A Conceptual Framework and Empirical Study
Abstract: With unlimited information available, customers today are highly sensitive to price. An organization can counter this situation of high price sensitivity if it has a large base of highly engaged and loyal customers. Such customers are associated with the organization because they are happy with the services and have an emotional connection with the brand.
Developing a reliable metric of customer engagement is as important as having a loyal base of customers because effective marketing intervention can be made only when the organization knows how engaged the customers are with its brand. Development of this metric has always been a challenge because it involves combining various behaviors of the customer into one scale. This paper discusses the importance of engaging customers, strategies to increase customer engagement and the use of factor analysis to measure customer engagement. Hence, this paper is not just a conceptual discussion but also an empirical study of customer engagement. The developed measure of customer engagement has been empirically tested for an online retailer and has been implemented to design marketing strategies.
In today’s competitive world, revenue and margins are not the only way to study the impact of marketing interventions. Sustained growth of the organization calls for investments in developing a robust customer relationship which is nurtured through an intelligent marketing program. The benefits of these marketing efforts may not be reaped in the short term but definitely have a powerful impact in the long term, which is reflected in a customer’s lifetime value. Customer engagement becomes a strong measure of the effectiveness of a marketing program. Engaged customers are loyal customers who promote the brand image of the business. They trust the quality of products offered and spread the good message around. These customers are less sensitive to price changes because the relationship which they have with the business is based on satisfaction over a period of time. Organizations with a loyal base of customers are less worried about customer retention and are able to focus on improving their services. Their resources are less diverted to resolving customer issues and more towards innovative solutions to customer requirements. According to a 2014 study by Rosetta Consulting, engaged customers spend three times more each year than other customers and are five times more likely to prefer the brand over others.
Given that the cost of acquiring a new customer is ever increasing and is higher than reactivating an existing customer, it becomes imperative for organizations to ensure that existing customers are satisfied and remain associated with the brand.
The recent digital disruption has given enormous opportunity for customers to communicate with organizations and also communicate with each other. A bad customer experience is not just limited to one person, but is spread to all their contacts through social media. Due to this multi-dimensional interaction, an organization must become customer centric. The digital marketplace sees a large number of new entrants every year, which makes it difficult to create a differentiated product. Competition of these types can be checked though an ongoing customer engagement program. Engaged customers are less likely to be swayed away by these competitive threats.
Review: Dr. Pankaj Gupta in ‘Enhancing Organizational Effectiveness Through Customer Engagement’, January 2012 highlights effective methods which online and offline companies are using to reinforce customer engagement. ‘A Generalized Multidimensional Scale for Measuring Customer Engagement’, December 2014 by Shiri D. Vivek, Sharon E. Beatty, Vivek Dalela & Robert M. Morgan, explains the usage of factor analysis to measure engagement. ‘Measuring and Influencing Consumer Engagement, February 2010 uses a combination of Cronbach’s Alpha, Guttman Split-Half Coefficient, Intraclass correlation and factor analysis to develop an engagement scale in healthcare domain. Justina Malciute, August 2012 discusses partial least squares method to estimate engagement in ‘Customer Brand Engagement on Online Social Media Platforms’
Strategies to increase online customer engagement:
Engaging customers does not mean a heavy investment on advertisements to create an all-pervasive presence through all communication channels. It needs a welldefined approach to achieve this objective, as discussed below:
i. Long-term engagement of the customer can be boosted by using the principle of the Four I’s: Interaction, Involvement, Influence and Intimacy. Regular interactions with a customer lead them to get him involved with the brand. This will influence them in making a purchase. If the services are delivered well, this goes a long way in establishing a sound relationship with the brand.
ii. It is very important to integrate various channels of communication to maximize the gains of each contact with the customer. It ensures that the customer is contacted at the right time with relevant content. A communication which is not of interest to the consumer is of no value to the website.
iii. The visit to the website should be a memorable experience to the consumer. The navigation path to reach the desired products should be seamless and hassle free. According to Kate Leggett of Forrester Research, 92% of companies view customer experience as one of their top priorities; 60% use customer experiences as a competitive differentiator.
iv. Communications to the customer should be linked to the customer’s lifecycle, their lifecycle with the business and their lifecycle of product repurchase. This strategy has tremendous opportunity in establishing a robust brand association and relationship. Product recommendations for a customer who is a senior citizen who has been associated with the business for five years and buys a specific product after every three months should be different than those for a 25 year old who became a customer six months back and has made a purchase just a week ago.
v. Social media should be used as means to improvise the products and not just a platform to build a brand image. Data on social media can be utilized to understand latest trends in customer preferences and requirements. This can be leveraged to anticipate the demand and offer it to the customer before the competition does.
