How to use big data thinking to conduct user research

Traditional product research, usually need to first select the user samples, and then spend a lot of manpower and resources to use different research methods to conduct user research. If you apply big data to user research, with a huge amount of historical data samples, for the research problem, you can use big data to pre-analyze the processing, and then manual selective intervention processing, not only to improve the efficiency of user research, the fastest response to user needs, but also can greatly reduce the cost of user research. Based on this, this paper attempts to use big data thinking to interpret the new changes in user research in the era of big data.

Description: This article provides only the idea of user research in the era of big data. If there is a specific user research needs, welcome to the author, the author will be in the next tweet, the specific case of the discussion.

Big data, as a means of production, is affecting human society more and more y. Now, big data in the field of e-commerce, through the similar consumer preferences for goods, to recommend to customers more in line with their personal preferences, this recommendation not only saves consumers the time to look for goods, but also improve the income of the e-commerce platform.

Similarly, in the fields of music, TV series, movies, advertising, and user research, big data is becoming more and more useful. So what changes has the era of big data brought to the way of user research?

Before big data was widely used, the traditional way of user research usually needed to go through the five steps of defining the research problem, formulating the research plan, synthesizing the research method, designing the research questionnaire, and summarizing the research results.

But after big data is widely used, with the massive historical data samples, the research problem can be pre-analyzed and processed with the help of a variety of publicly available big data tools, and then manually selectively intervene in the process, compare the two, and carry out multiple rounds of TEST, to help the product personnel to discover the truth of the problem.

One, set up a good research question, research will be half successful

Set up a research question, in the first part of the whole research, the importance of its natural self-evident. For example, some product managers may ask, "Why don't users pay for video?" or "Are there enough users who are willing to pay $15/month to watch a genuine HD video, if it's a lower or higher price? The former research question is too broad, while the latter is too singularly defined.

If the research question is defined as:

Which type of users are most likely to use the paid services of video websites? How many users would be willing to pay for the different price levels? Of all the video sites, how many users would choose that video site because of this service? What is the value of this approach as opposed to paying for video, such as advertiser sponsorship?

Of course, not all research is as specific as this:

Some is exploratory, where the goal is to get to the bottom of a problem and come up with possible answers or new ideas;

some is descriptive, where the focus is on describing some quantitative characteristic of the project's content;

and some is causal. The purpose of this kind of research is to detect whether there is a causal relationship between phenomena.

Two, according to the research question, the big data pre-analysis processing

The charm of big data is that the collection is not sample data, but all the data. For example, DDT launched DDT takeaway service, Meituan launched Meituan taxi business, thanks to the development of modern social networks, DDT and Meituan can almost statistical analysis of microblogging, weibo, wechat, and other social media for the newly launched services, so as to provide better services.

For example, the Baidu index can be used to understand the search behavior of netizens for this service, and at the same time tracking and analysis:

Of course, not all netizens will use the Baidu search, but they may also use 360 search, which is necessary to rely on the 360 index:

And or the user to take other ways to express the emotions and ideas, such as social media microblogging

Or perhaps users take other ways to express their emotions and thoughts, such as social media microblogging, WeChat, and may use the Weibo Index, a third-party public opinion monitoring and word-of-mouth analysis tool, and word-of-mouth analysis and text mining with the help of Sina Weibo:

Description: The above big data tools are only a list of 3 commonly used ones. In practice, the choice of big data tools also needs to be determined based on the user's specific research questions.

Three, artificial intervention, research issues for targeted processing

Can be based on big data analysis results, artificial intervention to the research problem, targeted research processing, this time you can use the traditional research methods. But unlike in the past, when using these research methods, do not need to spend a lot of cost for all kinds of research. The purpose of the choice of manual intervention is to more real feel the research process, involved in research issues to deal with up.

Traditional research methods usually have the following four ways:

1. Observation method

This method is to take an unobtrusive way to observe consumers using the product, in order to collect the latest data. Some strategy consulting firms are big believers in the observational approach to research.

The following is a clip from a well-known marketing consulting firm in China, Hua & Hua, on the use of this method in "Super Symbol is Super Creative", to understand:

"For example, if you are observing toothpaste consumption in a supermarket, and observing the people who are walking to the toothpaste aisle, you will see a process like this: a customer pushes a shopping cart and walks over, browsing the toothpaste aisle as he walks. Walking while browsing the shelves of toothpaste; stop, focus on a box of toothpaste for a moment, continue to walk forward; stop, pick up a box of toothpaste, after looking at put down; and pick up a box to look at, and then turn it over, take a closer look at the packaging, the text behind the shelf back; two steps forward, turn around and go back to the very beginning of the box of toothpaste, take a closer look at the text behind the packaging, put it back on the shelf; walk back quickly, the fourth step to look at the box of toothpaste still in the shopping cart, end of selection."

"No, it's not over; he may fold back after a while, put the toothpaste he just put in the cart back on the shelf, and replace it with the box he watched in step 2, or maybe both. This way you observe the entire process of his toothpaste purchase, which turns out to be seven actions."

2. Focus Group Interviews

This is a type of research in which six to ten people are carefully recruited on the basis of demographic characteristics, psycho-statistical characteristics, and other considerations, and then brought together for a discussion with those participants over a specified period of time, for which the participants usually receive some payment.

The researcher usually sits next door to the panel, in a room with a one-way mirror, and observes the panel discussion. It is important to note that real-time focus group interviews must be conducted in a way that makes the atmosphere as relaxed as possible for the participants and seeks to get them to tell the truth.

3. Behavioral data analysis

The behavior of the user in the use of the product can be used to observe the user's psychology, the researchers through the analysis of these data, you can learn a lot about the user's situation.

The user's browsing time and browsing content can reflect the user's actual preferences, which is more reliable than the user's verbal statements provided to the researchers.

4. Experimental method

The real cause-and-effect relationship between phenomena is obtained by eliminating all the factors that may affect the observation.

For example, a video website, which provides high-definition video service to its users, charges only $25 per month in the first quarter and $15 per month in the second quarter. If there is no difference in the number of users who use the service for the two different price charges, then the video website cannot conclude that the higher service charge significantly affects users' willingness to watch paid videos.

Four, after the research methodology is determined, you can start the design of the research questionnaire

Setting up the questionnaire is to collect first-hand information. However, because the questionnaire in the questionnaire format, order and the order of the questionnaire affect the questionnaire to fill in the effect of the questionnaire, so the questionnaire in the questionnaire to test and adjust the questionnaire is very necessary.

Precautions for questionnaire design:

Fifth, summarize the results of the research

The results obtained from the statistical pre-analysis of big data, with the results of the actual research of product researchers, to compare, so that the data and information will be converted into findings and recommendations.

Finally, the job is done, and specific marketing decisions can be made based on the results of the market research.

Description: Because of the use of traditional research methods in this process, there is no need to spend a lot of manpower and resources, and for suspicious results, multiple rounds of TEST can be conducted by controlling variables to help product personnel really find the truth of the research problem.