How to Analyse Marketing Surveys – Part 1

Marketers are faced with a deluge of data to deal with in their daily routine. Quite often, marketers conduct surveys to create a broad picture of the company’s existing market penetration, assess competition and the growth potential for their business, identify problems, opportunities, and existing market metrics.

The quality of insights derived drive business decisions. It becomes imperative for the marketer to understand that simply collecting market intelligence or market survey data is not sufficient but must be coupled with the skill of analysing the data. This will empower the marketer to use valuable data-backed insights to develop and optimize marketing strategies and make key business decisions.

Let’s consider the below survey dataset collected from a market research study by a B2B player and see how we can best derive significant insights from this data. 

This study aimed to survey professionals in the B2C vertical and understand how they are dealing with the post-COVID era to drive growth and also observe behavioural changes in their customers. 

Metadata for this Survey dataset:

1. Respondent ID: is a unique number assigned to each of the 120 respondents of the survey for quick identification or validation purposes
2. Job role: encompasses 7 unique categories representing the Profession or Department of the respondents
3. Have an app: 2 Unique Categories – Yes and No, to indicate whether the respondent’s company has an App
4. Industry: the respondents company’s vertical in which their app is classified
5. App MAUs: provides Monthly Active Users on the App 
6. Number of employees in company: Employee count of the respondents company
7. Revenue percentage for digital marketing: Revenue percentage allocated for digital marketing at the respondent’s company
8. Communication messages sent per week: Frequency of messages sent per week to their users
9. More than 50% of the revenue generated digitally: a flag to indicate whether more than 50% of the revenue is generated digitally
10. Time taken to implement Real Time Customer Insights for improving customer engagement: indicates time taken to generate customer insights
11. What is your company looking to achieve through customer data: This question allows the respondent to provide information about their company’s goal(s) regarding collected customer data. It is a multiple response question which allows the respondent to select one or more than one option
12. Top objectives for your customer engagement strategy: This question allows the respondent to list out their company’s objective(s) for customer engagement. It is a multiple response question which allows the respondent to select one or more than one option

How to Analyse the Survey Data

Data Type

To start with analysing any data, the first requisite is to know the data type of all the variables of the dataset under consideration, which in turn will help us choose the most appropriate way to summarize the variable(s). The data type can be known by looking at the values a particular variable takes or encompasses. 

Basically, there are two types of data possible for any variable — Quantitative Data Type and Qualitative Data Type:

• Quantitative (or Numerical) Data Type: This data type attempts to quantify things by assigning numerical values that make it countable in nature. For instance, the price of a smartphone, number of ratings for a product, revenue percentage, and so on.
• Qualitative (or Categorical) Data Type: This type of data cannot be counted or measured easily using numbers but can be divided into categories. It describes the variable under consideration using a finite set of discrete classes. For instance, gender, industry, job role, and so on.

Univariate Analysis

The next step would be to explore the individual variables of a dataset. This is known as Univariate Analysis, perhaps the simplest form of data analysis where the data is analysed by considering only one (“Uni”) variable (“variate”) at a time. 

Since it involves a single variable, it doesn’t deal with causations or relationships among variables. The major purpose of Univariate analysis is to describe the data by summarizing it and then find patterns that exist within it.

To explore any data based on the data type of the variables in a dataset, you can now proceed to assess the patterns in each of the variables using any of the below several options for describing the univariate data:

• Frequency Distribution Tables: Suitable for numerical data
• Bar Charts: Suitable for categorical data
• Histogram: Suitable for numerical data
• Pie Charts: Suitable for categorical data
• Descriptive Statistics: Measures of central tendency (mean, median, mode) and dispersion (standard deviation, range, quartiles)

Analysing the Data

For our survey dataset, we have 12 variables corresponding to the 12 questions of the survey and 120 observations corresponding to 120 respondents who took the survey. Of these 12 variables, we have 7 numerical variables and 5 categorical variables. 

Let’s explore and grasp the information hidden in this dataset by visually presenting and summarizing insights for a few of the relevant variables one at a time using the most simple and predominantly used graphing method: bar charts.

1. Respondent ID:

It can be used to track how many individuals were part of the survey. It could also be used to identify and eliminate duplicates, if any.   

2. Job Role:

The distribution of the respondent’s profession is reflected in the below plot:

Insight: Roughly 50% of the respondents are from Product & Marketing.

3. Have an App: 

Insight: 93% of the respondents have an app. 

4. Industry:

It is a categorical variable that encompasses 9 unique categories to represent the vertical in which the respondent’s app is classified. 

