E-commerce industry has witnessed a phenomenal growth in the past and is poised to grow even more in the ensuing future. Today, consumers are more focused than ever on finding the best prices. Therefore, e-commerce firms, for whom the utmost objective to make delivery of quality products and services at best available prices, need to segment their customer base and consequently customize, if they want to fully capitalize on this opportunity . By gaining a better overall understanding of customers, then grouping them into categories, companies are able to better optimize marketing programs and allocate marketing dollars more effectively. Defining and formulating these segments require careful analysis of customers and products using web analytics.
The article begins with the importance of understanding web traffic, different metrics that are derived from web analytics tools, and shows how analyzing this real time web data will help in user profiling. It concludes by showing how it will help the organization in personalizing and customizing the web experience with their customers and results in higher conversions.
Why Understanding Web Traffic is Important
For an e-commerce firm, the penultimate goal is the completion of a transaction- implies buying something. However, it seems unlikely to accomplish particularly in the case of potential customers who are first time visitors. Thus another useful goal in this context is to measure the number of people who add an item to the shopping cart, whether they purchase it or not- in other words, how many begin the shopping process.
Metrics derived from Web Analytics Tools:
How many daily visitors visit on website
Average conversion rate
Click through rate (CTR)
Top-visited pages
% Exit: the percentage of users who exit from a page
Geographic distribution of visitors
‘Stickiness’ of web pages
Defining Target Audience: RFM Law of Marketing
Recency, Frequency and Monetary Value (RFM) is the most widely known law of marketing. It is defined as follows:
Recency is the most powerful single factor affecting customer repurchase. The person most likely to buy again is the customer who bought most recently.
Frequency is the number of times that a customer engages in the desired behaviors. Experience shows that frequent buyers respond better than infrequent buyers.
Monetary Value involves categorizing all customers by the amount that they spend in a given time period.
The target is to increase the customers having high scores of Recency, Frequency and Monetary value.
Assume there are 3 variables: Recency (R), Frequency (F) and Monetary value (M). Each of the customers will be scored on each of these three variables. Assuming 1 to 3 scoring system is used to score consumers with 1 given to those who exhibited the least desirable action and 3 given to those who exhibited the most desirable action.
Recency Score Criteria
1 = Customers who purchased more than 12 months ago
2 = Customers who purchased more than three months ago, but fewer than 12 months ago
3 = Customers who made a purchase in the last three months
Frequency Score Criteria
1 = Customers who made a purchase in the past 12 months
2 = Customers who made between two and ten purchases in the past 12 months
3 = Customers who made more than ten purchases in the past 12 months
Monetary Value Score Criteria
1 = Customers with an average purchase amount up to Rs 1000
2 = Customers with an average purchase amount between Rs1000 and Rs 2500
3 = Customers with an average purchase amount greater than Rs 2500
After assigning scores to three variables, R, F and M, and sorting. It’s possible to identify best segments, worst segments and the segments in between. These segmented customers require different strategies:
Customers with RFM of (3, 3, 3): These are profitable customers and need to be thanked. While attempting to keep targeting these highly profitable customers with incentives to buy more, it may actually risk offending this segment and losing their business. This group does not require any incentive. A simple “thank you” and recognition will go a long way.
Customers with RFM of (1,1,1): These are customers which are not profitable.
Customers with RFM of (3,3,2) or (3,2,2) or any other combination having two 3s + one 2 or two 2s + one 3: These are the customers that need to be targeted. These segments need to be moved up to higher ladder from marginally profitable to highly profitable. It will require understanding the motivations, preferences, interests and expectations of each segment and then acting accordingly.
Understanding the preferences and interests of this “Target” audience requires analysis of web site measurement data. These target audience will further need to be segmented depending on different factors like session time, geographical location, source or channel of web page visit, pattern of visited pages etc. Here comes in the role of Web Usage Mining.
Segmenting Customers: Deploying Web Usage Mining
Web Usage Mining is defined as the application of data mining techniques to large web data repositories which aids in extracting usage patterns, namely the visitor behavior.
In the framework of Web Mining, User Profiling is one of the fundamental applications. User profiling is the act of building up a profile of who users are and what they want to do. These profiles are used to group visitors/potential customers and hence forth the priorities in their activities are identified. Knowing who the visitors are and what they want is a vital step in meeting their needs.
