The Cainiao double 11 "billion package" battle

Every year, the Double eleven is refreshing the world wonder of logistics, but due to big data and collaboration, the task that seems impossible to complete is accelerated every time. Taking a set of data from 2013 to 2016 as an example, from the time of signing, it took 9 days for the double eleven parcel to sign 100 million in 2013, 6 days for 2014, and 4 days for 2015, and 2016 for 2016. The speed increase is only 3.5 days.

As a platform company, Cainiao insists on not owning a car or a courier. He hopes to build a socialized collaborative logistics and supply chain company through data and technology. In this regard, Cainiao has been exploring for several years, and has made a lot of attempts and breakthroughs in the whole link of logistics, such as accurate forecasting of large package volume and flow through massive data, rational storage and storage through supply chain forecasting plan. Spread the goods, and do the "single unsold, goods first" goods sinking in advance and so on.

This article will focus on parcel forecasting, supply chain warehousing, order sinking, order routing scheduling, electronic quotation and smart quotation, as well as some technical exploration and practice in the end of the small pieces, through the power of data and technology, defend this The battle of double eleven in the field of billions of parcels.

1 Double eleven starts from parcel forecast

The volume of logistics orders on the day of the double eleven in 2015 reached 467 million historically, a substantial increase of 65% compared to 2014. By 2016, the amount of parcels during the entire double eleven period will face explosive growth, reaching nearly 1 billion. Therefore, how to use the big data forecasting technology to estimate the total amount of parcels in 2016 from the macro level and use the micro data information to get the flow and flow of all parcels during the double eleven period, so as to provide decision support for the courier company, in advance Planning to survive the double eleven peak is a crucial technical proposition.

The total package forecast is based on the total amount of double eleven GMVs estimated by the front desk in 2016, integrating historical double eleven GMV data and logistics order data, historically all major promotion periods and daily pen unit price trends, analyzing various macro factors, and Monitor the Group's pre-sales sales data during the warm-up period, introduce possible uncertainties, and robustly predict the growth of 2016 parcels. In the end, we estimated that the total amount of parcels on the double eleven day was 680 million, reaching 97% accuracy.

The parcel flow rate is accurately disassembled into the forecasting and forecasting 680 million parcels to different courier companies, dismantled to various cities across the country, and each line, and based on the timeliness forecast, the daily shipments of each line and the future double eleven period are given. The arrival volume can help the courier company prepare the capacity in advance, dispatch personnel, accurately layout, and rationally optimize resources so as to be able to cope with the upcoming logistics peaks.

In this big data forecasting model, we mine effective information from the sales data of six million merchants over the past five years, predicting the courier companies and corresponding shipping warehouses and cities that merchants may use during the double eleven period; The historical behavior, mining the forecasting address of the consumer's double eleven period; according to the participation and ability of different courier companies during the double eleven period, predict their market share of the double eleven. Finally, integrating all relevant information to establish a regression prediction model, we can predict the amount of each courier company on each line in the country with 80% accuracy in the first 3 months of the double eleven, and really help the courier company to do When the soldiers and horses are not moved, the grain and grass are first.

2 Supply chain storage plan

In the preparation stage of the Double eleven, in addition to how to help the courier company to make data predictions, to achieve advance expansion of capacity, personnel, how to coordinate the thousands of goods of thousands of merchants into the warehouse of hundreds of counts, has also been the previous double eleven The first class puzzle. The warehouse's blasting, the concentration of goods to the warehouse, the preparation ability is difficult to match the demand, the arrival of the goods can not be put into storage on time, and so on.

2016 rookie double eleven warehousing collaborative platform planning in advance, through the establishment of big data reservation platform, serial information from the appointment to the completion of the entire execution process of the warehousing, precipitation, warehousing, carriers, drivers, lines, vehicles According to the operation algorithm, the actual storage capacity and remaining storage capacity of the double eleven stocking period, combined with the actual resource preparation and the stratification of the merchants and goods, intelligently recommend the appointment warehousing method to achieve diversified co-ordination. reservation. In the end, the orderly warehousing of the goods of 100 million grades was realized, and there was no explosion in the whole warehouse, and the merchants did not wait.

