Secret Man Has His Own BI System. Big Data Can Play This Way.

The BI core team should do both yards and farmers.

Managers need to understand and analyze business.

Sydney every morning at 9 o'clock, the BI team led by Sydney will open the BI system data monitoring module on time and check whether the data of each business system are back to the group data warehouse according to plan.

Because the next day, thousands of employees from various departments of the group will use BI to make decisions that affect the various business modules of the company through BI's data support.

In the conversation with Sydney, I learned that

The group BI team currently includes engineers, product managers and data analysts, including 5 people.

In June last year, there were only 3 people in the team. The BI core team had to be a code farmer and a product manager. It had to understand and analyze the business. Finally, it went through 5 months of uninterrupted development, and completed the BI system that the general enterprise needed 1~2 years or even longer.

In my opinion, the BI system of Yin man can be described as "amazing" compared with other BI systems that I have seen before.

and

Like the employees, the BI system of Yin man has its own name, "no leakage", which is taken from the "net of heaven and earth, but not loose". It represents Sydney's hope that she can have all the business data.

Sydney has implemented many BI projects for many enterprises.

But in his eyes, not leaking BI is the leading BI system in the entire electricity supplier industry besides the BI system owned by Alibaba, Jingdong and vip.com.

Now let me tell you about this amazing BI in Sydney from four aspects.

Number - store all data, real-time operation data is the basis of analysis. In the era of big data, every business link produces data every second, without losing BI, which is greedy for data.

Without missing BI, a huge distributed data warehouse has been built. Every day, millions of data are synchronized from the internal ERP, WMS, CRM, OA, and finance through ETL program.

On the other hand, there is no leakage of BI through the platform interface, in real-time, Taobao, Tmall, Juhuasuan, Jingdong, Dangdang, vip.com, distribution, Direct stores, franchisees, Yin man official website, APP and other more than 20 different channels of orders, merchandise, membership data back to the data warehouse.

Finally, a malicious web crawler unit was built without BI, and the data of industry, competitors and social public opinion were collected and stored every day.

1: do not leak BI data integration schematic diagram, in order to make computing faster and more intuitive, all reports can be calculated and displayed in 10 seconds. It is a technical bottom line that does not leak BI challenge.

In order to achieve such a rigorous user experience, without leaking BI to abandon the inherent mode of the original business database line storage, the innovative distributed column storage database is more suitable for analysis, so that the aggregation operation of hundreds of millions of data is reduced from 60 seconds to 5 seconds.

People - customer 360 degree portraits and lifecycle management all brand efforts are ultimately designed to enhance the value output of the customer lifecycle, without leaking BI to play a very important role in customer portraits and customer life cycle management.

The seller of electronic business has limited data to customers because of the paction data of his own store.

In general, customer portrait analysis can only be done through limited dimensions such as RFM model, membership level, region, product preference, etc.

And the ability to leak BI in the customer portrait is amazing. She can check the number and preference of any brand consumers in the sinomex group. They buy other women's clothing brands (Figure 2), which is very helpful for brand leaders to think about brand positioning.

No leakage of BI customer preference map

Besides, BI can also give a portrait of consumers' preferences for styles, preferences of popular elements and preferences of fabrics (Figure 3).

As a designer brand pursuing unique design, the most difficult to control is the matching of design uniqueness and customer demand.

By not missing BI, the designers of emann can fully combine the characteristics of the project and the preferences of the customers under the big data, so that the "dead money" can be avoided to a large extent.

No leakage of BI customer attribute preference picture sketch map

{page_break}

Relying on the 9 million customers of the 12 brands of the group, Hui Mei has penetrated into the fashion trend of the Internet women's clothing industry. Once the potential market segments are found in the consumer's trajectory, it is likely that thirteenth brands will hatch.

Goods -- from the planning end to the sales side of the whole process management, all brand emotions are ultimately communicated to consumers through products.

The clothing brand usually starts from the design plan, subdivides to the color, the fabric, the category, the photography plan, then passes through the clothing development, the big goods production, the upper shelf sale, the physical distribution and so on the long link finally achieves our consumer, between the designer and the consumer has created the huge gap in the time and space.

The core competitiveness of the Internet brand is a quick feedback on consumer demand.

BI does not leak the real name that shortens the distance between the designer and the consumer. The first single product of each style is sold on any platform. The conversion rate of the style, the click rate of the main picture, the sales volume, the refund rate, the score of the product, the proportion of the old customer and the age structure of the buyer will be collected in real time.

Responsible for this style of designers, boards, QC QA, merchandisers, photographers, details page design, you can view the actual data of the style to decide whether to return the product and optimize the product when it returns.

With the continuous optimization of the whole link, BI does not guarantee the consistency of product experience and consumer expectations, so as to build the core competition of Internet women's clothing.

The age structure of each BI consumer is not omitted.

In the past, hundreds of designers of the 12 brands of Hui Mei Group developed at least tens of thousands of styles every year. Each Department collected fragmented data in different styles, and spent a lot of time on meaningless data collection rather than analysis and decision making.

Now, no employee needs to collect data, so he can analyze and make decisions on the basis of real-time viewing the data of each style.

Field - all channel data on the line are integrated with customers and there are no offline points, only switching from different consumption scenes.

Due to lack of time, I choose online purchase on weekdays, and I hope I can buy more clothes on the weekend because I have enough time.

By integrating online and offline consumer data, we can identify the complete consumption trajectory of customers online to offline and offline.

According to the receipt address of each customer, the distribution of customers in every street and each district can be meticulous drawn without leakage of BI.

According to the distribution of the existing customers, Yin man has a very good accuracy in the location of the franchisees, and every shop will send messages to the fans after the opening.

No missing BI customer distribution coordinate schematic diagram

In addition, online sales category structure will generally be ahead of the line, double 11 is the outbreak point of winter clothing online sales, and the most popular time for offline sales is December ~1 months, which is a huge test for online inventory allocation.

By dynamically adjusting the category structure data under the online and offline line without BI synchronization.

Different from the general clothing shop mode, Yin man initially chose data as the core of the O2O Road, and the sales volume, category structure, inventory and store flow of each store were monitored in real time.

The supervision of each district only needs to check the current sales situation of the store and the shops with better performance in each area without BI leakage.

Through benchmarking analysis, we can quickly adjust other stores in the region.

No leakage of BI stores distribution sales monitoring schematic map