IBM WebSphere Commerce (WSC) is a robust and highly scalable platform for powering omni-channel commerce. However, to monitor WSC effectively for the purposes of maximizing revenue through the delivery of a high-performance customer experience, it is not enough to just monitor PMI metrics, CPU, memory and disk of the underlying infrastructure that runs the IBM WSC engine. It requires collection of a rich data set that can be presented to various stakeholders to ensure that WSC is delivering a positive experience to digital shoppers. As with any e-commerce application, there are role-relevant metrics that are valuable to different stakeholders. These stakeholders can be divided into three categories:
- Digital Business Owners in the Line of Business group
- IT Operations
- Architects, Dev/Test, DBAs
Effective Collaboration between these three cross-functional teams requires a single toolset with views tailored to each stakeholder’s group. By using a single toolset, the data used by all three groups is from the same source, alleviating the need to determine which group has the correct data, which traditionally consumes precious time and effort.
As a business user, your main focus includes a real-time view of revenues as well as user struggles that prevent conversions. The key performance indicators you will be looking at are:
- Customer Experience: Measuring the end user experience along the full digital journey is a key component to successful sales. Faster, well-designed and more responsive sites and mobile apps translate into higher conversion rates, maximizing revenues, and driving brand loyalty.
- Key Page Health: As customers navigate through the e-commerce site, some pages or mobile app views matter more than others. In a typical transaction, a customer searches for and then retrieves a product list, chooses the product(s) to buy, and goes through the checkout process to complete the order. Real-time health and performance monitoring of the search, product display and checkout process are key to limiting the potential revenue impact due to performance issues.
- Customer Context: Personalization drives engagement and conversion. Visibility into customer behaviors across digital channels used at each interaction, time of the day, location, device, browser and network enables business owners to design more effective customer engagements.
- Visit Duration: It is well known in Retail that the more time customers spend on the site, the more chances there are for them to make a purchase. Business users use the visit duration not only to know the effectiveness of the site but also to compare it with the industry standard.
- Conversion Rate: It costs 6 to 7 times more to acquire new customer than to retain an existing one. Maximizing conversion is the most business effective way to impact the bottom line. Correlating user experience and conversion rate enables early detection of potential user struggles and infrastructure issues impacting transactions.
- Sales Conversion Funnel: The sales funnel is an intuitive way to understand if there are user experience issues at key business steps that drive down your conversions. We’ll look more closely at this in part II.
As an IT Operator, you are responsible for the overall health of your e-commerce platform. You are responsible for the platform availability, accessibility, normal operations and delivery of satisfying customer experiences. The metrics you will mostly look at are:
- Host Health: The health of the machines that support your IBM WebSphere Commerce (WSC) instances has a direct impact with your overall e-commerce platform health. Popular metrics that are used to gauge the host health are CPU, memory, disk and health.
- Application Process Health: Even though the host may be healthy, the application processes can suffer from high garbage collection rates, high memory utilization, slow disk I/O etc. Proactively monitoring them ensures good end-user response time.
- Key Page Health: Like business users, as an IT Operator you also care about key pages health since they are the backbone of your revenue stream. Early detection and effective remediation is key. Error count, response time and number of requests all matter and you need to compare these with your baseline.
Architects, Developers & Testers
You are in charge of delivering rich experiences with advanced capabilities, faster than ever to keep your ecommerce business ahead of your competition and in tune with your customer behavior and buying trends. Key metrics of interest include:
- Slowest BOD Pages: The list of slowest Business Object Document (BOD) pages for a given duration provide a quick preview into the set of web pages that are running slowly. Drilling down from this list with all the supporting technical context, you can quickly fix the application code causing the slowdown.
- API Breakdown: API breakdown provides a key insight about which layer in the code is taking the most amount of time on the slowest BOD pages. For example, if the API breakdown shows a high percentage of time spent on JDBC, you immediately know that the database calls are taking the most amount of time for all transactions for a given duration.
- Failure Rate: Proactive monitoring of the failure rate is essential to ensure overall e-commerce platform health. If the failure rate spikes up, you can quickly identify whether the application code is causing the issue or tiers that support the application are the culprits.
As most of the e-commerce data is persisted in a relational database, the health of the database is very important to the health of the response time. To ensure that all the queries are running properly and are using the right query plan, the metrics you care about are:
- Top Slow SQL Statements: This list of SQL statements provides a concise list of the queries that need some tuning. You can take this list, review the query plan and make appropriate database or query structure changes to improve query response time.
- Database Time vs. Count: This metric allows you to see the correlation between increase in SQL count and the query execution time. If there is a strong correlation between the two, one of the possible areas that you can look is table contention.
- Database Hotspots: This metric provides a view of the most frequently used queries and where they are coming from. This provides a quick view of where optimizations can be made to reduce the amount of time on the database.
When running IBM WebSphere Commerce, look beyond the typical metrics such as PMI and host level metrics in isolation, and instead focus on the user experience as customer transactions execute across the application tiers. There are many valuable metrics that can then be extracted and delivered to different stakeholders. The ultimate goal of running an e-commerce platform is to maximize revenues through conversions driven by great customer experiences. This goal can only be achieved if you have a unified view of your digital channels, from the browser or mobile app, to third parties all the way to supporting application code and databases.