How to use big data for marketing?

We can know very detailed real data of an individual, but how can this data help specifically in marketing ? This article tells you a new idea of ​​using big data for marketing.

1. About Big Data

In recent years, many people like to discuss the topic of big data with me at work and in life. My friends all feel that when we look at big data, it's like we have collective presbyopia: it looks very clear from a distance, but becomes increasingly blurry when we get closer. It's not the right way to do it.

We all know what big data means - a large amount of real data with high reading speed, multi-dimensionality and low value density .

It's difficult to read, but to put it simply, we can know very detailed real data of an individual, but it's hard to think of how this data can help with marketing execution (this is where the difficulty lies).

In reality, the biggest help that big data provides to managers is often a sense of control , which is actually similar to ancient superstition. When I see all kinds of data about my consumers, I feel like I know them very well and everything is under control.

But when it comes to how to use this data, it simply comes down to looking at what consumers have consumed/browsed in the past and then pushing them what they want (this is also the logic behind Taobao and Baidu’s push notifications).

Last year, I led a project to connect the data of retail companies with the most well-known data bank in the country. After accumulating a certain amount of data, we have data on consumers’ age, gender, occupation, income distribution, number of family members, whether they have a car or a house, mobile app interests, reading interests, etc. (and many more).

Then everyone got stuck in a quagmire: how can we make good use of this data ?

Going back to the starting point, I understand that big data is useful in two main aspects. One is to look at trends, markets, and the direction of development of consumer behavior (this is long-term). The second is to do sales, increasing marketing efficiency by analyzing consumer data that can be linked to one’s own business (this is short-term).

There are many large data organizations, such as Alibaba Cloud, CBNData, or consulting agencies such as Deloitte and KPMG, which regularly provide consumer reports on different markets. This kind of report is mainly used to analyze long-term trends.

Today's article mainly talks about the latter, that is, how looking at your consumer data every day can help a company improve its performance .

There are two main ways of applying big data in many large companies today. In addition to the simple and crude method of pushing duplicate information as mentioned above, a more advanced method is labeling + precision marketing .

Simply put, there are two steps:

First, label consumers differently based on age, gender, hobbies, habits, and life stages;

Second, determine which type of consumers the marketing information is targeted for and reach them directly.

However, this type of big data application method also has two disadvantages :

1. Data Source

Although big data obtains behavioral records of target consumers indiscriminately, when we use this data in a local environment , there is a great possibility that it will be interfered by inaccurate data sources .

For example, I once worked on a consulting project, using big data to capture the customer flow around a coffee shop and the data of customers entering the store, and to analyze why the coffee shop’s sales were always low at a location with a decent flow of people.

After two months of data collection, we found that the age distribution of consumers in this store is mostly middle-aged or even elderly, and they stay in the store the longest.

Then we fell into a misunderstanding, thinking that this store should have more elements that can attract middle-aged consumers, such as more conspicuous discount information, and even enlarge the words on the menu. However, this has little effect.

Until one day we decided to go to the store in person to see what was going on and found that these measures did not increase the transaction rate even a little bit.

It was also a hot summer. When we walked to the door of the store, we found a very strange phenomenon: many middle-aged and elderly people were sitting at the door of the store looking at their mobile phones.

It turns out that there is a bus stop at the door of this store. Because the store has air conditioning and Wifi, many elderly people sit at the door of the store to enjoy the air conditioning while waiting for the bus. This is the truth behind the so-called main customer group being middle-aged and elderly people .

This is not the first example where big data can mislead people. I also worked on an analysis project for a beauty online store, and the data showed that during a promotion, more than 50% of the transactions came from male consumers' accounts.

So this promotion seems to be more effective for men's products? But it is not. When we analyzed the shopping baskets of each order, we found that male accounts only bought women's products. In fact, the girl used her boyfriend/husband’s account to pay the bill.

Therefore, if we simply believe in the data presented on a panel, our judgment is likely to be misled, because the consumers we see in the data may not be what they are like in reality .

2. Fallacy of Inductive Reasoning

At present, many large companies' use of big data is still at the stage of inductive reasoning. That is, the data shows that most of the characteristics of your consumers are A, B, and C, and then you infer that the consumer's label is D, and then you reach out to them with information.

