When big data encounters unfathomable human nature

About the magic of Big data marketing, there is a story in the big data that the term has not been red through the Internet marketing area has been talked about.

The story goes like this:

At the beginning of 2012, a man stormed into a target supermarket in the outskirts of Minnesota State, singling: Why did the supermarket send his or her high school daughter a discount coupon for baby diaper samples and formula milk? "Are you encouraging her to be pregnant?" Angry father questioned Target supermarket manager. A few days later, the supermarket manager called to apologize to the father, and the father's tone became peaceful, and he apologized in return for the fact that his daughter was pregnant and was due in August. This is the story of how a retailer applies large data to marketing, a story that has been reported by the New York Times, and the power of big data has been sensational all over the United States.

Why can target make such a miraculous prediction? This is because Target has established a very standard large data management system, it has a data analysis team, after viewing the consumer records of expectant mothers, identified more than 20 kinds of associations, through these related things to the customer "pregnancy trend" forecast, and send the corresponding coupons, to fuel consumption. Whenever possible, Target's large data system will give each customer an ID number. You swipe your credit card, use coupons, fill out questionnaires, mail return orders, call Customer service calls, open AD mail, visit the website, all of which will be recorded in your ID number. And this ID number will also be taken with your demographic information: age, whether you are married, have children, live in a city, address from Target, salary, recent move home, credit card in your wallet, frequently visited URLs, etc. Target can also purchase your other information from other relevant agencies: race, employment history, favorite magazines, bankruptcy records, marital history, home purchase records, school records, reading habits, etc.

In the traditional marketing era, Lasswell mode is the most basic mode of advertising communication mode, it is 1948 by American political scientist, psychologist Harold · A representative linear model proposed by D. Lasswell also referred to as "5W mode", the advertising information dissemination process contains the five key elements are: Who (WHO), what to say (dour what), through what channel (in abound Channel), for whom (to whom), what effect achieved (with what multiplying).

Who refers to the dissemination of the subject, that is, the brand and the product itself, even in the traditional marketing era is a difficult job. In fact, in the traditional marketing era, the most difficult to solve is the dissemination of who (to whom), through what channel (in abound Channel), achieved what effect (with what multiplying) three major issues, so in the traditional marketing era, Brand managers often question the question of their agents: "I know that half of my advertising costs are wasted, the problem is I don't know which half," and from Target's point of view, large data can be very accurate to lock in and even predict the next consumer behavior, This makes the brand manager in the traditional marketing times confusing users ' needs with a sharp weapon.

With the advent of the big data age, through what channels (in abound Channel), who (to whom), and what effect (with what multiplying), these three points in the traditional marketing era can not solve problems, seems to be solved, Because the essence of large data marketing is to affect the target consumers before the psychological path of shopping, it is mainly used in three aspects: 1, large data channel optimization, 2, precision marketing information push, 3, online and offline marketing connection. Under the influence of large data marketing, one of the most difficult problems in marketing is how to accurately predict the needs of target users and provide solutions, which is the value of large data marketing, so the 2014 domestic DSP platform developed in full swing.

One, to whom? --large data can pinpoint target audiences. Traditional marketing mostly by demographic characteristics to summarize the target consumers, such as consumption habits, psychological characteristics, hobbies such as the depth of data need to rely on professional market research companies, and with the help of large data technology, marketers can infinitely close, almost accurate judge of each person's attributes. Some enterprises through the collection of massive consumer information, and then using large data modeling technology, according to consumer attributes (such as the region, gender) and interest, purchase behavior and other dimensions, mining target consumers, and then the classification of the label layer, based on these to the individual consumer marketing information Push, If the mobile DSP crowd according to 3 kinds of label layer division, analyzes the user's personalized demand, by this provides the personalized product and the service, or realizes the more accurate advertisement:

1, the property label layer. These properties can be from the population attributes (gender, age, occupation, income, etc.), equipment properties (equipment prices, equipment systems, equipment models, etc.), operator attributes (China Mobile, Unicom, Chinese telecom, etc.), urban attributes (development level, population number, regional location), business district attributes (function, location, etc.) Some of the main attributes of the label, the number of attribute labels and a platform of the technology and experience has a direct relationship, the more mature technology, the more accurate capture of attributes, the more experienced, the classification of attributes more reasonable.

