Shen Du, a sharer, has rich experience in Internet R&D and marketing planning, and is good at finding more marketing value points in enterprise services through data analysis. I have been responsible for the digital marketing of Great Wall Motor, Colgate, L 'Oreal and other projects, and have rich practical experience in data, analysis, mining and application.
Editor's note: 20 18165438+1October 30-65438+February 3, the 7th Global Software Case Study Summit was grandly opened in Beijing National Convention Center, and the "List of 100 Cases" of 2018 was interpreted on the spot. This article is a case record of "Digital Marketing Solution for Service Fast-moving Brands" shared by Mr. Shen Du, the consulting director of big data products.
As we all know, advertising is divided into two branches, namely, effect advertising and brand advertising. Effect advertising is very common in the field of mobile terminals. For example, you see a message that encourages you to install and download an APP, or leave a message to register. This is the effect advertisement.
Brand advertising refers to all kinds of advertising information that people see in traditional media, whether it is asking people to speak or outdoor posters, it belongs to brand advertising. The biggest difference from effect advertising is that it does not seek to produce immediate results in an advertisement, but makes the brand image deeply rooted in the hearts of the people through long-term communication with consumers. Brand advertising can be divided into more detailed views according to different categories, such as durable goods, luxury goods, fast-moving consumer goods and so on.
For durable goods, the common way for general brands is to publicize their functionality, reliability and durability. Luxury goods are completely different. Luxury goods have their own brand concept and texture, which embodies the identity characteristics of luxury consumers. FMCG is a broader category, and consumers' ideas about FMCG will change with various factors.
The competition in FMCG market is fierce, and consumers change their ideas.
For FMCG, it is influenced by many factors. Let's talk about brand factors first:
Brand itself is a factor of consumption and decision-making. For example, brand factors such as Nike and BMW are deeply rooted in the hearts of the people. However, there are many activities of FMCG, and we often say: "No matter how sensational the advertisement is, it is not as good as double 1 1 half price." This is the activity factor. In addition, there are environmental factors. Now some FMCG products are sponsored in TV series/variety shows, and some people will buy them on impulse. At this time, its purpose has been achieved. Finally, the psychology of buying, such as coffee, you will want to try it as a consumer when you see people around you drinking it.
Consumer characteristics of fast-moving consumer goods;
1, buying habits vary from person to person.
2. Consumers' purchase of FMCG is a relatively emotional product. Now some online celebrities and bloggers bring goods. If you have a good impression on him, you are easily influenced by him.
The value of FMCG is not high, so the loyalty of consumers is not high.
Problems faced by brand marketing
These influencing factors of FMCG will affect the brand's performance for a long time, so how can we help brands better understand consumers? This is what brands care about. If you want to do brand advertising, you must pay attention to making the brand image deeply rooted in people's hearts. Therefore, in order to understand consumers, the traditional way is questionnaire or interview. A small number of users are selected to represent the whole group, and their internal motives are explored through their communication. However, this has three disadvantages.
First, the research cost is relatively high. Effective questionnaires need a lot, which may require thousands or even tens of thousands of questionnaires. At this time, the screening rate of the questionnaire is very high, a product needs a round of investigation, and the time line will be very long.
Second, when we do some interviews now, we need to get some news claimed by consumers, but in fact these news are not their real thoughts.
Thirdly, when we selected some media for contact, because the previous period was a sampling survey, when we put it in, the users of the volume could not cover the characteristics of the whole group.
In the face of these drawbacks, can we solve them in some digital ways?
Although big data emphasizes the total amount, advertisements must be separated from brands and effects. Traditional advertisements carry out large-scale promotion at some key points through rounds of brand impressions. In the digital field, we should also follow such rules, instead of making brand advertisements into effect advertisements after we have some digital tools. This is a misunderstanding that many brands have from the beginning.