Web usage mining activity involves following activities:
Web data pre-processing
Pattern discovery and analysis
Web Data Pre-processing
The main task of user profiling is to organize data such as to collect consistent information about the users. Information is bifurcated as explicit and implicit information: the former is derived from the data provided by the user in filling any kind of application form while the latter one is collected from website tracking tools through log files, cookies etc.
Pattern Discovery
The output of web data pre-processing will aid in performing pattern discovery. Pattern discovery activity involves deciphering how the web pages are viewed.
Following patterns potentially emerge:
Association: Identifying groups of pages that are jointly consulted within a set of sessions in navigation.
Sequential: Used to find pages that are accessed in sequence.
Episode: Episode is a subset of page views in a particular session.
Classification/ Segmentation
Data generated and reformatted in pre-processing stage and patterns observed will be useful in making classification rules. In the e-commerce industry, aim of the management is to understand those factors that discriminate the web visitors between buyers and non-buyers.
Web Personalization
The output of web usage mining is web visitor profiling which becomes the input for web personalization. Web personalization involves customizing the web relationship with individual consumers based on understanding their preferences, needs and value. Segmentation is the intermediate step between web usage mining and web personalization.
Types of personalization
Memorization: It is based on memorization of some information about the navigation which is used as soon as the user returns on the web site.
Customization: This type of personalization considers the information filled by the user in the application form and in this respect it takes into account of the privacy and depends on the cooperation of the user.
Personalization models follow two main approaches:
Content-based filtering: This exploits the profiling methodologies to match the user profile and the contents, the products and the services proposed on the Web site.
Collaborative filtering: This is a collaborative approach where the user is invited to express his interest or rate about the visited or downloaded contents. All information is memorized such that the contents are proposed to all users of the same typology.
Finally, the recommendation engine uses all the available information to provide a suitable suggestion for the visit. Two measures are considered to evaluate the performance of the personalization model:
Precision: Computed as the ratio between the "correct" recommended pages and the overall number of recommended pages.
Coverage: Defined as the ratio between the correct recommended pages and the overall number of visited pages, thus measuring the impact of the overall personalization system against its absence.
Conclusion
Web analytics is much more than hits, visits, and page views. The segmented buckets derived from web mining activity helps organizations to target them with specific marketing campaigns and promotions.
E-Commerce Company should follow the AMAT approach:
A: Acquisition of visitors
M: Measurement of performance
A: Analysis of trends
T: Testing to improve
The article begins with the importance of understanding web traffic, different metrics that are derived from web analytics tools, and shows how analyzing this real time web data will help in user profiling. It concludes by showing how it will help the organization in personalizing and customizing the web experience with their customers and results in higher conversions.
For an e-commerce firm, the penultimate goal is the completion of a transaction- implies buying something. However, it seems unlikely to accomplish particularly in the case of potential customers who are first time visitors. Thus another useful goal in this context is to measure the number of people who add an item to the shopping cart, whether they purchase it or not- in other words, how many begin the shopping process.
Metrics derived from Web Analytics Tools:
How many daily visitors visit on website
Average conversion rate
Click through rate (CTR)
Top-visited pages
% Exit: the percentage of users who exit from a page
Geographic distribution of visitors
‘Stickiness’ of web pages
Defining Target Audience: RFM Law of Marketing
Recency, Frequency and Monetary Value (RFM) is the most widely known law of marketing. It is defined as follows:
Recency is the most powerful single factor affecting customer repurchase. The person most likely to buy again is the customer who bought most recently.
Frequency is the number of times that a customer engages in the desired behaviors. Experience shows that frequent buyers respond better than infrequent buyers.
Monetary Value involves categorizing all customers by the amount that they spend in a given time period.
The target is to increase the customers having high scores of Recency, Frequency and Monetary value.
Assume there are 3 variables: Recency (R), Frequency (F) and Monetary value (M). Each of the customers will be scored on each of these three variables. Assuming 1 to 3 scoring system is used to score consumers with 1 given to those who exhibited the least desirable action and 3 given to those who exhibited the most desirable action.
