Customer Analytics – Effective Data Collection Methods

Surveys have proven to be an effective method of collecting information about customer sentiments, purchasing habits and product preferences. There are two ways that customer feedback is collected, either after the product has been purchased or during the purchase process. Online tools such as Survey Monkey enable creation of Free or advanced surveys to collect and analyze data. While not all customers might be willing to complete surveys after product purchase, it is more effective to collect information during the purchase process.

What are the pros and cons of using surveys for data collection?


  • Free or low cost
  • Relatively easy to create and modify
  • Effective method of learning about potential customers


  • Hard to write non-biased questions
  • Hard to determine the correct respondents

What kind of questions to ask?

Categorizing the responses to get the satisfaction rating to enable selection of answers from very satisfied, quite satisfied, neither satisfied nor dissatisfied, quite dissatisfied or very dissatisfied. The only downside to this approach is that there is no way to compare responses.

Comparing the responses to determine product quality compared to other products or competitor products such as very inferior, neither inferior nor superior, or very superior. The only downside to this approach is that there is a big loss of information as there is no way to rank the responses.

Ranking the responses to determine how important are certain features of a product such as the most important to the least important. The only downside is that it provides many comparisons and ranking types. It is recommended to not exceed 8 rankings.

Paring the responses such as two products for comparison to determine which one they like while displaying few information about each product will force the customer to select one or the other. The only downside is that customers might dislike both products.

The Rensis Likert Scale named after its inventor is a useful method if you want people to think about lots of different aspects of a product and giving them the option to chose from a ranking of Definitely Agree, Generally Agree, Slightly Agree, Slightly Disagree, Generally Disagree, or Definitely Disagree.

The Continuous Scale is also a useful method to give people a scale to select from “Do not prefer” to “Strongly prefer”.

How to check for validity and reliability of questions?

Each type of question has pros and cons so think about what the end goals are for each product.

Predictive validity – such as measuring customer satisfaction to see if they will refer your product to other customers

Test and retest reliability such as how stable is what you are collecting? Does it vary a lot? If it does, then it is not stable. The volatility does not change.

How to determine the overall customer satisfaction rating?

Using Net Promoter Score companies can determine customer satisfaction rating related to profitability. The higher the satisfaction rating leads to more positive outcomes.
Questions based on the likely hood to promote the product or company and giving the option to select from scale of 1 to 10.

Respondents fall into three categories:

Promotors select responses from score 9 to 10.
Passives select responses from score of 7 to 8.
Detractors select responses from score of 0 to 6.

To determine the Net Promoter Score (NPS):

NPS = % of Promotors – % of Detractors

Other ways to collect information from customers?

Mining for Big Data on the internet, online product reviews and words of mouth on social media sites to determine what customers say about the product is also another method of data collection. This is a big topic on its own that we will discuss in another article.

How to design an effective survey?

Survey Respondent Population Selection

Select the appropriate population representative sample that have knowledge and understand the questions. Provide incentives to motivate them in answering the questions. Typically, the response rate is 5%, so collect data about the non-respondents by testing the differences between those whom respond and those that do not respond. Follow-up with non-respondents to get them to respond.

Determine Question Formatting

Open-ended response formatting are questions that start with “Why”. Respondents will provide general reaction to questions. Responses are given in real world terminology. It can also help interpret closed-ended questions. The problem with open-ended questions is that they are not good for self-administered surveys.

Closed-ended response formatting are questions that start with “How”. Respondents will be provided with pre-determined descriptions and selects one or more of them. They are easy to fill out, easy to code and cheaper to administer.

Sequence and Layout of Questions

Randomize the order of the questions so that they are easy to do. Start by asking easy questions without asking about income or personal information. Make the flow of questions smooth and logical. Go from General to specific questions.

Test the Questions and correct problems

Check to ensure that there is variation in questions by the type of questions selected. Ensure that the questions have meaning and keep respondent’s interest.

[vntd_button label=”Learn more about Data Science” align=”center” url=”|title:Data%20Science||”]