The bar plot reflecting vertical distribution:

Insight: Ecommerce, Edtech, and Media & Entertainment constitute about 50% of the respondent verticals

5. App MAUs:

App MAUs is a numerical variable. But to enable the response and for easy analysis, it has been broken down into intervals. This changes the variable to a Categorical variable. This data is useful to understand the TAM (Total Addressable Market).

Insights: 

• 49% of the respondents have less than 100 thousand MAUs. This serves an interesting data point to the B2B player to develop the GTM (Go-to-market) strategy
• 6% of the respondents have more than 10 million MAUs        

6. Number of Employees in Company:

It is also a numerical variable but to enable the response, this data is broken into 7 non-overlapping intervals. 

Insights:
This data gives us information about the size of the respondent’s companies. The key insights from this plot are:

• 48% of the respondents have an employee count of less than 100 
• Only 3% of the respondents have an employee count of more than 10,000

7. Revenue Percentage allocated for Digital Marketing:

It is a numerical variable but the data is divided into 10 unique and non-overlapping intervals. 

Insights: 

• 60% of the respondents allocate up to 30% of revenue on digital marketing
• 15% of the respondents allocate more than 50% of revenue on digital marketing

8. Communication messages sent per week:

It is a numerical variable but for ease of respondents in selecting the response, it has been measured using intervals. 

Insights: 

• More than 50% of the respondents prefer the thresholds of sending between 1 to  10 messages per week to customers. That implies that the respondents are not sending a higher number of messages indiscriminately but are maintaining a balance between engagement and spamming.
• 7% of the respondents send 16-20 messages per week.

9. More than 50% of revenue generated digitally:

It is a categorical variable with 2 unique categories, TRUE and FALSE. 

This data gives an idea about the digital penetration within the respondent’s company. 

Insight: 81% of the respondents’ companies generate more than 50% of their revenue digitally. 

10. Time taken to implement Real Time Customer Insights for improving Customer Engagement:

It is a categorical variable of 7 unique categories demonstrating the ability to provide real-time customer insights by their companies. 

Insights: 

• 33% can’t provide real-time customer insights in the same day
• 14% of the respondents can’t implement real-time customer insights

These insights give an idea as to the respondents’ inability to generate real-time insights with the tools available to them. 

Multiple Response Option Variables Analysis 

Some questions do allow the respondent to select multiple options. These options are not mutually exclusive.

The best way to analyse multiple responses variables is by considering each of the selected options as a separate response.

For example: 

User 1 has selected response (a) and response (b)

User 2 has selected response (b) and response (c)

If there are only 2 respondents, response (a) has 50% share, response (b) has 100% while response (c) has 50% share

11. What is Your Company looking to achieve through Customer Data:

It is a Categorical variable that allows the respondent to select one or more option(s) regarding their company’s purpose / focus to achieve via Customer Data. 

Insights:

 Most of the respondents mainly focus on the below 2 aspects:

• An overwhelming 80% of respondents would like to understand customer behavior better so as to engage differently with different groups.
• 65% of the respondents would like to build better customer journeys.

The above insights point towards the need for a better segmentation and engagement tool.

12. Top Objectives for your Customer Engagement Strategy:

It is a categorical variable that allows the respondent to select one or more than one option to represent top objective(s) as strategy for their customer engagement. 

Insights: 

• Retention is the key objective for over 70% of respondents 
• Creating loyal users is the key objective for 60% of respondents

With the Univariate Analysis via Graphs, we have demonstrated how easy it is to describe and explore any given dataset and even make comparisons easy to derive meaningful insights. 

Putting It All Together

The key takeaways from this survey data are: 

• 50% of the respondents cannot generate insights in a day due to limitations with their current tools
  
• Marketers are not looking to spam but to send more personalized and contextual messages and limit the number of overall communications to their customers
  
• Marketers need real-time customer insights to understand customer behavior and craft better customer journeys to drive retention, create loyal users & stickiness, reduce churn, and provide an omnichannel customer experience
  

The Road Ahead

Univariate analysis can be used effectively to derive meaningful insights. Though univariate analysis does give a bird’s eye view of the data and insights, it will be interesting to see how two or more variables are related to each other. 

For example: How does industry/vertical and time taken to implement Real Time Customer Insights for improving Customer Engagement relate to each other? Is the relationship significant?

Stay tuned for Part 2 of this post, where we will conduct Bivariate analysis (analyse 2 variables at a time) to extract more insights.