In order to ensure the consistency of resource preparation and merchant reservation requirements, S&OP also played a great role in the storage end. The system combined with the big data algorithm to predict the amount of inbound and outbound storage, the amount of business reservations, and the historical actual amount of inbound and outbound information, rolling analysis The estimated amount of storage and storage in the warehouses before and after the 11, the data is transmitted to the warehouse, distribution, and transportation system in advance, and multiple roles are coordinated to ensure resource preparation for the upstream and downstream links.

In addition, Cainiao further makes the goods directly to the nearest consumer, maximizing the consumer's logistics experience. Based on the historical sales data of the merchant's product granularity, the socialized warehouse network system, the line aging and the cost data, the real-time operation simulation system of big data is used to intelligently recommend the optimal goods warehousing plan and the warehouse inventory distribution plan. The data is transmitted directly to the replenishment system, ensuring that the goods are directly replenished to the warehouse nearest to the consumer.

Open the data of the warehousing link, through the selection of warehouse binning, big data reservation platform, S&OP, to achieve the supply and demand matching of consumer demand, merchant reservation, warehouse resource preparation, and ensure the optimal realization of goods in the warehouse. Distribution is another tool for rookie to make data play value and realize social resource synergy.

3 Order pre-packaging and inventory sinking

Every time the big promotion, the most concerned issue for consumers is when the package purchased is delivered. In addition to allowing the goods to be directly replenished to the nearest warehouse when they are in storage, the technology continues to contribute to the way the goods are closer to the consumers, and truly “single and unsold, goods first”.

What are the products that consumers are most likely to purchase, which products are most likely to be purchased together, and the sales forecast of large-scale merchandise and merchandise combinations is a must-do for rookie before the big promotion. Through the big data analysis and algorithm model, combined with the commodity inventory distribution of the warehouse, the historical delivery order situation, the data of the shopping cart and the favorites of the consumer on the transaction website, it is predicted that the order structure of the merchant during the double eleven period will be advanced in the warehouse. The goods are pre-packaged and ready to be shipped out. When the actual warehouse production is carried out, due to the pre-packaging, the production process will save a lot of time.

The pre-location and sinking outlets all over the country are an important part of Cainiao warehouse distribution network. After the sales forecast data is calculated, the automatic transfer system will also be launched. According to the sales forecast results, the inventory in the warehouse distribution network at all levels is lacking. A series of factors such as goods, turnover, timeliness, cost, etc., the system will automatically give the allocation advice, loading recommendations, the goods will sink to the front warehouse or outlets in an intensive manner.

For the reverse situation that may occur in the sinking, Cainiao also made predictions in advance. The big data algorithm sinking wind control model will give a refund probability score based on the consumer portrait, the product portrait, and the refund model. The high risk order will not sink. Low-risk orders sink, achieving an optimal balance of timeliness and cost.

The amount of parcels during the double eleven period is often ten or even dozens of times. Through the prediction and co-ordination of big data algorithms, pre-packaging, transfer, and sinking in advance can make full use of the prime time of the double eleven days and increase the parcel speed. Operational efficiency and enhance the consumer experience.

4 Global Intelligent Distribution Routing

Considering that the 2016 Double eleven coincides with Friday, a large number of orders will be delivered on Saturdays and Sundays. Due to the limited resources of the core area, plus the double eleven is generally more than 5 times the usual amount. If no effective measures are taken, It is highly likely that the end distribution network will be overloaded and cause congestion.

In addition, in the commercial office area, the weekend is generally rarely signed, the delivery will lead to waste of delivery resources, such orders will be issued during the delivery phase, and the delivery pressure will be reduced during the peak weekend, which can save the repeated delivery due to the failure to sign. Resources.

By analyzing the Gaode geographic information database and the daily rookie mass distribution data, based on the user's signing cycle behavior, public address population distribution, POI word formation structure, etc., the machine learning model is constructed, and the accuracy rate and office address recall rate are 95. %the above. Based on the address recognition technology, Cainiao can flexibly control the time of the warehouse packing and delivery of the office address order, which greatly reduces the possibility of “explosion”.

In terms of international logistics, Alibaba's Aliexpress export business and Tmall International's import business, in 2016, the double eleven singles were more than double the double eleven last year. Behind the cross-border overall single-index growth of Alibaba, how to choose the right carrier and route for each package, so that each package is optimized for cost, timeliness and service. It is a business. Great value and challenging things.

