5 steps to help you avoid confusion when analyzing data, starting today

"Why doesn't analysis lead to results?"

Most marketers today analyze large amounts of data on a daily basis. However, many of them feel that they are not actually analyzing the data, saying things like, "I create reports, but they don't lead to any concrete actions," or "I think I'm analyzing, but I end up just reporting the facts."

The root cause of these problems may not lie in analytical methods, tools, or skills, but in the order and attitude of thought, in other words, the way in which one "interacts with data."

Therefore, in this article, we have divided the important "how to deal with data" prerequisite for analysis into five steps. No matter how excellent the analytical method or tool, if the foundation of the analysis is not well-established and well-structured, the results will not be fully utilized. However, the five steps we will introduce below are simple and essential, and can be put into practice immediately by anyone in the marketing field. By keeping these steps in mind, you will acquire the ability to conduct analysis that leads to decision-making and improvements.

How to deal with data that is the foundation of analysis

Below are five steps to transform your analysis, which tends to end with you just thinking you did it, into something you can actually use. By thinking through this process in an orderly manner, you will be able to clear up any vague doubts you may have had about your analysis and conduct analysis that leads to results.

• Step 1: Arrange the data
• Step 2: Put your discomfort or realization into words
• Step 3: Check the background
• Step 4: Formulate a hypothesis
• Step 5: Select an appropriate method for your hypothesis and verify it

This time, we'll use the recently concluded Osaka-Kansai Expo as an example to show how to implement the five steps. This event was held for a limited period of time, so the start and end of the data are clear, making it easy to review and verify the whole picture. Furthermore, there is data available that shows consumer interest and behavior, such as "search volume (interest) on Google Trends" and "actual ticket sales (behavior)," and comparing these data is likely to yield some interesting discoveries.

If you find yourself in a position in the future where you are responsible for planning and executing a large-scale event or other marketing initiative, understanding what attracts people's interest and how to get them to take action will be a great help when designing initiatives and considering strategies for disseminating information.

Things to keep in mind before you start your analysis

Before proceeding with the five steps we'll introduce below, it's important to first clarify your purpose for conducting the analysis: "Why are you doing the analysis?" It's important to make the big goal of your company or organization (for example, selling millions of tickets) your own, and then break it down into specific objectives for how you will use the data to achieve that goal. If your purpose remains unclear, you'll likely just be looking at the data aimlessly, or be overwhelmed by analytical methods and tools, which will often not lead to actual results.

It is also essential to clarify the issues that need to be resolved in order to achieve your objectives. Once you have identified the issues, you will know which data to look at and from what perspective you should proceed with the analysis.

For example, the following can be considered for the Osaka-Kansai Expo.

[Objective] Increase ticket sales by XX% in September and October to achieve the goal of selling XX million tickets.

[Issue] Despite growing interest, there have been periods when ticket sales have been sluggish, and the reasons for this have not been identified.

By clarifying the purpose and issues in this way, the way you view the data and the direction of your analysis will become clearer, and the quality of the insights you obtain will naturally improve.

Now, let's look at the five steps to actually proceed with the analysis. The first thing you should do is organize the data related to the set purpose and issue, and "view it side by side."

Step 1: Arrange the data

The first step in analysis is to sort through the data you have and take a good look at it. First, organize multiple pieces of data in line with your objectives and issues by time series and channel, and then compare them.

For example, let's compare the number of tickets sold for the Osaka-Kansai Expo with the Google Trends search volume for the general word "Expo" as shown below.

Step 2: Put your discomfort or realization into words

When you look at all the data side by side, there will be moments when you think, "Huh?" It's important not to miss any "discomforts" or "changes" such as a drop in results, an increase in interest, or a sudden change in the numbers. Also, be sure to "verbalize" any points that catch your attention so you can share them with someone. By verbalizing things, the team will be on the same page, and the next decision will be faster and more accurate.

In the case of the Osaka-Kansai Expo, we compared the number of tickets sold and the search volume on Google Trends over time, and discovered the following:

① There are periods when the number of tickets sold is clearly increasing rapidly, but the search volume remains flat.

②There are times when both ticket sales and search volume increase sharply.

3) There are periods when ticket sales have not increased compared to search volume.

