Syuppin: Automating dynamic pricing strategies in the resale camera market with Vertex AI
About Syuppin
Syuppin Co., Ltd. (Syuppin), is a company that manages and operates e-commerce platforms, which sells both new and used goods such as cameras and luxury watches. Its platform, Map Camera, which deals in new and used cameras as its main business, has developed a dynamic AIMD pricing system that automatically forecasts purchase and sales prices.
Tell us your challenge. We're here to help.
Contact usAbout SIGMAXYZ
SIGMAXYZ Inc. (SIGMAXYZ) is a business consulting firm that supports corporate businesses in the development of their transformation journeys. It offers consulting services to help create new value, including business strategy planning, development of new businesses/services, digital technology, SaaS/cloud solutions and project management, and is a Google Cloud Service Partner.
With Vertex AI's AutoML function, Syuppin is able to achieve highly accurate price predictions for products in the used camera market.
Google Cloud results
- Achieves rapid growth of 135.7% in segment profit
- Completes PoC to release in just 8 months
- Increases the frequency of price revisions by up to 6x to achieve more accurate pricing
- Provides automatic price revisions for 1,250 best selling items
Achieves rapid growth and increased sales with AIMD
Syuppin is a company that manages and operates e-commerce platforms, which sells both new and used goods such as cameras and luxury watches. However, as its main platform Map Camera, which buys and sells new and used cameras, kept growing, Syuppin found it challenging to revise prices in a timely manner. Without a proper process in place for price revisions, the staff in charge sets the price based on market trends and past sales data, making it very manual and dependent on the staff.
To solve these problems, AIMD was developed using the AutoML capabilities of Google Cloud's Vertex AI. With AIMD, AI automatically determines purchase and sale prices based on past records. The price determined by AI is then automatically reflected on websites and stores without manual pre-checks.
Makiko Matsuzaki, Executive Officer and General Manager, System Development Department, Information System Department, Syuppin, led the AIMD project. "Because our business model is not based on small profits and high sales volumes, it is important to secure gross profits. However, the development of AIMD was not an easy task. Before adopting Google Cloud, we worked with several vendors for about three years. Although we worked on development, we could not achieve our desired goal," shares Matsuzaki.
The company then turned to SIGMAXYZ Inc. (SIGMAXYZ), a Google Cloud Partner who has been supporting the company in building its core system, and has a deep understanding of Syuppin's business.
"Because our business model is not based on small profits and high sales volumes, it is important to secure gross profits. However, the development of AIMD was not an easy task. Before adopting Google Cloud, we worked with several vendors for about three years. Although we worked on development, we could not achieve our desired goal."
—Makiko Matsuzaki, Executive Officer and General Manager, System Development Department, Information System Department, Syuppin"SIGMAXYZ’s proposal was based on an analysis of past purchase and sales data, and I felt that SIGMAXYZ had a good understanding of our business model. In contrast to what was previously done, where AI directly predicted the purchase and sales prices. SIGMAXYZ's new proposal predicts the maximum number of sales, and then adjusts the purchase and sales prices based on that sales number. This proved to be a better solution based on a good understanding of our business, and not just implementing AI," says Matsuzaki.
Masatoshi Furuta, Director of Program Management Sherpa at SIGMAXYZ explains, "In the case of second-hand goods trading, both the sales and purchase prices must be predicted, so if you try all possible price combinations, the numbers will be astronomical, and take too much time. Based on our experience, we knew that predicting the sales volume and then using the price as an explanatory variable, rather than directly predicting the price, was more agile for Syuppin’s business. By using a model that predicts the number of transactions, we were able to determine the price in a short time."
Unlocking highly accurate price predictions with AI
AIMD uses Vertex AI's AutoML function to create price prediction models and execute predictions. BigQuery is used to store and process data for analysis, and Cloud Storage is used to link data with business systems and store intermediate files. Google Cloud's managed services are fully utilized throughout the system, including Dataflow for data processing execution infrastructure, as well as Cloud Run for executing scripts required for using Dataflow.
"We only had eight months to release the service, so using Google Cloud was the most reasonable choice for the project’s success, so we decided to use BigQuery. We also focused on development speed and fully used serverless managed services. We set our policy to develop the service on Google Cloud. As a result, we no longer had to consider non-functional requirements and redundancy on our own, and we were able to release the product in a short period," says Furuta.