Cross-border logistics is characterized by long distance, long link and many roles, which leads to a large cost per ticket. Logistics costs account for a large proportion of the overall customer price of e-commerce transactions. In order to achieve the preferred goal, Cainiao began to think in 2015. How to solve the routing problem through technical means, and finally choose the optimal architecture scheme through manual strategy and big data intelligence.

The manual strategy allows the operating class to customize the routing strategy according to its own operating experience, and to perform various types of restrictive filtering according to the package's own attributes. The big data is based on the performance of historical parcels to do machine learning, from cost, timeliness, and service quality. The dimensions are optimized for each package, and the best technical solution is to achieve intelligent routing, manual strategy and big data optimization.

Since the launch of the global intelligent routing platform GRC (Global Routing Center), the business indicators have been significantly improved. In terms of timeliness, through this intelligent routing platform, Aliexpress successfully completed the aging improvement goal of Russia's Top20 city from 25th to 15th in 3~4 months, and the routes to Europe and the United States are basically the same. There is a 20% improvement in aging. In terms of cost, because the logistics line can be optimized according to the category, weight and size, the freight cost saved by the merchant is more than 15%; in terms of service, after the platform is launched for more than one year, Cainiao export logistics experience is analyzed through merchant research and DSR evaluation. Ranked among the best in China's export e-commerce platform.

5 Application of electronic face sheet

In the face of increasing sales volume of merchants, the traditional paper entry and printing efficiency is low, and the information between merchants, courier companies and rookies cannot be effectively interoperable, which is not conducive to logistics informationization and technology optimization based on this. In 2014, Cainiao launched an electronic face-to-face service, which will make the face-to-order electronic, greatly improve the delivery efficiency of the merchants, and carry the intelligent data distribution and other big data-based help logistics companies to improve the distribution center, outlets and other operations. effectiveness.

Cainiao electronic face sheet is a kind of online waybill generation, printing and management service launched by Cainiao. The system can provide 15 domestic mainstream express companies, many landing companies and many international logistics companies. Open Cainiao electronic face-to-face service to all merchants and complete access to more than 300 mainstream ISV systems. The electronic single-single system is like a HUB. Through the three-party system of HUB merchants, rookie and logistics companies, the information is interconnected.

Based on Cainiao electronic face sheet, we can provide a unified face-to-face service for the merchants. The merchant or ISV only needs to connect with Cainiao to have the ability to use a plurality of electronic companies (complete CPs with Cainiao docking)

In addition, we worked with major logistics companies to develop a standard electronic face sheet template, which greatly simplified and standardized the template format of the electronic face sheet, including the face sheet size and the face sheet content format. At the same time, we developed Cainiao electronic face sheet. The printing tool has developed and opened Cainiao standard printing specification to ensure that the format of the single sheet printed by Cainiao electronic surface system is unified.

Through the electronic face sheet, we can carry a large number of logistics services, and help the logistics companies, such as the currently widely used two-segment code, three-segment code, in the electronic face sheet generation, through Cainiao big data intelligent system, real-time calculation of the current package corresponding The routing network, the terminal distribution center, the terminal dispatching network, and the corresponding dispatching small-member information, when the surface sheet is generated, all the routing lines and node information of the package life cycle are completely planned and calculated, greatly improving the express delivery. Operational efficiency.

6 Smart order service

In the past big promotion, express delivery companies often have a phenomenon of bursting. In 2016, in the 11th, in the face of a record-breaking 650 million pieces of parcels, why do you rarely see all kinds of media reports?

Before the parcel is delivered to the consumer, a large-scale sorting operation is carried out at the distribution center and the outlet. In the past, sorters sorted by memory at the address on the list. Due to the huge amount of addresses, the probability of sorting errors is high, and the quality of parcel distribution is difficult to guarantee. At the same time, the sorting workers need to be trained for half a year before they can be employed. In the case of double-eleven and other large-scale activities, the amount of parcels increases sharply, and manual sorting Will become an efficiency bottleneck, leading to various phenomena of the explosion.