For example, a hotel finds that most of its guests have these characteristics:

• Parking space required
• Mostly stay for one to two nights
• No or little consumption of the room minibar

Based on these three conditions, it is easy to draw a conclusion: the main customers of this hotel are families on short family trips. Therefore, hotels can increase user stickiness by adding family meal packages and family packages for nearby attractions. Then we will promote it through family travel forums and public accounts .

Although this may seem like normal reasoning, this label may be completely wrong.

It is not just family tourists who meet the above three conditions.

There may also be business people who come to a nearby company for a meeting. They will bring their own good drinks as a treat and even go out for a night out.

It is also possible that there is an Internet-famous store nearby, and couples nearby drive over specifically to check in, which is a short-term purpose.

The best way to avoid this kind of inductive reasoning error is to find " key indirect evidence " in different data dimensions. It's like Sherlock Holmes seeing that the burnt part of a pipe is on the right side and inferring that the user is left-handed.

For example, in the example above, none of the three conditions can filter out an accurate label. But if we add an extra data dimension "these guests' rooms will require extra beds", then we can basically confirm that they are customers traveling as a family.

Therefore, when using big data to label target customers, identifying key labels can effectively increase subsequent conversion rates . After all, if the labels are wrong, there will be no precision marketing.

However, this "label + precision marketing" approach still has a big flaw.

From the perspective of consumer behavior, even if the label is correct, it is useless to push marketing information at inappropriate times and places. Just like a newborn mother is accurately labeled, it does not mean that she needs to buy milk powder and diapers anytime and anywhere. Not to mention that being a middle- to high-income fashionista doesn’t necessarily mean a consumer has to pay for a particular trendy brand.

Fundamentally, it is because consumers have a variety of purchasing motivations.

Another disadvantage of labeling consumers and marketing to them is that it reduces the likelihood that people who are not in the target audience group will buy your products . For example, just because I am a college student who loves studying and working hard for postgraduate entrance exams doesn’t mean I won’t be interested in trendy brands like Su pr eme. Or being a high-income senior manager doesn't mean he doesn't like the simple and plain Toyota cars .

However, if these groups are not reached due to precision marketing, it is foreseeable that the sales opportunities that should have been there will slip away.

So why not understand big data and precision marketing in a different way?

Sometimes, it is more effective to label marketing scenarios and develop several different sets of copy based on the scenarios.

2. Marketing Scenario

Next, we will talk about the labels that can be used for marketing scenarios from the three dimensions of instinct, emotion and cognition .

The reason for choosing these three dimensions is that in motivational psychology, the intrinsic motivation of a person's behavior is mainly influenced by these three aspects, among which instinct and emotion are hereditary, while cognition is learned.

When these labels are marked independently or simultaneously in a certain marketing scenario, they can tell us what information the product copy needs to highlight.

1. Instinct

Anyone who has read Maslow's hierarchy of needs theory knows that the bottom two levels of needs are human instincts. The bottom layer is related to survival and reproduction, such as breathing, water, food, sex, etc. The second layer is related to a sense of security, such as health, assets, morality, etc.

The reason why a sense of security is so important is that humans naturally seek certainty from their surroundings and spend our entire lives increasing this certainty .

In gathering societies, owning a cave of one's own meant being sure to be free from harassment by large predators, and through sacrificial rituals one could try to ensure a favorable natural environment.

The constant pursuit of certainty has inspired human progress from thousands of years ago to the present day.

To this day we are still trying to increase certainty in the world

Countries increase certainty in the international community through international organizations such as the United Nations

Individuals can increase their income and career prospects by joining a Fortune 500 company

The mother-in-law increases the certainty of her daughter's quality of life by asking her son-in-law to buy a wedding house

Without exception, these certainties give every stakeholder a sense of security.

Researchers have found that when we feel safe in an environment, we tend to choose more personalized products. When we are in an environment where we lack a sense of security, we tend to prefer mass-produced products.

For example, the marketing channel of an APP is before a movie on a certain video website. For many businesses, all movie advertisements may be a channel that attracts people who like to watch movies. But what is more effective is to label TV series and movies.

If the audience is watching thriller or suspense movies, then the marketing information of this APP should highlight that many people are already using it , such as " the second-hand car platform chosen by 30 million people".

On the contrary, if the theme is something like romance or science fiction, the copy of this APP should highlight its uniqueness , such as "no middlemen to make a profit, and it is guaranteed to sell out within three days."