When it came to price determination, which is the core of AIMD, Syuppin opted to build two models, one that predicts the purchase price, and another that predicts the selling price, to make timely pricing decisions. Furuta further shares, "When we compared the results of using multiple AI services in preliminary verification, Vertex AI's AutoML function was highly accurate and cost-effective."
In addition, leveraging Cloud Dataflow helped the release reach completion within the time frame of eight months. "Our strategy was to conduct a proof-of-concept (PoC) during development and proactively use PoC outputs in production. Cloud Dataflow played a major role in this process. Because we were able to easily reuse the scripts created in the PoC in the production environment, we were able to seamlessly progress through the cycle from requirements definition to PoC, design, and implementation, leading to a significant reduction in development time," adds Furuta.
"We only had eight months to release the service, so using Google Cloud was the most reasonable choice for the project’s success, so we decided to use BigQuery. We also focused on development speed and fully used serverless managed services. We set our policy to develop the service on Google Cloud. As a result, we no longer have to consider non-functional requirements and redundancy on our own, and we were able to release the product in a short period."
—Masatoshi Furuta, Director, Program Management Sherpa, SIGMAXYZ Inc.Achieving rapid growth with AIMD
After implementing AIMD, Syuppin significantly increased the segment profit of its camera business to 135.7% YoY in its consolidated financial results for the fiscal year ending March 31, 2022. Syuppin's original goal of speeding up the revision of selling prices has been greatly enhanced, and the number of revisions has increased by about six times. Syuppin also conducts direct marketing, such as notifying users who have added the product to their "wish list" by email when the selling price or purchase price of the product changes. Matsuzaki explains that the increase in the number of price revisions has directly led to an increase in the number of transactions of target products.
"In the past, there were hundreds of email notifications per day for items on customers’ wish lists, but after the introduction of AIMD, the number of emails increased to about 40,000. This has had a positive direct impact on sales. The sales team also appreciates the validity of the prices derived by Vertex AI's AutoML function. While it was a big decision to not conduct manual pre-checks, it has helped speed up price revisions," says Matsuzaki.
Currently, about 1,250 items, mainly bestsellers, are subject to automatic price revisions by AIMD, and Syuppin is looking to expand this in the future. Hirotaka Futami, Manager, System Development Department, Information Systems Headquarters, Syuppin, department manager, explains, "In order to use AI for price determination, we need a sufficient amount of trading data. There is also the question of what to do with new products for which only the most recent trading data exists. If we can standardize the process of adding and selecting target products, we should be able to use Vertex AI's AutoML function more effectively."
"Since AIMD is built with serverless products on Google Cloud, there are no operational burdens such as changing specifications or applying patches. The throughput of Vertex AI's AutoML is also higher than expected, so we are considering increasing the number of products to be forecasted. We have expanded our scope by adding AIMD price data to our newly added content recommendation function. With Google Cloud services, we were able to be more flexible," says Futami.
He goes on to add that this high level of scalability and maintenance is one of the strengths of adopting Google Cloud. "Google Cloud had all the services that could realize the AIMD workflow. We believe that this is a great advantage for users in terms of continuous operation and expansion. By consolidating all AIMD functions in Google Cloud, we were able to build a new system infrastructure that will lead to future growth."
"Since AIMD is built with serverless products on Google Cloud, there are no operational burdens such as changing specifications or applying patches. The throughput of Vertex AI's AutoML is also higher than expected, so we are considering increasing the number of products to be forecasted. We have expanded our scope by adding AIMD price data to our newly added content recommendation function. With Google Cloud services, we were able to be more flexible."
—Hirotaka Futami, Manager, System Development Department, Information Systems Department, SyuppinTell us your challenge. We're here to help.
Contact usAbout Syuppin
Syuppin Co., Ltd. (Syuppin), is a company that manages and operates e-commerce platforms, which sells both new and used goods such as cameras and luxury watches. Its platform, Map Camera, which deals in new and used cameras as its main business, has developed a dynamic AIMD pricing system that automatically forecasts purchase and sales prices.
About SIGMAXYZ
SIGMAXYZ Inc. (SIGMAXYZ) is a business consulting firm that supports corporate businesses in the development of their transformation journeys. It offers consulting services to help create new value, including business strategy planning, development of new businesses/services, digital technology, SaaS/cloud solutions and project management, and is a Google Cloud Service Partner.