4. There are times when search volume is on a downward trend, but sales suddenly increase.

These discrepancies are immediately noticeable just by looking at the graph. At this stage, there is no need to come up with a clear answer. Rather, the starting point for analysis is to ask yourself, "Why?" In the next step, we will use this sense of incongruity to explore "what was behind it?" 

Step 3: Check the background

If you notice something odd or a change that concerns you, try to investigate the background. By exploring the "context" behind the data, you can uncover a story that cannot be seen from the numbers alone.

For example, in the case of ① "the number of tickets sold has clearly increased sharply, but the search volume has remained flat," to find out the reason for this, we will look into the events that occurred around the time of the sudden increase in ticket sales. In fact, when we look into related news, we find that the sale of super early bird one-day tickets ended on October 6, 2024. This information suggests that the increase may have been due to a "last-minute rush of demand just before the sale ended."

In this way, for other anomalies, you can gather material to support your hypothesis by researching the news, events, and social media topics that occurred before and after. The important thing here is not to seek the correct answer, but to simply gather material to form the foundation of your hypothesis, and imagine that you are building a hypothesis by collecting hints and pieces.

① There are periods when the number of tickets sold is clearly increasing rapidly, but the search volume remains flat.
[Background] Super early bird one-day tickets will be sold out on October 6, 2024 → This is likely due to last-minute demand

②There are times when both ticket sales and search volume increase sharply.
[Background] The Osaka-Kansai Expo will open on April 13, 2025. → This suggests that people are interested in going to the expo because it has become a hot topic.

3) There are periods when ticket sales have not increased compared to search volume.
[Background] Detection of methane gas and mass outbreak of midges → (Multiple factors involved, making it difficult to explain)

4. There are times when search volume is on a downward trend, but sales suddenly increase.
[Background] New TV commercials will begin airing on June 14, 2025, coinciding with summer vacation and school trip seasons → (Multiple factors are involved, making it difficult to explain)

On the other hand, in cases ③ and ④, it has become clear that it is difficult to explain with just one piece of background information, and that there is a high possibility that multiple factors are involved. In such cases, it is necessary to formulate several hypotheses based on the background information and verify each possibility.

Step 4: Formulate a hypothesis

The next step is to formulate a hypothesis based on the information and insights gained.

A hypothesis is your own idea of ​​"why a certain phenomenon occurred," and it serves as the basis for your next action or verification. If you start your analysis with an unclear hypothesis, you will often find yourself overwhelmed by methods and tools, and wondering, "What exactly did you want to know?"

For example, in ③④, the following hypotheses can be considered:

3) There are periods when ticket sales have not increased compared to search volume.
[Hypothesis] Negative reports about methane gas and midges attract interest, but do they have a negative effect on purchasing?

4. There are times when search volume is on a downward trend, but sales suddenly increase.
[Hypothesis] The effect of the TV commercial encouraged the behavior of those who were "aware" or "interested," leading to a purchase without going through a search?

The next step is to verify the validity of the hypotheses formulated here using data.

Step 5: Select an appropriate method for your hypothesis and verify it

It is at this stage that the question finally arises as to how to analyze the data.

When it comes to analysis, we tend to think of "choosing a method first," but in reality, it is far more practical and efficient to think of it in the order of "choosing an appropriate method for your hypothesis."

For example, in response to hypothesis 3, "Will negative news coverage of things like methane gas and midges attract interest but have a negative effect on purchases?", a simple test such as segmenting Google search trends can provide sufficient insight. Specifically, we obtain and compare the search volumes for negative words related to the Osaka-Kansai Expo, such as methane gas and midges, and compare the periods when these topics attracted attention with the number of tickets sold and the overall trends in search volume. Even with such a simple analysis, it is possible to verify the validity of the hypothesis to some extent.

However, caution is required when multiple "different" factors overlap, as in ④. Because different events of different natures are occurring simultaneously, such as the airing of a new TV commercial and the overlap with the summer vacation/school trip season, it is difficult to simply determine which factor had the greatest influence.

In a situation where multiple different factors overlap, if you want to understand the impact of each factor numerically, you can use the "Multiple regression analysis" is one effective method. Multiple regression analysis is a type of statistical method that clarifies the relationship between a target variable that is an outcome (for example, the number of tickets sold) and multiple explanatory variables that are factors that affect the outcome (for example, the presence or absence of TV commercials, the summer vacation/school trip season), and because it can be handled using tools such as Excel, it is a relatively easy method to get started with.