2, the behavior of the label layer. Refers to the label layer generated by the user's analysis of the landing page or the app's behavior during a specific time period or location. Behavior Tag layer classification based on the frequency of the occurrence of the statistics to make a label, if the user's behavior is only a few times within a period of time, and will not be listed as a label, only the occurrence of the behavior of a regular frequency or cycle will be considered as a label. For example, often play mobile phone games, often using tourism software, such as subdivision of business travel crowd, hand visitors group, financial people, car family, cosmetics audience, education audience and so on. Because of the variety of user behavior, this kind of label will have tens of thousands of ads, the accuracy of the delivery is undoubtedly a big advantage.

3, target crowd level. This is the most directly related level of advertising, the target population layer is mainly based on the attribute tag layer and the behavior tag layer after the combination of the label layer, this combination will produce a large label, while a user is labeled with a number of tags will become a comprehensive label body, but also to ensure the accuracy of the target population, For example, an advertiser needs to locate in the 35-year-old male car audience, you can pass the first level of age, gender plus the second level of the car browsing behavior combination to get the target crowd, so that the brand with the most relevant people, so the ads targeted very strong, excellent results.

Ii. in abound Channel? Large data can optimize the transmission channel in real time. For public media resources, in order to cover as many audiences as possible to create contact opportunities with the audience, advertisers often need to spread across the media. But how is the budget allocated? Large data at this time is the best decision reference, based on mass user data, in the distribution of marketing channel allocation on the adjustment, get the best portfolio.

Online real-time optimization channel: by inserting cookies in different Web pages, according to the user's Internet traces of channel marketing optimization, is based on the Internet customer's behavior trajectory to find out which marketing channels of the most customers, which source customers actually buy the most, whether the target customers and so on, In order to adjust the marketing resources in various channels of delivery. Dongfeng Nissan, for example, uses tracking of customer sources to improve marketing resources in various network channels such as portals, search and microblogging.

Online real-time feedback: such as fast fashion brand Zara in the store layout of a number of cameras, such as customers in the store to respond to shop assistants "I do not like this skirt zipper", immediately through the manager summary to the headquarters of the design staff, after the summary and analysis, according to customer demand to make product optimization immediately.

Online offline collaborative implementation of the effect of closed loop: Even some enterprises will be the internet Shanghai volume of consumer behavior traces of data and offline purchase data through to achieve online and offline marketing channels of Synergy. For example, Dongfeng Nissan, online and offline collaborative marketing approach: Its portal brings order clues, and through these clues, the service personnel to make a return call, so as to promote the customer on-line transactions. In this process, Dongfeng Nissan recorded the consumer enter, browse, click, register, telephone return visit and purchase all aspects of the data, to achieve a line across lines, with large data analysis for support, the marketing effect of the continuous optimization of the closed-loop marketing channel.

Third, with what multiplying? Large data can be real-time feedback effect, large data is a real-time analysis engine. According to the actual data during the launch process, such as audience behavior, flow composition and other real-time data, to find the target audience focused on the point of time, to find the best response to the creative version of the audience, to determine the depth of users, mining new potential consumers and so on, the advertising timely judgments and adjustments, and the above process is dynamic, Real time.

The ideal large data age enables accurate evaluation of each activity through the processing of data, in the past, the fear of "I don't know which part of the waste" was feared by the brand manager was never to be found, and the purchase of advertising, in addition to relying on the experience of media buyers, each step has accurate data presentation, All the decisions of brand managers will see more marketing opportunities and strategies that have been overlooked in the big data age.

The traditional marketing dissemination of the ex-post investigation and evaluation methods for consumer feedback is very slow, timeliness is also very poor, but "big data" marketers almost real-time feedback data of a variety of communication effects, information detailed, and has tracking, which provides a great value for marketing communication optimization decision-making.

Large data marketing can solve some of the most difficult problems in traditional marketing (communication crowd, communication channels, the effect of communication, so the industry almost in a very short time to reach a general consensus: digital advertising industry is moving towards programmatic purchase, more activists have put forward the "traditional advertising is dead" view, There is also a view that large data can be used to achieve a more accurate and more panoramic picture of consumers.

But is big data marketing really omnipotent? In my opinion, the simplest mode of communication in 5W "what to say (dour what)" is exactly the last problem that big data marketing can't solve. Because "what to say (dour what)" is based on an insight into the real needs of consumers, and this insight is derived from the detection of the human nature of consumers, which is not the experience of machines and processes that can replace human beings at this stage.

