The priorities for the CIO
The priorities of CIO of an organization include delivering on business solutions, business intelligence and analytics and improving the customer experience with the brand. The IT operations team within the CIO is responsible for the IT systems, which include,
- development and innovation of business solutions
- production management and uptime
- operational efficiency, providing with the analytics from operational and transactional data for the consumption of both IT and business
The priorities for the retail CIO
Additionally, the retail CIO is grappling with the world of omni-channel, building a single customer experience, and working towards delivering personalized services. This involves building a mobile strategy as the statistics continue to show higher exploration on mobile and higher conversions on desktops/tablets. This also means the focus on the Big Data paradigm, to gain competitive advantage.
In the following blog, I will outline the layers of monitoring that an IT Operations team will need to setup as the commerce site is ready to Go-Live. The key operations data is required as soon as the system becomes available to the customers. The retailer must get their data in view and the thresholds and alerts set out prior to the launch. The alerts should be set based on the performance cycle for the launch. The other business data can be configured based on the priorities and focus of the business.
The metrics and measurement
1 – The key performance indicators (KPI) which a retailer wants from their digital channels are
- Visitor summary: page views, average session length and time spent
- Visitor behavior: bounce rate, new visitors and conversions
- Product views, top search terms, search hits and misses, cart abandonment rate
- Paid search sessions, marketing, referring site sessions
- Browser types, devices
- User behavior across different channels
- Performance against industry trends
There are several analytics tools which offer integrations with websites. IBM’s Digital Analytics is the SaaS offering which provides insights on this data and also publishes industry trends across holiday seasons and events. Some of the reports that any retailer will find very insightful to plan for upcoming 2015 events and holidays are the 2014 Black Friday report, 2014 Cyber Monday Report, Worldwide Online Holiday Mobile Shopping Report 2014, UK Retail Online Christmas Shopping Recap Report 2014.
2 – Retailers have to manage their operations and deliver to business based on the uptimes, scalability and non-functional requirements (NFR). Some of the metrics which are must-monitor for a site to deliver client satisfaction are:
- Application performance metrics like CPU usage, JVM heap usage, garbage collection, Web Container threads
- Backend performance and response times from 3rd party integration systems like payment gateway, or external applications like inventory, availability, address verification
- Response times for the various operations like browse, search, cart, order completion
- Requests per second for various incoming requests
- Number of orders, order-size, order-value
- Cart abandonment
The Application Performance Management tools are specifically focused on monitoring and management of performance of the software applications. The tools provide a view on most of the above parameters, by either using agents or are based on log analytics. The out-of-the-box features have to be augmented by providing visibility to other metrics, like the order information will be extracted from the database. The agent-based tools need to be configured correctly and tested to ensure that they do not impact the performance of the host application. We will explore some of the tools under APM in future blog topics.
3 – Application specific monitoring aspects
Java applications have standard mechanism for managing application resources using Java Management Extensions (JMX), is managed bean (MBean). The WebSphere Application Server can be extended with custom JMX MBeans. One of the examples to demonstrate a key performance resource for WebSphere Commerce is the caching efficiency. The performance of the WebSphere Commerce can be optimized by building in a strong caching methodology at various levels. The application level caching is built upon the WebSphere Application Server Dynacache which caches JSP/servlets and data object. The cache statistics of cache hit, cache miss, hit/miss ratio, the cache eviction, invalidation, provide insights into the site usage and its performance. The dynamic cache service provides an MBean interface to access cache statistics. The Extended Cache Monitor tool pulls the mbean statistics for cache instances and displays the following data:
- Display contents of servlet and object cache instances
- Display the Dynamic Cache mbean statistics for cache instances
- Display the dependency ids of the cache instances
- Display the disk offload statistics and contents
- Invalidate cacheids
There is a lot of data in the commerce system, and the IT Operations can harness the information for insights. The data ranges from the application performance metrics to the business data and analytics for business insights. The downtime, slow site, unknown issues in the infrastructure will catch an operations team blind-sided, impacting in loses and brand dilution. The monitoring and alert enables the operations team to manage and troubleshoot the system in a predictable context. The operations team should evaluate and chose the monitoring tool based on the data they want to view. The tool should be tailored with correct thresholds, alerts, messaging to provide the effective and actionable insights.