Swiftype to Elasticsearch Serverless Migration For An Airline Group
Client
The Client is a major American airline group modernizing its enterprise search infrastructure and multi-domain content strategy.
Project Context
The project successfully migrated a legacy Swiftype instance to a modern Elasticsearch Serverless architecture to eliminate technical constraints and reduce operational costs. The architecture utilized a hybrid ingestion layer—integrating the Elastic Open Web Crawler with custom Python scripts for Contentstack- to achieve a unified, high-precision search experience across 5+ web domains and five global languages.
Project Objectives
- Replace managed legacy systems with a serverless architecture for better scalability.
- Leverage advanced ranking algorithms to achieve higher precision and lower search latency.
- Support a hybrid ingestion model for both legacy web pages and new Headless CMS content.
Challenges
- Overcoming the limitations and high operational costs associated with the legacy managed Swiftype platform.
- Consolidating content silos from 5+ different web domains into a centralized index.
- Managing a dual-stream data pipeline to handle both legacy web pages and new Headless CMS content simultaneously.
- Delivering high-precision search results across five different languages (English, Italian, Spanish, Korean, and Japanese)
Solution
- Hybrid Ingestion: Deployed Elastic Open Web Crawler for domain scanning and custom Python scripts for Contentstack CMS integration.
- Configuration & Tuning: Established Index Templates, custom Ingest Pipelines, and relevance tuning using Synonyms and Query Rules.
- API Integration: Developed dedicated Autocomplete and Search APIs to provide instant suggestions and full-text results to the frontend.
