The Ask phase is the first step in the data analysis process. In this phase, we typically define the problem we are trying to solve and understand the expectations of stakeholders. It is important to have an open line of communication to collaborate with stakeholders and take a step back to look at the problem in context. This involves using structured thinking, where we recognize the current problem or situation, organize available information, reveal gaps and opportunities, and identify options. Problems can be categorized into different types.
Types of Problems:
Making Predictions: Use data to make future predictions. Example: Predicting future sales based on historical data.
Categorizing Things: Cluster data based on common features. Example: Grouping customers by purchasing behavior.
Spotting Something Unusual: Identifying data that deviates from the norm. Example: Detecting fraudulent transactions in a financial dataset.
Identifying Themes: Grouping information into broader concepts. Example: Analyzing customer feedback to identify common complaints.
Discovering Connections: Finding similar challenges faced by different entities and combining them. Example: Linking marketing campaign results across different regions to find strategies.
Finding Patterns: Using historical data to understand trends and make predictions. Example: Analyzing seasonal sales patterns to forecast inventory needs.
In the Ask phase, importance is given to asking the right questions. If you are unable to craft effective questions, you might end up with unreliable information.
Examples of Poor Questions:
“Our company is the best at customer service, do you agree?”This is a leading question, as it prompts the respondent to agree, leading to unreliable information. It's also a closed-ended question, which doesn't provide any insights beyond a yes or no answer.
“Do you prefer vanilla or coffee?”
This question is too vague as it lacks context. Are we talking about ice cream or coffee?
Crafting Effective Questions:
Crafting effective questions can be challenging, and one popular technique used is the SMART methodology. SMART stands for Specific, Measurable, Action-oriented, Relevant, and Time-bound.
Specific Question: Simple, significant, and focused on a single topic or a few closely related items. Example: “What specific features do customers most frequently request in our software updates?”
Measurable Question: Questions that can be quantified and assessed. Example: “How many customers reported satisfaction with our new product in the last quarter?”
Action-oriented Question: Questions that encourage change. Example: “What actions can we take to improve our response time to customer inquiries?”
Relevant Question: Important and significant to the problem at hand. Example: “How does the current marketing strategy align with our target demographic's preferences?”
Time-bound Question: Questions that specify a time period. Example: “What are the sales trends for our product over the past six months?”
Using the SMART methodology ensures that your questions are clear, actionable, and relevant to the problem you are trying to solve. This leads to more reliable and useful data for your analysis
By asking questions, we are defining the problem. Defining the problem is very critical, as Albert Einstein said, “If I had one hour to save the world, I would spend 59 minutes defining the problem and one minute resolving it.” Although this might sound extreme, it emphasizes the importance of problem definition. If we proceed with a project without completely defining the problem, we may spend hours resolving it only to realize the problem was not properly defined. This would necessitate going back and redoing the work. To prevent this and save money, time, and resources, it is best to spend sufficient time defining the problem.
This can be done by identifying the problem domain using SMART questions. The problem domain is the specific area of analysis that encompasses every activity affecting the problem and is affected by the problem. This requires structured thinking to recognize the problem, organize available information, reveal gaps and opportunities, and identify options. A popular technique used for this is the Scope of Work (SOW).
Scope of Work (SOW):
SOW is an agreed-upon outline of the work you’re going to perform on a project. It includes deliverables, timeline, milestones, and reports.
Example:
For a data analysis project aimed at improving customer satisfaction, the SOW would include deliverables like customer feedback analysis reports, timelines for each phase of data collection and analysis, milestones such as initial findings presentation, and final recommendations report.
Another important aspect to consider is understanding the context of data, as the data might also have problems. Asking questions such as:
Who collected the data? Example: Was the data collected by a reliable source or an inexperienced intern?
What could have impacted the data? Example: Were there any external factors like holidays or promotions affecting the sales data?
Where was the data collected? Example: Was the data collected from a specific region that may have unique characteristics?
When was the data collected? Example: Was the data collected during a particular season or event that could skew results?
How was the data collected? Example: Were standardized methods used for data collection, or were there inconsistencies?
Why was the data collected? Example: Was the data collected for a specific purpose that could introduce bias?
Asking these questions will reveal the context under which the data was collected and help identify problems in the data. By asking SMART questions and creating a Scope of Work, we can clearly define the problems and goals/objectives of the project. This ensures that the project is well-focused and that the analysis conducted is accurate and relevant, leading to effective and actionable insights.