Measuring customer engagement:
The importance of having a strong base of loyal customers cannot be negated, and hence it becomes important to have a holistic measure of the same. A challenge faced by all organizations is to define the metrics on which customers’ engagement with its product can be measured. Several parameters can be used to define and measure engagement like activity onsite, frequency of visits and customers referring the website to a friend or colleague. A rulebased approach can be used to define if a customer is engaged or not. For example, if the customer has made a purchase in last four weeks or has browsed on the website more than ten times in last six weeks then they are engaged, or else the customer is not engaged. Rule-based methods are simple to implement but given their nature they cannot be used to rank order the customers. A set of engaged/ disengaged customers may vary in their degree of engagement/disengagement, so marketing interventions must be designed according to the degree of engagement/ disengagement. Statistical methods provide a meaningful solution to this problem. This paper discusses the use of factor analysis to define the engagement score. The score was developed using weekly data for an online retail company wherein the population size was 0.5 million. The results were validated on a predefined hypothesis using a profiling exercise.
Factor analysis was used to measure engagement because it enables combining different dimensions of customer behavior to get a measurable scale of engagement. Factor analysis identifies latent factors summarizing the characteristics of the observed variables. Factors or the latent variables are a linear combination of correlated variables. Different factors are independent of each other. It is important to use only the variables impacting customer engagement because factor analysis gives junk results if junk variables are inputted. The steps outlined below explain the methodology used to define engagement score.
1. All possible variables which reflect customer engagement were identified. Some of these variables were frequency of visit, depth of visit, email open rate, clicks rate, etc. The hypothesis was set on the direction of relationship of the variable with customer engagement.
2. Factor analysis typically groups similar variables into one factor. However, when factor analysis was done on the exhaustive set of variables it showed that dissimilar variables had higher factor loading in the first factor. Since factor analysis was not able to create the factors in desired manner, the exhaustive list of the variables were grouped into different variable categories to manually create factors of different categories of variables as below:
i. Browse Behavior: Gallery page views, product views, etc.
ii. Purchase Behavior: Credit/cash purchase, average value of order placed, etc.
iii. Device Mix: Interaction with the website through laptop, smartphone, tablet, etc.
iv. Channel Mix: Responsiveness through different channels of communication
3. Within each variable category, the objective was to select the most important variable. Hence, factor analysis was done for each factor separately with number of factor=1 because a variable category captures a particular characteristic of the customer. Within a category, a variable was selected on the basis of the conditions below:
i. Variables with higher absolute factor loading than other variables were selected because such variables capture high variability within that variable group
ii. Variables with VIF<5
iii. Variable which made very good intuitive sense were retained even if the variable had low factor loading or high VIF
4. Factor analysis was done on all the selected variables. Variables were selected on the same logic as in the above step. The intent was to have diversity in the selected variables so that the engagement score captures all facets of customer engagement.
5. On the above set of selected variables, a final factor analysis was done to get the factor loading of variables, a linear combination of which explains customer engagement.
6. Factor loading of the variable is the weight for the variable. A sum-product of the factor loading and the value of the variable is the engagement score.
The engagement score derived using the linear combination of the shortlisted variables was tested against hypothesis for the variable. This was done through a profiling exercise to check for the direction of relationship of the variable with the engagement score. The validation results of some hypotheses are presented below.
Hypothesis: More engaged customers drive up the sales. The chart below shows that as engagement score increases, the amount of sales also increases.
Hypothesis: Customers who have placed an order in last 3 months are more engaged than customers who have not placed an order. The table below validates this hypothesis.
Hypothesis: Customers who have placed an order in last 3 months and have not returned the order are more engaged than others. This is validated in the table below.
Hypothesis: Customers who have opened the email are more engaged than customers who did not open the email.
The study shows that factor analysis is able to create a measurable score of customers’ engagement with the online retailer and can give expected results if it is tweaked as per business requirement. This scale can be used to design loyalty programs, reactivation programs and offer optimization strategy.
Organizations can be tempted to make heavy investments in engaging customers, but it is equally important to track this metric against a financial metric which has direct impact on bottom line. The discussed metric of engagement is positively related to sales, which show that higher engagement improves the performance of the business.
The engagement scale discussed here was developed using behavioral data of the customer which was available in-house. A further enhancement to this scale can be done by using data available through social media and survey data on degree capturing connect of the customer with the business. This will help to build a 360° view of customer engagement.
We would like to thank Divya Chowdhary, Chhavi Gupta, Anshul Goyal and the entire team at EXL whose feedback and support helped in making this paper more informative.
- Rosetta Consulting, ‘How Technology Marketers Can Better Engage Customers’
- Magento.com, ‘The Rules (and Tools) for Successful Customer Engagement’