Based on this multiple regression analysis, the "MMM (Marketing Mix Modeling)MMM models not only marketing strategies such as TV commercials and online ads, but also external factors such as demand, seasonality, and competitor strategies, making it possible to quantitatively visualize the impact and effect of each.

For example, by using MMM, you can break down the degree of influence of each factor into numbers and determine whether the significant increase in ticket sales was due to the influence of TV commercials, the effect of summer vacation or school trip season, or actually due to excitement on social media. This can be used to verify the effectiveness of measures and to allocate budgets next time.

Some people may wonder, "What is MMM?" or think it would be difficult to use right away, but by starting with multiple regression analysis, you can acquire the ability to interpret data relationships. The following materials introduce the concept of multiple regression analysis and an analysis method that you can learn by actually doing it in Excel.

Click here for the material version of this article

A guide to multiple regression analysis in Excel that empowers marketers
~ Understand the correlation between marketing measures and business results ~

Download Now

Additionally, the following white paper summarizes everything from the basics of MMM to examples of its use in actual business settings. If you would like to learn more about MMM or are considering introducing it in the future, please make use of this as well. 

Click here for the material version of this article

What is "MMM" that all marketers should know about?
~ Benefits of Implementation: Insights from Three Company Case Studies ~

Download Now

From data to action, designing your next move based on analysis results

The five steps we've introduced so far are designed to help you develop a better way of interacting with data, allowing you to derive important insights and hypotheses from the data.

However, this alone will not lead to results.So, finally, we will introduce the first step in the practical phase, which is to link the hypotheses and suggestions obtained to measures and decision-making, and turn the analysis into "action" rather than just "thinking about it."

For example, suppose that after verifying ④ with MMM, the following was discovered: (Note: The following is a reference image and is not an actual analysis result.)

④ There are times when the number of tickets sold suddenly exceeds the search volume.

[Hypothesis] The effect of the TV commercial encouraged the behavior of those who were "aware" or "interested," leading to a purchase without going through a search?

[Results] The increase in ticket sales was most significantly influenced by television commercials (approximately 40%), followed by seasonal factors such as summer vacation and school trips (approximately 20%). Social media also had a small influence (approximately 10%).

Based on these results, the following measures can be designed as the "next step." In doing so, it is effective to organize the identified options along two axes: "impact (magnitude of effect)" and "feasibility (possibility of implementation and cost)," and then proceed by prioritizing them.

• Strategically review the timing of TV commercial broadcasts (e.g., a specific period before the summer holidays), the amount of advertising, and the content of the creative.
• Clarify the next actions that people are likely to take after being exposed to a TV commercial (e.g., searching for related keywords or visiting a website), and strengthen collaboration with online advertising and websites.
• Consider investing in channels other than search (e.g., social media ads and influencer campaigns) for those who purchase without using search.

In this way, the ultimate goal of analysis is not to determine "why," but to clarify "what should be done" derived from that "why" and incorporate it into the next strategy or action plan. Only by utilizing the suggestions gained from the data and linking them to concrete action can analysis contribute to the success of a business or project.

In conclusion

Analysis is often thought of as a matter of method and skill, but in reality, the quality of the analysis depends on whether you can approach the data in a proper order. Furthermore, while methods evolve with the times and tools, the fundamentals of "order of thought" and "hypothesis-based analysis design" remain the same.

The five steps we've introduced here are simple yet essential ways of thinking that can be put into practice even without advanced knowledge. Whenever you feel a sense of discomfort, such as "I can see the numbers, but I can't put them into action," stop and keep these steps in mind as you work. "Before choosing a method, get your 'way of thinking' in order." This basic approach will transform your marketing from "vague" to "convincing decision-making."

If you have concerns like "I'm analyzing, but it's not leading to the next step" or "I want to use data more strategically," please give this a try.Please contact XICA.

For over 10 years, XICA has worked with over 280 companies on analytics that support marketing decision-making. By utilizing various data sciences, including MMM, we can derive actionable strategies even in complex environments.

We also accompany our clients through the entire process, starting with clarifying the business issues they face, asking "What questions should we ask in the first place?", and guiding them through how to incorporate analysis into their decisions and how to make it take root in their organizations. Through "usable analysis," we improve our clients' decision-making capabilities and provide support that directly leads to business results.

XICA Co., Ltd. Marketing Team