Tiago Fonseca & Rui Velho - Syone & Leroy Merlin - OSL19
1 like202 views
Leroy Merlin Portugal implemented an Elasticsearch solution to enhance their e-commerce search engine, aiming to increase sales and reduce costs. This allowed for personalized searches among 250,000 products and improved response times to around 100ms, achieving a 40% cost reduction. The solution was developed using open-source technology and involved significant input from internal teams, enabling future innovations like machine learning and analytics.
![SOCIAL
MOBILE
INTERNET
ANALÓGICO
[1923]
DIGITAL
The Speed of Change](https://image.slidesharecdn.com/tiagofonsecaruivelho-syoneleroymerlin-osl19-191015130438/85/Tiago-Fonseca-Rui-Velho-Syone-Leroy-Merlin-OSL19-6-320.jpg)
Tiago Fonseca & Rui Velho - Syone & Leroy Merlin - OSL19
- 1. How Leroy Merlin uses Elasticsearch to Drive Sales and Increase Revenue on their E-commerce Rui Velho – Leroy Merlin Tiago Fonseca - Syone
- 2. Agenda • Context • Initial Problem • Challenges • Solution • Benefits
- 3. Context
- 4. Leroy Merlin Portugal – Adeo Group
- 5. 410 MILLION CUSTOMERS SERVED 23.1 TOTAL TURNOVER BILLION € 114000 LEADERS 14 COUNTRIES 800 STORES N° EUROPEAN PLAYER onthe global DIY market ON THE GLOBAL DIY MARKET 1 N°3 WORLDWIDE BRANDS and banner brands 3 USEFULTO INHABITANTS ALL OVER THE WORLD
- 7. Time2Market oriented Open Agile Product Design Thinking With end users Mobile-first approach Internal team involved Start SmallFeedback Modular Architecture Build & Run Fast Continuous delivery short cycle Drivers
- 9. Initial Problem • SaaS Solution Implemented • No optimized Search Results • Expensive Software Maintenance Costs • Lack of Product Support
- 10. Challenges
- 11. Challenges • New Search functionalities needed • Start small (MVP Product) • Build Fast (AGILE) • Innovate (+ Use Cases) • You Build It You Run It
- 12. Solution
- 13. Solution MVP Product: Implementation of search capabilities for the Project Kiosks for physical stores. • Elastic Search: • 3 x Master / Data Nodes • Logstash: • 1 x Logstash Node; • Data ingestion: • 1 x Elastic Pipeline configuration: • Data Collection, through the Logstash HTTP REST input plugin; • Data parsing; • Implementation of two indexes that will swap on each update, in order to avoid service outage
- 14. Solution DATA LAYERPRODUCT INFORMATION INGESTION LAYER Elasticsearch Master/Data Node 1 Elasticsearch Master/Data Node 2 Elasticsearch Master/Data Node 3 KIOSK Logstash Node
- 15. Solution
- 16. Solution 2nd Project: E-commerce Search Engine – Implementing an Elasticsearch solution, in order to provide an efficient product search engine to increase sales. The Search component used a Elasticsearch Cluster deployed on Azure with the following characteristics: • Elasticsearch: • 3 x Master / Data Nodes • Two indexes that will swap on each update, in order to avoid service outage • Logstash: • 1 x Logstash Node
- 17. Solution • Search Engine Requirements: • Research Term; • Search Criteria; • Sorting Criteria: • Clicks Number; • “Promo” Price; • Etc.. • Tiebreaker; • Research Weight; • Ordination Weight; • Total Product Weight.
- 18. Solution DATA LAYERINGESTION LAYER Elasticsearch Master/Data Node 1 Elasticsearch Master/Data Node 2 Elasticsearch Master/Data Node 3 Website Logstash Node Product Information Physical store Sales Information Online Store Sales Information
- 19. Solution
- 20. Solution • Implemented with Open Source Solution and Syone expertise • MVP + Project had a combined effort of 65 FTE’s • A small investment (only services). • The elastic cluster is also used to infrastructure logging • Data integration layer for the website(data + content) Start 3rd Month1st Month 2nd Month
- 21. Benefits
- 22. • Personalized search (250K products within Kiosks and the Website) • Fully deployed on Azure with "unlimited scalability" • Enhanced performance (response time around 100ms on the website) • Cost reduction around 40% for overall business case • Introduce innovation Benefits
- 23. Next Steps
- 24. Next Steps • Subscriptions acquisition to take advantages of the following features: • Graph; • Machine Learning; • APM; • Security; • Alerting; • Analytics; • Support; • Innovative Use Cases • Next Best Offer • Similar Products Presentation
- 25. THANK YOU!
- 26. supported by 2019 powered by syone