The above picture shows the marketing rhythm that we often do with brands. For a brand, marketing time is separate. In some time periods of the working day, what it needs to do is to find the corresponding crowd and put in some corresponding content, so as to establish the brand impression. At the time of great promotion, relationship is like a reservoir. Usually, the brand impression is a reservoir, and the promotion is equivalent to a switch.
How can we help brand advertisers do digital marketing well?
In the digital age, everyone is a netizen. No matter what we do, such as consumption, entertainment, traffic and ordering food, we should use mobile phones. On average, there are 33 apps actively downloaded and installed on each mobile phone. So with such a mobile foundation, we can say that we can build a mobile marketing tool with people as the unit.
The above picture can be understood as follows: through big data, we can interpret consumers' behavior preferences on the mobile side, so as to fully understand consumers, and finally deduce marketing strategies through "people".
There is no shortage of data in this era. A brand has its data in various fields. For example, the e-commerce platform will have a lot of CRM information, order information, consumer information, consumer purchase time, purchase amount and other data. In addition, online media will leave a lot of consumer data, such as which consumers have seen advertisements where, whether they have clicked, and how sensitive they are to certain advertisements. But there is a big problem here. In different scenarios, the data is precipitated in different ways, so the data can't be opened.
So, how do we know that consumers will not only shop but also go to offline stores and see advertisements?
Now, BAT has a very strong account system. All apps and websites are logged in with WeChat/Taobao accounts, so I can know where the same person appears. Our method is through ID. We match these data, get through the ID with equipment as the unit/similar to human as the unit, and mark our own labeling system.
The matching of data is actually a relatively low-level work, which advertisers can't see. Since it is a product that provides insight research for advertisers, we should do more visualization to make the insight results more obvious.
The picture above shows the sales of a product in the flagship store. As you can see, 57% of the orders were completed during "6. 18", which is a very typical consumption pattern of FMCG. All its consumption is not evenly distributed, but suddenly completed at a certain explosion point, which falls within the time periods of "6 18" and "double 1 1". So it basically leads to shopping twice a year and solving the annual sales twice.
Everyone will have a stereotype. What kind of consumers will my product attract? Are they extremely price-sensitive people? If you take this cut-in as the main idea, then this is a misunderstanding of the follow-up user research. Let's take a closer look. We use statistical clustering algorithm to extract the characteristics of people who consume during the promotion period. As you can see, after clustering with this tag, five people have significant characteristics.
The second is the pursuer of quality life. These people are relatively older, they will have leisure activities and go to some nursing places.
The third kind is the shrewd shopper. When we talk about consumption upgrading, everyone wants to use better products, but they don't want to spend too much money. This group will become more and more common.
The fourth kind is otaku and otaku. Most of these people are single and secondary, but they are more satisfied.
The fifth is the student party. Students are relatively idle, and they often have some rich social behaviors, whether it is the exchange of visits between offline universities or various social behaviors online. We call it a place where hormones break out. For this group of people, we should convey a brand concept of youth and vitality.
In the field of brand advertising, it can be divided into two categories. The first category is consumer goods in a relatively narrow sense. In the domestic market, minorities have contributed a lot of sales, mainly high-end goods. To do this kind of brand marketing, there is no need to do large-scale/big media promotion, because most people will not buy even if they see advertisements. Therefore, we need to find high-energy and high-net-worth people to promote and maintain long-term effective influence. The second category, mass consumer goods, is the fast-moving consumer goods we are discussing today. In fact, anyone can afford such a product, but why choose your product? Therefore, it is necessary to instill different brand concepts for different groups, so that consumers can think that your products meet the current needs.
Developing Visual User Analysis Interface
In addition to the characteristics of online performance, users are often offline. For example, when we go to the airport, we often see some high-end advertisements, because the net worth of the airport is relatively high and people gather. However, the urban population is complex, and no corresponding consumers can be found to do user research and analysis. So under this condition, we made a tool to do offline crowd analysis. Let's look at two cases.
