the actions taken to resolve the incident in the tool directly.
  • As a support operations manager, you can also analyze the cross-dimensional impact. How did the support headcount drop in the specific geography impact the various products/services in your organization’s portfolio? 
  • As a support operations manager, you can also analyze cross-metric impact. How did the support headcount drop in the specific geography impact the Customer Wait time in that geography? 

 

High-Level Reference solution architecture:

Qiyang_2-1639554392341.png

 

 You can refer to the solution architecture to achieve the E2E (end to end) scenario described above. The first step is data preparation we talked about in the earlier section” Data preparation & Onboarding” to do that, you will need a streaming processing framework like Azure Data Bricks to transform the raw data from the supply chain system into the required schema. The cooked dataset then is stored in data stores like Azure Data Lake or Cosmos DB. Then Azure Metrics Advisor can connect directly to those data stores and leverage the dataset for model building and parameter tuning for anomaly detection and root cause analysis. Finally, those results can be available for visualization through Power BI or notification to users via emails or other channels. 

 

Microsoft internally is adopting Metrics Advisor to identify Support operational risks to provide a better customer experience. 

 

Get started today 

 

Go to Azure portal to create your new Metrics Advisor resource here. You can also read the Metrics Advisor document to learn more about the service capabilities. 

 

Posted at https://sl.advdat.com/3F1jpJo