Upgrade to Pro — share decks privately, control downloads, hide ads and more …

Patents that Support Yahoo! JAPAN Shopping’s Ranking

Patents that Support Yahoo! JAPAN Shopping’s Ranking

Ryu Inagaki (Yahoo! JAPAN / Production Division 1, Shopping Service Group, Commerce Group / Backend Engineer)
Ayane Konno (Yahoo! JAPAN / Production Division 1, Shopping Service Group, Commerce Group / Designer)

https://tech-verse.me/ja/sessions/169
https://tech-verse.me/en/sessions/169
https://tech-verse.me/ko/sessions/169

Tech-Verse2022

November 18, 2022
Tweet

More Decks by Tech-Verse2022

Other Decks in Technology

Transcript

  1. - Online shopping service - Wide variety of stores &

    items - “PayPay” can be used What is Yahoo! Shopping?
  2. - Ranking is determined based on order, review, and stuff

    like that. - Provide information to aid the user's buying experience What is “Ranking”?
  3. Number of patents applications in the commerce field by Z

    Holdings Source:https://www.z-holdings.co.jp/sustainability/stakeholder/18/
  4. - Anyway, only the lowest priced ones come out on

    top. Problem that the ranking faced - Whenever I look at them, I always see the same products. Not a ranking I'd want to see over and over again!
  5. Changed the feature used for ranking Logic Coefficient Feature Approaches

    to Problem Solving Change the logic used to generate rankings Tuning coefficients Perfect Ranking
  6. The result was a failure vs Likes good deal items

    & Follows trends Likes high quality and long-selling items User A User B
  7. Policy Conversion-rate supremacy Change the decision making process to look

    at multiple indicators. before after Creating the "Perfect Ranking Logic" Search for "Optimal UI/UX with Ranking"
  8. Policy Conversion-rate supremacy Change the decision making process to look

    at multiple indicators. Creating the "Perfect Ranking Logic" Search for "Optimal UI/UX with Ranking" before after
  9. Policy Conversion-rate supremacy Change the decision making process to look

    at multiple indicators. Creating the "Perfect Ranking Logic" Search for "Optimal UI/UX with Ranking" before after
  10. 1. When you perform a search on an e-commerce site

    or auction site The patent essentials 2. Provide search results and rankings with tab
  11. The ranking were once isolated Search Item Cart Ranking Main

    Purchasing Site Flow Category Ranking Brand Ranking Top Ranking Yahoo! Shopping TOP Page Other Flow
  12. 1. Read the intent of the search terms retrieved by

    the user The patent essentials 3. Provide a flexible combination of constructed section 2. Construct various section depending on the intent
  13. Organize so that you have a clear idea of what

    you want Ambiguous categories Ranking classified by category Ranking classified by price range Ranking classified by brand Ambiguous brands Ambiguous price range
  14. When searching for “ふるさと納税” * Category Section (e.g. Steak, Meats,

    Sweets, Fruits …) Price range Section * hometown tax payment (e.g. ~10000yen, 10001~15000yen … )
  15. 1. Intent is inferred according to the search query The

    patent essentials 3. The details of the "format" are free 2. Display a different form of ranking depending on the judged intent
  16. Item Name Image Price Point Review Spec Store Name Review

    Excellent Store Badge Gift wrap option Delivery Postage Delivery date Shipment date Arrange a scheduled Wide variety of information What Konno focus on What Inagaki focus on
  17. Display format branching by search keyword Search keyword or Interpret

    the intent of search keyword Ranking System List Grid
  18. - Pros. Show diverse information - Cons. Not good at

    showing a large amount of items at the same time LIST Format
  19. - Pros. Display at the same time that a large

    number of items - Cons. Not good at showing diverse information GRID Format
  20. 1. When generating rankings The patent essentials 3. Ranking those

    two indistinguishably. 2. Mixing "products" that are grouped together and "products" that are not grouped together of the same commodity
  21. We provide mall-type e-commerce Store A Store B Store C

    Yahoo! Shopping MJTUTIPQ MJTUTIPQ MJTUTIPQ
  22. This is not a useful ranking for users who know

    the item or already have it. Ranking Issue © Nintendo / HAL Laboratory, Inc. / Pokémon / Creatures Inc./ GAME FREAK inc. Source: https://shopping.yahoo.co.jp/searchranking/?p=nintendo+switch+%E3%82%BD%E3%83%95%E3%83%88&rcid=50522
  23. This is not a useful ranking for users who know

    it or already have it. Ranking Issue © Nintendo / HAL Laboratory, Inc. / Pokémon / Creatures Inc./ GAME FREAK inc. Source: https://shopping.yahoo.co.jp/searchranking/?p=nintendo+switch+%E3%82%BD%E3%83%95%E3%83%88&rcid=50522
  24. Solution - The approach of aggregating the same products -

    Scoring as a single "product", taking into account the ranking of products, etc. © Microsoft / Mojang / Nintendo / HAL Laboratory, Inc. / ARMOR PROJECT / BIRD STUDIO/ SQUARE ENIX / Pokémon / Creatures Inc./ GAME FREAK inc. / BANDAI NAMCO Entertainment Inc. Source: https://shopping.yahoo.co.jp/searchranking/?p=nintendo+switch+%E3%82%BD%E3%83%95%E3%83%88&rcid=50522
  25. 1. When a user executes a search process based on

    a query The patent essentials 3. Generate and provide rankings appropriate to user information 2. Based on user information and query information entered by the user
  26. Get ahead of the ranking you want #1 Provide a

    ranking of the best deodorizers for cars deodorant They likes driving ×
  27. Get ahead of the ranking you want #2 Provide a

    ranking of "foods that go well with balsamic vinegar" ふるさと納税 They bought balsamic vinegar × * hometown tax payment
  28. Dynamically change the scoring depending on the trigger. Number of

    sales and sales history Transition in number of products sold PV numbers and review information SNS BUZZ Indicators 1 2 3 4 And more... Trigger Example
  29. Use of boosted rankings "Things that will sell in the

    future" "SNS Buzz" "What's hot right now" Ranking of... e.g. Super Realtime Ranking And more...
  30. '21/10 '21/11 '21/12 '22/01 '22/02 '22/03 '22/04 '22/05 '22/06 '22/07

    '22/08 '22/09 New Ranking Conventional Ranking Conventional Ranking vs New Ranking (CVR)
  31. '22/04 '22/04 '22/05 '22/05 '22/06 '22/06 '22/07 '22/07 '22/08 '22/08

    '22/08 '22/09 '22/09 '22/10 SP PC Trends in external inflows