The flow pool thinking turned out to be a rapid replacement of the "growth hacker" and was regarded as the new purpose of the marketing industry traffic game. Subsequently, a wave of crazy science waves swept the circle of friends and took a look. After a preliminary understanding of the traffic pool, everyone also looked around for ways to build a traffic pool.

Once the traffic pool is in operation, the data base + strong and fast system is needed to achieve the purpose of operation, discovering the value of data, and bringing more traffic. In this issue, I will share some ideas on “Building a Data Pool”.

First, why do companies need data pools?


Businesses have more or less their own customer data, but they have a hard time answering “what data do we have?” “The data we have is able to reflect the true state of the customer” and “whether the data dimensions used by different departments are consistent”.

1. We can take a look at what data does the company have?

First party data

Transaction order data: Various types of transaction information generated from ERP, CRM, and e-commerce systems. Including card coupons, orders, shopping carts, return orders.

Behavioral data: A large amount of behavioral data generated by customers on various first-party contacts such as WeChat, websites, apps, and applets. For example, follow WeChat, submit forms, visit pages, and more.

Business object data such as products: These data are not customer data, but are extremely relevant to the analysis. Inventory and product prices, for example, are the data points necessary for many retail customers to analyze.

Data generated by external tools: Modern marketing relies on a large number of external tools, such as registration forms, mail, micro-classrooms, micro-stores and other systems to generate large amounts of data.


Second party data

Data returned by the cooperative system: If the email or SMS is sent, the customer has read, clicked, etc.

Cooperative media data: data provided by ad serving, video, portal, vertical media, etc.

Third party data

Data provided by third-party data providers, such as data platforms, operators, etc.

2. Where do companies put data?

Each business unit relies on different aspects of customer data, and they all have their own application scenarios. The sales department relies on CRM (Customer Relationship Management Platform), the after-sales department mainly looks at the customer service system, the marketing department cares about the WeChat platform, and the data analysis team uses the data warehouse or customer behavior analysis tools.

Each department has its own main system, each system has different concerns, and it is also aimed at customers at different stages, so the things that are seen are of course different.

Just like the blind people feel like each other, each department sees only the part that they care about, not the complete situation of the customer. The tools used by different departments each generate new, isolated, one-sided customer data that cannot be quickly synchronized.

The fragmented data poses a significant challenge to operations.

For example, an enterprise wants to be a customer service applet. It turns out that Billing data is in ERP, order information is in the e-commerce system, and behavior data is in the background of the website. Although the features they do are very simple, companies need three developers to take values from three different systems.

After finally understanding the value rules of different systems, you also need to write a lot of logic to merge. Then, the next time you do another small program, you have to repeat the above steps again...

Therefore, enterprises need a data pool as the basis to ensure subsequent traffic operations. To be clear, the data pool is not a product, you can understand it as a type of data asset for the enterprise.

Second, I have other data tools, can I use it as a data pool?


The answer is, no.

1. Can the Behavior Analysis Tool be a data pool?

Many people ask the Behavior Analysis Tool to collect customer behavior data as well as provide an extensible data structure. So what is the difference between "data pools"? The core difference is the granularity of the data.

For example, a company has multiple apps. When the "data pool" is stored, the data is stored in different categories according to different apps. When viewing, you can see the inflow and outflow of data from each channel, or you can slice a person according to different channels.

However, the Behavior Analysis Tool generally recommends that you isolate the data for each app because it is not designed for cross-channel data integration.

2. Can “Data Lake” be used as a data pool?

It should be emphasized that the “data pool” is different from the data lake in the traditional sense.

The Data Pool focuses only on collecting customer data and can standardize and standardize these customer data assets on a large scale. Through the "data pool", the collected data is not a simple merge, but can be saved by saving the data of each channel.

The Data Lake is a library that stores a large amount of raw unstructured and structured data, so the data lake is difficult for people who don't understand IT.

The “data pool” is a substitute for the previous business system. It is a data asset that is reorganized by the company's existing operational traffic.

For enterprises, the data collected by the Behavior Analysis Tool and the Data Lake are different types of data assets, but they are not applicable in the scenario of building a traffic pool.

Third, how should the data pool be built?

1. Open data barriers

Without relatively complete omnichannel data, operations will be blocked, so companies need a shared data source that connects every customer interaction on each channel, from WeChat to websites, from stores to ERP, payment services, customer service systems, and even It is CRM.

Then, pass the data to the system used by each department.

This allows data to flow between departments of each system, breaking the company's “department wall”, giving everyone a comprehensive understanding of the customer and helping to build a company-based, fact-based culture.

In turn, it saves development costs, improves operational efficiency, and makes the team more focused.

2. Data normalization

Various duplicate records, missing fields, and mismatches in cross-system data values have been a problem that plagues operations.

So companies need to implement common data standards across the organization, define what constitutes good data, and remove erroneous data from the source, so that the entire organization believes that the data is correct.

3. Building an image

The above two steps can be said to be the data preparation phase, and then it should be the data insight phase.

Enterprises can translate the collected customer attributes and behavioral data into labels, and give each customer a unique “branding”, such as their favorite product categories, frequency of purchase, etc., and analyze and count these characteristics. To explore potential value information and outline customer images.

After labeling, companies can also group labels based on custom business conditions, so that when communicating with customers, they know what to say and provide a unique experience for customers.

4. Manage complete behavioral data

Customer behavior data can be complemented by customer portraits. With the proliferation of customer-enterprise interaction channels, it is often not enough to know the customer's preferences. Sometimes it may be necessary to know when the customer's behavior occurs, so as to more accurately prescribe the right medicine.

By capturing the customer's complete behavioral data, companies can gain a keen insight into customer intent, develop operational strategies based on different phases, and fully exploit customer lifecycle value. With complete behavioral data, companies can also target targeted operational strategies for the same type of customers.

The data pool is the data foundation of the traffic pool, so high quality data is essential. High-quality data is an important reference for analysis and decision-making, and refined operations, so that enterprises can better serve customers and achieve the purpose of exploring data value and bringing more traffic.