Value of Social Media Marketing

Social Media Data

A marketing manager from the firm has selected scanner panel data from a two-year period for this analysis. In the first year (weeks 1-54), the firm did not have a social media presence, while in the second year (weeks 55-108), the firm maintained its social media presence. The analysis assumes that the event — consumer participation in social media — takes place in week 55 for all consumers in the treatment group. The manager randomly selects 200 consumers from each group and observes their purchase behaviors.

Total observations: 14030

Total weeks: 108

Total customers: 400

Data Dictionary

Variable	Description
ConsumerID	Identification code for consumer
Group	$Group=\begin{cases} 1, & \text{if $ConsumerID \in TreatmentGroup$}.\\ 0, & \text{otherwise}. \end{cases}$
Period	$Period=\begin{cases}1, & \text{if $Week \geq 55$}.\\ 0, & \text{otherwise}\end{cases}$
Week	Purchase occasion of consumers at weekly level
Spending	Dollar spending by individual consumers during each purchase occasion

Task: Capturing Value of Social Media Marketing

Determining the effectiveness of social media marketing requires moving beyond simple metrics like likes and shares to examine tangible business outcomes. A comprehensive evaluation must consider both the immediate financial impact and the longer-term relationship dynamics between the firm and its customers.

Question: Analyze whether the firm’s social media marketing is working.

You can approach this question by analyzing the customer value generated along two dimensions

Transactional Value: Assess changes in immediate purchase behaviors, such as purchase frequency, average transaction size, and total spending.
Relational Value: Evaluate shifts in customer loyalty and engagement, such as retention rates, purchase consistency, and customer lifetime value.

Solution

We can capture the customer value generated along two dimensions

Transactional: using Customer Spending
Relational: using Customer Visit Frequency

Transactional Value: using Customer Spending

Relational Value: using Customer Visit Frequency

Regression Analysis

\[ CustAvgSpend = \beta_{0} + \beta_{1} Group + \beta_{2} Period + \beta_{3} Group \times Period + \epsilon \]

DiD Regression: Customer Spending
term	estimate	std.error	statistic	p.value
(Intercept)	34.89	0.32	107.50	0.00
Group	0.59	0.46	1.29	0.20
Period	-0.29	0.46	-0.62	0.53
Group:Period	15.43	0.65	23.75	0.00

\[ CustVisitFreq = \alpha_{0} + \alpha_{1} Group + \alpha_{2} Period + \alpha_{3} Group \times Period + \nu \]

DiD Regression: Customer Visit Frequency
term	estimate	std.error	statistic	p.value
(Intercept)	17.18	0.61	28.34	0.00
Group	-1.03	0.86	-1.20	0.23
Period	0.54	0.86	0.63	0.53
Group:Period	2.83	1.21	2.33	0.02

Breaking Down Regression

\[ CustAvgSpend = \beta_{0} + \beta_{1} Group + \beta_{2} Period + \beta_{3} Group \times Period + \epsilon \]

Group	After	Before	Difference
Treatment	$\beta_{0} + \beta_{1} + \beta_{2} + \beta_{3}$	$\beta_{0} + \beta_{1}$	$\beta_{2} + \beta_{3}$
Control	$\beta_{0} + \beta_{2}$	$\beta_{0}$	$\beta_{2}$
Overall			$\beta_{3}$

Group	After	Before	Difference
Treatment	\(\beta_{0} + \beta_{1} + \beta_{2} + \beta_{3}\)	\(\beta_{0} + \beta_{1}\)	\(\beta_{2} + \beta_{3}\)
Control	\(\beta_{0} + \beta_{2}\)	\(\beta_{0}\)	\(\beta_{2}\)
Overall			\(\beta_{3}\)