Recency Score Criteria
1 = Customers who purchased more than 12 months ago
2 = Customers who purchased more than three months ago, but fewer than 12 months ago
3 = Customers who made a purchase in the last three months
Frequency Score Criteria
1 = Customers who made a purchase in the past 12 months
2 = Customers who made between two and ten purchases in the past 12 months
3 = Customers who made more than ten purchases in the past 12 months
Monetary Value Score Criteria
1 = Customers with an average purchase amount up to Rs 1000
2 = Customers with an average purchase amount between Rs1000 and Rs 2500
3 = Customers with an average purchase amount greater than Rs 2500
After assigning scores to three variables, R, F and M, and sorting. It’s possible to identify best segments, worst segments and the segments in between. These segmented customers require different strategies:
Customers with RFM of (3, 3, 3): These are profitable customers and need to be thanked. While attempting to keep targeting these highly profitable customers with incentives to buy more, it may actually risk offending this segment and losing their business. This group does not require any incentive. A simple “thank you” and recognition will go a long way.
Customers with RFM of (1,1,1): These are customers which are not profitable.
Customers with RFM of (3,3,2) or (3,2,2) or any other combination having two 3s + one 2 or two 2s + one 3: These are the customers that need to be targeted. These segments need to be moved up to higher ladder from marginally profitable to highly profitable. It will require understanding the motivations, preferences, interests and expectations of each segment and then acting accordingly.
Understanding the preferences and interests of this “Target” audience requires analysis of web site measurement data. These target audience will further need to be segmented depending on different factors like session time, geographical location, source or channel of web page visit, pattern of visited pages etc. Here comes in the role of Web Usage Mining.
Segmenting Customers: Deploying Web Usage Mining
Web Usage Mining is defined as the application of data mining techniques to large web data repositories which aids in extracting usage patterns, namely the visitor behavior.
In the framework of Web Mining, User Profiling is one of the fundamental applications. User profiling is the act of building up a profile of who users are and what they want to do. These profiles are used to group visitors/potential customers and hence forth the priorities in their activities are identified. Knowing who the visitors are and what they want is a vital step in meeting their needs.
Web usage mining activity involves following activities:
Web data pre-processing
Pattern discovery and analysis
Web Data Pre-processing
The main task of user profiling is to organize data such as to collect consistent information about the users. Information is bifurcated as explicit and implicit information: the former is derived from the data provided by the user in filling any kind of application form while the latter one is collected from website tracking tools through log files, cookies etc.
Pattern Discovery
The output of web data pre-processing will aid in performing pattern discovery. Pattern discovery activity involves deciphering how the web pages are viewed.
Following patterns potentially emerge:
Association: Identifying groups of pages that are jointly consulted within a set of sessions in navigation.
Sequential: Used to find pages that are accessed in sequence.
Episode: Episode is a subset of page views in a particular session.
Classification/ Segmentation
Data generated and reformatted in pre-processing stage and patterns observed will be useful in making classification rules. In the e-commerce industry, aim of the management is to understand those factors that discriminate the web visitors between buyers and non-buyers.
Web Personalization
The output of web usage mining is web visitor profiling which becomes the input for web personalization. Web personalization involves customizing the web relationship with individual consumers based on understanding their preferences, needs and value. Segmentation is the intermediate step between web usage mining and web personalization.
Types of personalization
Memorization: It is based on memorization of some information about the navigation which is used as soon as the user returns on the web site.
Customization: This type of personalization considers the information filled by the user in the application form and in this respect it takes into account of the privacy and depends on the cooperation of the user.
Personalization models follow two main approaches:
Content-based filtering: This exploits the profiling methodologies to match the user profile and the contents, the products and the services proposed on the Web site.
Collaborative filtering: This is a collaborative approach where the user is invited to express his interest or rate about the visited or downloaded contents. All information is memorized such that the contents are proposed to all users of the same typology.
Finally, the recommendation engine uses all the available information to provide a suitable suggestion for the visit. Two measures are considered to evaluate the performance of the personalization model:
Precision: Computed as the ratio between the "correct" recommended pages and the overall number of recommended pages.
Coverage: Defined as the ratio between the correct recommended pages and the overall number of visited pages, thus measuring the impact of the overall personalization system against its absence.
Conclusion
Web analytics is much more than hits, visits, and page views. The segmented buckets derived from web mining activity helps organizations to target them with specific marketing campaigns and promotions.
E-Commerce Company should follow the AMAT approach:
A: Acquisition of visitors
M: Measurement of performance
A: Analysis of trends
T: Testing to improve
No comments:
Post a Comment