These two tags are not only applicable to advertisements in streaming content. For example, advertisements placed in office building elevators should convey popular information, while those placed in residential building elevators should convey personalized information, because people often feel safer near their homes than in the office .

2. Emotion

Different marketing scenarios give consumers complex emotions. In order to simplify complex issues, the labels are divided into two categories: familiarity and unfamiliarity.

There is a concept in psychology called the priming effect , which means that when our eyes see anything, our brain will begin to associate all the concepts related to this perceived target.

For example, when I see a meadow, my subconscious mind will mobilize my cognition and memory to start associating various concepts related to the concept of "grassland".

At this time, because everyone is watching the World Cup recently, the concepts related to football and the World Cup will be easier to be "triggered". At this time, things like the national football team and even the football players will give us a sense of familiarity.

When consumers are faced with familiar concepts, they tend to start thinking about cost issues, that is, what will prevent me from taking a certain action.

When faced with unfamiliar concepts, we tend to start thinking about the benefits, that is, what benefits this thing can bring .

Therefore, in this dimension, before labeling a marketing scenario, we need to think about how directly the product is related to the content of the marketing scenario for the target group .

For example, if we sell clothes and invest in a public account with emotional content, then this marketing channel will be highly familiar to the target consumers. At this time, the first thing consumers will think of is “Oh, there are clothes for sale, let me see how much they are first.”

Similarly, in a public account with emotional content, if the product to be marketed is tea sets, then it is considered to be of low familiarity. At this time, consumers will immediately think, "What's so good about tea sets sold in this place?"

3. Cognition

In the previous article "We often talk about drainage , what are we talking about?", I mentioned the concept of cognitive closure mode .

Simply put, we have different levels of acceptance of ambiguity in answers to questions in different scenarios.

For example, when a certain online shopper buys clothes on Taobao , she is thinking "I want to buy a skirt so that I can go shopping with my boyfriend next weekend." At this time, every time I see a style, I have a lot of questions in my mind:

“Is this style a standard for this season?”

“Will this fabric seem see-through/hot/wrinkle easily?”

"Is this store having any sales soon? Will I lose money buying it?"

"This style is similar to the one I saw before. How will it be different when worn?"

“…”

At this time, there is a high demand for cognitive closure , because in order to perfectly complete the task of buying clothes, these questions must be answered accurately before a decision can be made .

When consumers browse information and think they are completing a task, they will have a high need for cognitive closure.

Under what circumstances are consumers completing tasks?

The answer is transactional scenarios, such as e-commerce platforms such as Taobao and JD.com , or offline retail stores such as supermarkets and personal care stores, or even websites for buying air tickets and booking hotels.

On the contrary, in content-based scenarios, such as video websites, short video apps, public accounts, and Moments , consumers will enter a low-cognition closed demand state, which means that they are prone to make impulsive decisions due to the advantages of one or two products.

(PS. This is also the fundamental reason why the conversion rate of WeChat business is so high)

So when we label a marketing scenario as a transactional scenario, the marketing information should be a list and comparison of detailed data, and address consumers' main concerns. This scenario is more suitable for the marketing of supplementary products, such as daily care products.

For content-based scenarios, marketing information should be as concise as possible and stimulate with emotional connections. This type of scenario is more suitable for the marketing of new and unique concept products, such as newly released foreign imported products or an unknown romantic hotel.

Conclusion

Today I talked to you about a new idea for using big data.

Nowadays, many large companies usually use the methods of repeated push of browsing/purchase history and "label + precision marketing" to conduct big data marketing. These methods are affected by data sources and inductive reasoning fallacies and cannot effectively improve conversion rates.

Therefore, here we give you another option, which is to first use big data algorithm to label marketing scenarios, and then push personalized information in different marketing scenarios.

The advantage of doing this is that the marketing scenarios and product information that need to be communicated are all objectively determined.

Merchants only need to customize several different sets of marketing copy based on the label combinations of marketing scenarios, and then deliver them to different channels through big data.

Simply put, it means "using corresponding baits in different ponds so that the fish in that pond are most likely to bite the hook under the current water temperature and lighting conditions."

Author: Mr. He, authorized to be published by Qinggua Media .

Source: Yuanwai He (ID: Yuanwai-HE)