Case 1:
The picture above is called red and blue ocean. Two years ago, I did an offline position analysis for a famous bicycle in China. At that time, there was a big challenge, that is, the distance of the bicycle determined whether the user rode or not. We will pick out the people who may use bicycles, then make a big market, and then count how many people have installed the bicycle APP of this brand. Finally, we draw it as a picture with a threshold in the middle. If the value exceeds, it will turn red. Red indicates that the competition in this area is very fierce; Blue indicates that the concentration of vehicles in this area is low, but there are many potential users in this area, so the brand should increase the number of vehicles in this area.
Case 2:
The above four small pictures are made by us for Microsoft Surface. The first picture on the left is a sample analysis of Beijing. In the second picture, you can see that people with relatively high consumption are mainly distributed in the northwest corner and east of Beijing, namely Zhongguancun and Guo Mao. The third picture shows the distribution of typical users of Surface in Beijing, including students and IT Internet industry practitioners. The fourth picture shows the distribution of apples. The third picture is very different from the fourth picture. Surface and Mac are not directly paired. Therefore, when doing some activities and publicity, we prefer the Surface crowd, because Surface is more used by IT Internet practitioners.
As shown in the above figure, each line from top to bottom represents the charging state of the mobile phone for a period of time, and the rightmost time is from late to early. Every vertical line is made up of thousands of lines. Red indicates that the mobile phone is consuming power, and the power consumption in normal use is called active, green indicates that it is charging, and blue indicates that the power is relatively balanced, reaching or approaching 100%. We can see that a line is drawn at the junction of red and green or green and blue. The picture above is called a sleep chart on Twitter, which means that different people have different sleep conditions.
From traditional delivery to refined delivery
The traditional way of contact is guided by media and content. But this method has three disadvantages: high price, low exposure and fierce competition.
Contact with "TA" instead of media-runs through every day of digital life.
Now the mainstream way of programming is DSP, which is a platform and is responsible for docking various media. When a user opens an APP, if it is a screen advertisement, the request will be displayed to the device ID within a few milliseconds, and then sent to our system, and the system will judge whether the device ID is suitable for delivery. If appropriate, the system will answer yes and send it to the user; If it is not suitable, the system will answer no and fill it with another advertisement.
This way is to intervene in the advertising exposure process as a third-party data service role. For brands, all data will not be uploaded to the media, because we build a private cloud for customers, and the data is stored in the customer's private cloud, which will only be queried when it is used and uploaded when it is queried, which is also a kind of protection for user data.
Performance analysis of self-help media
It is very important to do media analysis after the launch, because the media performance and media-related traffic/quality in a launch process are all related to the brand display effect.
Let's look at an example. In one launch, we chose a group of people, half male and half female. After going online, we found that 60% of the people who clicked were men. Can you conclude that men are more willing to click? No, one factor is ignored here, exposing the crowd.
The above picture shows a system we have made for brand advertisers, and the contents mentioned above are all cases based on this system. Starting from the most basic data warehouse, we not only docked the data of the brand, but also docked the data of the brand's partners. Most of the data are connected in series, stored in the data warehouse and marked in our way. On this basis, we need to do analysis and insight.
The most direct way is to do media docking, like Ali and Tencent. There are direct API interfaces, and our people can send them directly to Guangdiantong, and then put them in after internal docking.
Media monitoring and anti-fraud, this is the cycle of data. This is not only a shell for the brand, but also the number of times used by the brand is constantly circulating. The more circulation, the richer the data.
Case enlightenment
? Data is one of the most important assets of brand owners. Now more and more brands are developing in the direction of data. FMCG brands at home and abroad have a clearer concept of data.
? Tools are only auxiliary, and marketing pursues essence. Marketing methods will never go out of fashion. Data and tools make the marketing process more efficient and can cover the blind spots of traditional methods, which is the significance that data tools bring to marketing.
? Data products should be developed in combination with actual business and cannot be built behind closed doors. Facing different customers, we should develop different tools according to the actual situation.