The only way to achieve 5W communication mode large data marketing closed-loop industry is the electric business industry, because such ads like electricity dealers in most of its digital media budget through programmatic purchase, is perfectly reasonable. Because this kind of company mainly puts on the effect advertisement, pays attention to the consumer to read the advertisement the conversion rate, namely whether will buy immediately online. Programmatic purchase can get a cheaper advertising position, through the optimization algorithm to achieve more target audience.

However, for products, services and other core brands, large data marketing can not help them solve the problem of brand building, in this year, Ari Brandt published in the "Advertising Age" article "why the programmatic purchase of advertisers can not bring the desired marketing results?" Says he "commissioned Millward Brown to conduct a special survey of 300 digital marketers and marketing decision-makers: whether programmatic buying can be a tool for branding." The findings confirm my suspicions, reflecting the confusion of most respondents, who have many concerns about programmatic buying, including the Invisible banner ads, high-quality media resources, click Cheating, Brand security and Non-human traffic. ”

For non-electric business products, whether it's FMCG or consumer durables or a 3C product or a fashion industry, their most important work, in addition to sales, is to consistently build and maintain meaningful relationships with consumers, a relationship that is known as "brand building and Maintenance", It comes from the deepest insights into human nature, not other cold data analysis and procedural screening.

Large data can help us find the target consumer group we want to communicate faster and more accurately, in the past we used to study consumers in a sampling way, that is, to find consumers by random or quota principles, and to use surveys to get data; but, in the big Data age, It is very efficient and low-cost to monitor or track the mass of behavioral data generated by consumers on the Internet.

After capturing the exact target population, if we simply make programmatic purchases through big data marketing, we don't really know what consumers really want, because consumer demand is like an iceberg, and what you can easily observe is the tip of the iceberg; consumers ' real motives lie deep beneath the ice, Deep insight is needed to shake the entire iceberg. According to the iceberg theory, most of the potential human consciousness has an impact on the surface consciousness and behavior, and the user's potential demand is the product's real purchase motivation.

There is a story about instant macaroni that best reflects the fact that there is no good way to detect human nature at this stage, and that the story of target supermarkets is two distinct methods of consumer demand detection.

The story goes like this:

An instant macaroni brand has done a market research, they have obtained a very fresh discovery, the consumer is cooking the instant macaroni the time to add a little onion, so the considerate macaroni factory invented a new product, in the instant seasoning bag for the consumer to add some onion. As a result, when it was actually sold, no onion-added pasta was better sold than the new product added to the onion, which left market researchers baffled. There is, in fact, a human-based insight hidden here: housewives who cook instant macaroni for their family have a sense of guilt that they do not do a housewife's duty, and in order to dispel this guilt, they will choose to add a little of their own prepared onions to the cooking pasta, indicating that the meal was prepared by herself. They are not a lazy, incompetent housewife, so they choose to buy instant macaroni without onions.

To be sure, large data can be crawled by grabbing microblogs, Renren and a variety of forum data, access to consumer brand to the product of the immediate view and attitude, but the behavior of consumers are always their own reasons, these reasons, some consumers are willing to hang in the mouth and you say, mostly some superficial reasons, And lurking in the consumer subconscious they can not say but drive his behavior is the factor, is consumer insight, this is the brand and consumers to establish a "meaningful" relationship between the indispensable link, large data cannot replace the human. The vast majority of consumer insights are not derived from quantified research data and written research reports, it comes from direct, deep contact with consumers, such as street investigation, consumer behavior observation, conversation with the target group, more grounded, more primitive methods, rather than a string of cold digital symbols, the crowd label can be replaced.

Big Data marketing is not a substitute for human-based consumer insights, but it can change the pace at which advertising companies have been doing a creative job over the past few weeks, with large data based on real-time data mining technology to promote advertising companies to create hot-related content, communication companies can change their creativity according to performance. For example, in the recent World Cup marketing, every few hours, according to consumers in the social media on the hot topic to create a new idea, and to determine the extension of the topic of ideas, is from the sea, such as the social large data mining out the hottest topic.

In order to establish a loyal and lasting good consumer relationship, advertisers must return to the essence of communication, that is, to create valuable and innovative brand information, which is also a popular "content marketing" in mobile internet communication, the creation of content is the most important dour what in 5W communication, has always been the core competitiveness of advertising companies. However, in the big Data marketing tide, the advertising company must learn how to use the big data reasonably to reach the target crowd, and through the intelligent way to pass this information to the user, in order to deepen the emotional relationship with the consumer.

Wen/Zhong Weishan

(Responsible editor: Mengyishan)