Extraction of product attribute information and automation of master registration

  • #Automation
  • #product master
  • # Product attribute extraction
  • # Product attribute estimation
  • #extend database
  • #Wholesale
  • #EC
  • #CAX for Product Master Database

Business challenge

商品登録業務において、商品登録担当には専門性が求められ、多くの人員をアサインすることが難しい。 ビジネスを拡大するためには商品点数・商品属性情報を充実させていくことが必要だが、手作業の商品登録では限界だった。 商品登録業務をAIによって自動化することで、現行の数倍の商品数量と商品属性を登録できる体制を構築し、人的工数を削減することを目指す。

Task

Product master data is an important asset that will serve as the basis for marketing and sales trend analysis in future DX. As marketing methods and data analysis methods become more sophisticated, it is becoming more and more important to increase the number of products handled by the company and to enrich product information. On the other hand, "increase the number of products and product attribute information = increase the information registered in the product master", The registration work to the product master becomes a bottleneck, and it is not possible to increase the number of products as desired. There is a status quo.

Registration work to the product master

Problem-solving process

① It is not possible to increase the number of people in charge of product registration

In order to increase the number of products handled in product registration and the types of product attributes, it is necessary to assign new people to be in charge of product registration and carry out the work with more people. On the other hand, while there are cases where human resources cannot be allocated due to a shortage of manpower, it is becoming more difficult to allocate human resources to product registration operations. As a result, it is not possible to increase the number of products and product attribute information as desired. In order to continuously expand the business, it is necessary to radically reform the product registration process and start thinking about improving efficiency and productivity using technology.

(2) It takes time to register products

Persons in charge of product registration work are required to have certain skills. The product registration process can be roughly broken down into the following three steps.

STEP 1
Collect and classify product information
STEP 2
Read product information and input product information to database
STEP 3
Content confirmation/inspection/content correction
Product registration work
Challenge Result

(1) Collection and classification of product information: The process of collecting and sorting product attribute information of the products handled by various methods.

There are cases where product information is sent from business partners and cases where information is collected by referring to websites, etc. In each case, the format of product information varies.

  • ✔ The product catalog will be sent in PDF format
  • Product price information is sent in Excel format
  • Browse websites and collect necessary information

異なる情報収集先、異なる商品情報フォーマットを取り扱うには、担当者が実際に手を動かし、商品情報の収集と分類を行う必要がある。1商品あたり3分の作業時間が必要な場合、100商品で300分、10000商品で30000分(500時間)もの工数が必要。

(2) Reading product information and entering product information into the database: The process of reading the collected information and entering it into the database

収集した商品情報を一つ一つ確認しながら、データベースへ情報の入力を行っていく必要がある。現在、多くの企業では、商品登録業務を担当者が手作業で実施している。商品情報が載っているテキストファイルやEXCELファイル、カタログ等を目検で確認・内容を理解し、その情報をデータベースに写しかえるといった作業が発生している。
商品数量が増えれば増えるほど、商品登録業務の工数は増大となる。また、商品属性情報を多く登録すればするほど、一商品に対して必要な商品登録業務の工数が必要。

  • 100 products x 10 product attributes = 1000 column information input
  • 1000 products x 10 product attributes = 10000 column information input
  • 1000 products x 30 product attributes = 30000 column information input

Reading product information and inputting product information into the database are operations that occur continuously throughout the year. In addition, the work load temporarily increases during the busy season, such as when the manufacturer's part number changes or when the product is revised. there is

(3) Content confirmation/inspection/content correction: A process of checking the consistency of the entered information, correcting incorrect product information, and final confirmation.

データベースに入力された情報を、別の担当者が改めて目検で確認を行い、最終的に作成されたデータの正確性を担保する。このステップに関しても、現在多くの企業では、担当者が手作業で業務を実施している。このステップに関しても、商品数量が増えれば増えるほど、商品情報の確認・修正の工数は増大になる。また、商品属性情報を多く登録すればするほど、一商品に対して必要な商品登録業務の工数が必要になる。

Problem-solving approach

(1) It is not possible to increase the number of people in charge of product registration. (2) Product registration work takes time.
In order to solve these problems, what kind of approach can be considered from the perspective of utilizing technology?

Approach using AI (Artificial Intelligence)

By automating the work in product registration with AI, it is possible to reduce human man-hours and shorten the lead time of product registration work.
The advantage of AI is that there is no need to set rules for product registration, and the AI model itself finds rules through learning using training data. Therefore, even if the number of business partners increases, the AI model can be reused to a certain extent, realizing automation of more general product registration work, reducing many man-hours, and improving product quality. Registration work time can be shortened. However, AI also has its own problems, and whenever an AI problem-solving approach is taken, it is necessary to prepare training data for AI models to learn. However, in many cases, there is already product attribute information for products that have been registered in the product master in the past, so it is possible to develop AI models using them as training data.

Fully customized AI solution service that automates product registration work
AI implementation example in CrowdANALYTIX for Product Master Database (hereafter CAX PMD)

AI implementation example in CAX PMD

Click here for specific solutions using AI.