To solve this problem, Cainiao algorithm team developed a big data-based intelligent ordering system called “three-segment code” in the industry. The system accurately predicts the dispatch site and widget code at the time of shipment and prints the code on the order sheet to guide subsequent sorting operations. Through the smart order, the sorting workers directly distribute the code on the face sheet, and the accuracy of the distribution can reach 99.9%, which greatly improves the distribution efficiency. At the same time, it no longer relies on professional sorting workers for sorting, which reduces the operating costs of enterprises. At present, the system has been connected to domestic mainstream express delivery companies, saving more than 600 million yuan per year for the industry.

Last year, Cainiao launched the "two-segment code" to help the distribution center to quickly sort, which produced a huge effect. On the basis of the two-segment code, Cainiao 2016 double eleven launched the "three-segment code" to advance the predicted granularity to the courier and help the outlet to quickly distribute the parcel to the courier.

Cainiao can accurately assign the best courier to each order, and print the courier information on the order sheet in the same way as the two-segment code. When the end point is reached, the courier no longer picks up the parcel belonging to him by the detailed address on the sheet, but sorts it according to the "three-segment code" printed on the sheet.

In 2016, the double eleven and three-segment codes played an extremely important role in helping the outlets to quickly sort and digest packages. Temporary sorters no longer need cumbersome training to get up and running quickly; skilled sorters no longer need to peruse every address and sort by manual memory. The three-segment code simplifies the sorting operation and improves sorting efficiency, saving an average of 2.4 seconds per order. In 2016, the double eleven, the stay time of the parcel at the outlet was reduced by one hour compared with last year, and the total save time for the small staff was 160,000 hours.

Based on the electronic face and smart order, the mainstream express company enabled the large-scale automatic distribution device in 2016, which greatly improved the efficiency of the core node. Therefore, consumers can continue to perceive the double eleven package to send faster and faster.

7 Last mile of smart delivery

In the scene of logistics end-of-sentence, each small-member team generally has POI collections with natural boundaries, such as multiple communities, schools, factories, industrial parks, office buildings, etc., according to the community, school, factory, industrial park. The office buildings and the like are brought together so that the actual walking distance of the gathered packages is approximately equal to the linear distance.

Let the gathered packages be concentrated to complete the distribution, thus greatly reducing the redundant paths caused by frequent bypassing of natural obstacles (main roads, viaducts, rivers, community walls, mountains, etc.), so the package aggregation in the end scene of the logistics The efficiency of the courier is obviously improved.

Now, in the morning, the courier is full of parcels from the outlets, open the courier APP, through the background parcel aggregation algorithm, to inform the small members of the current distribution of parcels in the community, school, office building and other dimensions, easy for the courier to each after the aggregation The package realizes the batch notification, signing, checking, searching for the operation of the cell empty self-receiving cabinet (self-receiving cabinet), greatly improving the processing efficiency of the courier's dispatching task, and calculating the optimal delivery according to the aggregated parcel in real time. path.

The reservation parcels at a certain time of the day are grouped together according to a certain block (community, school district, office building, etc.), and are grouped together and assigned to the same courier. If a single order has been received but not taken If there is a new order in the small area (the same cell, office building, school, etc.) nearby, the new order will be directly traced to the list of small orders of the current order.

Through the merger and single-tracking to achieve the bulk collection of parcels, reduce the cost of the parcels to collect parcels, increase the income of the courier, double-track period and the amount of single-accounting accounted for 40% of the total receipts.

The courier can also take the parcel in the process of dispatching. In the process of dispatching, through the change of the parcel status in the dispatch list and the real-time position of the courier, the system can determine which community the courier is currently in, and The next-door community to go to, directly distributes the current parcels of the next-door community to the current dispatcher, thereby increasing the income of the courier.

8 Conclusion

The annual double eleven, huge amount of parcels, is a shock and baptism for Cainiao, sharpening the knife without cutting the wood, Cainiao has been thinking in all aspects of various fields, exploring the use of data technology to improve efficiency optimization services. The supply chain, order scheduling, smart ordering, etc. mentioned above, as well as other optimization practices not mentioned, are the power of data and technology to once again defend Cainiao double eleven "silky smooth" technology. Support provides consumers with a “silk-like smooth” logistics experience.