Mining and Analytics in E-Commerce
Stephan Kudyba, New Jersey Institute of Technology
It seems that when anyone considers conducting analytics in e-commerce, whether it be reports, dashboards, or mining, the first thing they think of is website activities. However, given the evolution of the electronics products market, wireless communication, and social media applications, the term e-commerce has taken on a vastly new and complex meaning. Consider marketing initiatives in today's bold new e-commerce world. Only years ago, analysts focused on click rates and banner views and traditional email activities. Today the e-commerce marketing professional must be versed in such concepts and strategies as:
- Deriving product likes on social media
- Developing apps to engage customers
- Maintaining active Facebook accounts
- Understanding what platforms people are using to engage their marketing tactics
- Sending text messages
- Better understanding email formats
- Managing consumer comments on blogs
- Engaging customers to acquire insightful market feedback
- Understanding consumer activity on websites
These are just some of the noteworthy activities to keep in mind when considering strategies to achieve success for an organization. One major advantage all these activities involve is the generation of data that measure corresponding activities that underpin many e-commerce-based initiatives, and of course with data comes analytics.
In all the concepts mentioned above, the analyst must make the decision as to what level of analytics is necessary for his or her operation. Many times simple graphic displays in the form of dashboards (with access to real-time data) may suffice in providing pertinent information. More dynamic multidimensional cubes may be required, or perhaps more complex mining applications to estimate likelihoods, associations, and predictive power may be the solution. The key to any of the analytic methodologies lies in their robustness in providing a well-rounded view of the process they are addressing. For example, consider the simple notion of understanding effectiveness of an organization's mobile app initiatives.
Assume that your organization has produced a well-functioning application that it wants to send to users in order to maintain an active engagement. One way to leverage analytics to better understand the success of that initiative is to include relevant data variables that provide information that fully describes that scenario. Merely sending apps to consumers and monitoring whether it is downloaded does not imply a successful endeavor. A better performance metric is utilization of the app. However, utilization must be clearly defined as to meaningful utilization (active and continuous engagement on a daily/weekly basis). Other variables that must be considered include:
- Type of technological platform the consumer uses (Android, iOS, etc.)
- Consumer's response to alerts/messages
- Type of consumers that are either active or nonactive engagers
These variables, if analyzed in the correct format over relevant time horizons, can provide strategists valuable information as to effectiveness of mobile app initiatives. Poor consumer activity/engagement of a downloaded app can quickly give an organization a better understanding of whether the functionality of the app is relevant to the marketplace, or perhaps the wrong target market was chosen as a primary user base.
Basic Email Reports
Sometimes more basic analytics are adequate for strategists to quickly extract knowledge of a particular initiative, where more routine activities simply require a timely and accurate report of basic metrics describing results from an initiative. Consider the highly utilized activity of text messaging or traditional email messaging of a consumer base. Perhaps an organization's focus is more on effective and timely sending of these communications to large segments of the marketplace. Here general metrics regarding responses are necessary. A simple report with core metrics to this routine activity with supportive graphics is all that an organization may require. Consider metrics within an email marketing initiative:
- Number of emails sent
- Emails delivered
- Number of emails bounced
- Number of emails opened
- Amount of clicks on links in email
- Conversions of clicks
The metrics above provide core information to consider regarding email marketing initiatives. A simplistic but noteworthy variable involves the number of emails delivered. A report that illustrates that if you send 100,000 messages, but only 90,000 are delivered, provides the strategist with basic but very actionable information. Simply because an email address has been acquired doesn't imply that it is still active. In the case above, 10,000 emails may need to be scrubbed from the active list. Bounce rate also provides a mechanism to filter or optimize future email endeavors. The rate of email opens provides another source of vital information. If only 5% of emails sent are being opened by recipients, perhaps the marketer needs to consider more enticing subject headings or email IDs that send the messages.
Analyzing Web Metrics
Websites provide a host of trackable metrics that depict visitor activities that provide decision makers with information to optimize site layout, enhance marketing effectiveness, or simply enhance the communication effect of the organization with the marketplace regarding its products and services. Prominent metrics that often are leveraged in site analytics include the following:
- Page views
- Time spent on pages
- Navigation route with a website
- Click rates on banners or icons
- Bounce rate
- Conversion or interaction rates
- Source of traffic (geographic, referral sites, unique or repeat visitors)
All these metrics describe traffic during different times of the day that may be initiated by a host of marketing tactics. Initial page views may be a result of paid and generic search initiatives, email and text messaging, or even traditional advertising initiatives, such as radio, TV, or print. In order to fully understand basic metrics generated at the site level, we must consider a combined effort of ongoing strategic marketing tactics (e.g., e-commerce, TV, direct marketing). Remember, just because an organization has built a website doesn't imply that consumers will automatically access it; the marketplace usually needs a little incentive to visit it.
Regardless of the marketing push to the site, analysts can leverage available metrics to better understand their website effectiveness. In order to do this, insightful consideration of those metrics needs to implemented. For example:
Bounce rate can be a powerful metric in describing a market's response to your organization. A high bounce rate can imply that either the site display is poorly designed, as viewers simply aren't interested in what they are seeing, or the marketing initiative that pulled a particular market segment to the site landing page was not appropriate for the organization's products or services (e.g., the message may have been misleading). However, the analyst must consider not just total bounces, but a bounce rate or number of bounces per views.
A total bounce count of 10,000 may be high in a given day, but if a TV initiative drives 2 million visitors to a site, the bounce rate (bounces/views) may be in the realm of normal activity.
Of course one of the most popular areas of analytics for websites remains simple traffic patterns throughout a site layout. This includes analyzing how site visitors navigate around a site once they have engaged it. A common essential piece of information for decision makers involves such navigation elements as where traffic came from, hot spots in select pages, time spent on pages, and route of navigation, to name a few. These areas provide decision makers with actionable information on where to display advertisements or invoke interaction with visitors, optimize site content, and focus marketing tactics in attracting more traffic.
Time spent on particular pages could signify either a source of interesting content to visitors or a possible glitch in the site (poor information that is not understandable, or difficult options to navigate out of an area). Other metrics, such as back page, should be analyzed in such cases. More time spent on pages due to interesting or hot areas provides actionable information to leverage marketing tactics.
Analyzing sources of traffic, or where traffic originated before coming to a particular website, can provide a great source of strategic information. Geographic areas of traffic origin can provide decision makers with a better understanding of the target market of a company's products and services, and once again can help fine-tune marketing initiatives. A quick view of analytics can illustrate whether a target market is local, regional, or national, where paid search and banner display activities can be optimized accordingly. For example, strong traffic at the county level can provide insights for the following:
- Keyword selection
- Better selection of complementary, local sites to display banners
- Filter out IP addresses outside the target market
Figure 1. Heat map hot spot. (From http://provim.net/heat-maps.)
Visuals can play a vital role to strategists when analyzing web metrics. Although pure numbers that depict metrics as described above are essential to conducting fine-tuned, detailed analytics in pursuing strategic initiatives, it is usually a well-designed visual that provides decision makers with a quick understanding of web-related activities. Hot spots in websites and traffic sources through heat map visuals quickly give analysts a robust view of what is happening. Hot spots in websites (see Figure 1) generally depict traffic activity in certain sections of a page (e.g., upper left or middle right side of a particular page). Sifting through the core numbers that underpin this could be a more time-consuming process and provide less immediate knowledge impact to the analyst/strategist.
Read more IT Performance Improvement
Certain names and logos on this page and others may constitute trademarks, servicemarks, or tradenames of
Taylor & Francis LLC. Copyright © 20082014 Taylor & Francis LLC. All rights reserved.