Jump out of the Big data analysis error

"I know everything that's been removed from the shelves, and I know a lot about the consumers who have a membership card,

But when we put a product similar to what consumers buy on the shelves, we don't see the expected revenue growth. "And why?"

Big data, large numbers, is a quantitative message based on the behavior of people on the Internet and social media. With the gradual explosion of such data, the desire of companies, academics and the media to use the data has been stimulated. Among them, the company's executives are fascinated by the customer's activity details (such as who they contact, what they like, etc.) to discover the customer's propensity to buy. Through the computer classification, filtering and modeling, can make the data analysis based on the internet become a reality, but also stimulated the company to practice the idea of large data analysis. There is no doubt that big data will be the next game-changer in the marketing arena, and in this change, experienced companies will gain more.

Since automated software for data mining has become ubiquitous today, many executives think they can easily spot the previously inconspicuous trends and dynamics. But the analysis itself contains much more than knowing a fact, and it requires smart analysts to ask the right questions and make the right decisions; just answer "What's" data analysis is not analysis, you must learn to ask "why", "what is next."

To answer these questions, companies should return to the basics, rather than continue to be mired in misunderstanding, to really use the full potential of big data. The author of 25 years of marketing data analysis in the field of work experience that there are three basic principles has been to achieve effective data analysis of the core. A. Rely on a theoretical approach rather than blindly data mining; B. Maintain a good picture of customers and markets; C. Stick to the middle school.

Theory assumes that the foundation

Without a theory telling analysts how consumers form preferences and act, analysts will soon be flooded with data, and any advanced data-processing power in the world will not help analysts get out of the vast ocean of data. Therefore, the first step in data analysis should be to develop a clear and clear hypothesis on "customer needs" and "how to create value for them". The assumption may be for new products that have a sweeping potential in the laboratory, or for customers who are not loyal to any brand in the market (also known as "undecided voters")--often companies can seize these potential customers by fine-tuning only a few strategies.

Once the required data is obtained to validate the assumptions, the results will guide the company in formulating specific plans to develop, refine, and direct the value proposition to the market. With reasonable assumptions and validation, companies will be able to perform niche market segments (which will classify target customers with similar preferences and behaviors), helping to position the company strategy more effectively.

A drug company, for example, has recently sold poorly, and the company is trying to increase sales of the drug. The manager then assumes that the company's current sales plan is not well targeted at physicians, who are the most likely target groups to use the drug when prescribing.

To verify this hypothesis, the company collects a wealth of data on the circumstances in which doctors will prescribe the drug. For example, physicians prescribe how many prescriptions they prescribe each year, whether their prescriptions are increasing or decreasing, and who are more loyal to their recipes (the company's own or its main competitor). This allows companies to find an optimal location in the marketplace: which physicians prescribe more prescriptions, who prescribe more prescriptions per year, who have no specific preference for pharmaceutical formulations or lack of loyalty.

Although the drug sales performance is not good, the relevant sales staff plummeted, but through the above data analysis, the sales force targeted directly toward the data point of the opportunity strategy, the results are far beyond their imagination.

"A Day in life"

In the course of the development of marketing science, there is an important lesson, that is, "the" is not advisable. When a new data source emerges and becomes available, people flock to it and use it heavily. Smarter companies, however, remain calm, reserved and look at customers and markets in a more holistic perspective. On the one hand, these smart companies will be enthusiastic about mining new data sources, on the other hand, they still attach importance to other information, to complement each other, avoid missing important information, impact analysis.

You know, this is not the first time the marketing sector has faced the subversion of the data revolution, and the transition has been bumpy. In the the mid 1980s, the advent of barcode scanning enabled companies to gather information through the checkout desk. Previously, the data were very limited. The company only knows what goods they deliver and asks at most what products the consumer has purchased. However, with the advent of scanners, companies can actually know every point of sale.

In the early days of the technology invention, the awareness of using the new data led to many mistakes. Executives have become overly concerned about the impact of price promotions on sales, and have lost the basics of marketing: Brand value and brand building. However, over time, the company honed a more complex statistical model, and on this basis to adjust the focus of the focus of energy, the scanner has become the consumer market in the last 30 years and the retail industry, a great weapon of God. Today, the method of collecting data and understanding customers by point of sale has been extended to the membership card, which can help retailers understand what each family buys and what their specific online shopping behavior is.

Just as there are flaws in the investment return model for barcodes in the early years, the latest big data analysis could lead to a misguided path. Many retailers say, "I know everything that's been removed from the shelves, and I know a lot about the consumer who owns the membership, but when we put a product similar to what the consumer buys on the shelves, we don't see the expected revenue growth." ”

What is missing? It is likely that retailers have stumbled into one-dimensional perspective due to the excessive focus on the latest data sources. In fact, to get to the customer's consumption behavior, need a broader vision. In the field of marketing, this broad vision is often called "a day in life." It means: You should have a more comprehensive understanding of your customers, know how customers interact with their companies, and how interactions between customers and other retailers are combined, or how customers interact with their own companies to suit their customers ' behavior in other businesses, shopping channels, and activities. If there is no deep insight into what motivates consumers to go shopping elsewhere rather than their own, the company's growth plan is risky.

Learn to walk before you run

The first step in analyzing new types of data is to learn to open mind and coordinate methods. While digging new data can almost always find exciting new conclusions, companies should not rush into new methods of data analysis with an open mind and gradually challenge practices and rhetoric that have previously been the truth.

In general, new information obtained with new data sources can cause companies to question certain products, services, or strategies. Sometimes it would be counterproductive to believe in mass adoption. Therefore, it is not recommended that companies do their best to analyze and use relevant conclusions in the presence of new data sources; it's best to take a step-by-step approach so that the company learns to walk before running: You can start with a product, a certain region, or a problem, and then evaluate the new data, The new method of data analysis and related improvement of the benefits and costs, to determine whether the new data source analysis is worthwhile.

A global energy giant, for example, has decided to use more advanced analytical methods to address quantitative research issues and improve the return on marketing investment. Top leaders selected two business units from three countries to carry out pilot projects across developed and developing markets. The conceptual framework and objectives for each project are the same, but the operation of European gas stations and the sale of oil in Asia requires different data sets and analytical tools. This diversity allows companies to experiment with more analytical methods and gain enough experience to decide which analysis method to use. In addition, once these pilot projects are successful, confidence can be established for other operational modules and national affiliates to condense the enthusiasm for adopting new data analysis methodologies. With the continuous application of new analytical methods, the model will be more complex and practical, gradually accepted by the company, and finally used in the global scope.

Back to Basics

Many executives are interested in using large data but often have little direct experience with this latest analysis tool and technology. So, in the beginning, they often ask how much the cost of this analysis method is. And the author's answer is generally: "What is the cost of making the wrong decision?" What would be the cost of not reacting to digital photography in time like Kodak? "In a gentle, less direct way, it would be like this:" Data analysis first requires a large amount of money to assemble and coordinate data; then the company needs trained data specialists to carry out higher-profile work to find patterns behind the data, to interpret them, and eventually to turn into ideas and insights that the company can use. " ”

However, as indicated in the previous three basic principles, the adoption of a new data analysis approach is a manageable process, with proper handling that can yield significant potential benefits. In fact, when most companies start investing in data analysis, they are basically not going to stop, as the results of the data analysis are far more costly and effort-enhancing than the data analysis. It can be seen that data analysis has become an important way for companies to improve their market position.