Enhancing Search Engine Visibility of an E-commerce Single Page Application with Server-Side Injected Structured Data
Metsoila, Tuomas (2026)
2026
Tietotekniikan DI-ohjelma - Master's Programme in Information Technology
Informaatioteknologian ja viestinnän tiedekunta - Faculty of Information Technology and Communication Sciences
This publication is copyrighted. You may download, display and print it for Your own personal use. Commercial use is prohibited.
Hyväksymispäivämäärä
2026-06-02
Julkaisun pysyvä osoite on
https://urn.fi/URN:NBN:fi:tuni-202606016725
https://urn.fi/URN:NBN:fi:tuni-202606016725
Tiivistelmä
Modern web applications are often developed as Single-Page Applications (SPAs) using Client-Side Rendering (CSR) for dynamic user experiences. However, this architecture introduces significant challenges for technical Search Engine Optimization (SEO) and organic visibility, which are critical for e-commerce competitiveness. The primary concern is the visibility issue, where essential contents, specifically JavaScript Object Notation for Linked Data (JSON-LD) structured data, cannot be indexed by crawlers that lack JavaScript execution capabilities, including most emerging Artificial Intelligence (AI) search engines.
This study follows the Design Science Research (DSR) methodology with the primary objective of resolving the visibility issue by designing, implementing, and evaluating a hybrid rendering architecture. To set a baseline, the legacy architecture was analyzed, revealing separated and hardcoded JSON-LD generation logic across different frontend components, causing maintainability issues and schema vulnerabilities. To evaluate the artifact while accounting for seasonal traffic behavior, a comparative quasi-experiment was conducted on two geographically separated sites of a Nordic e-commerce SPA. SEO performance was measured using Google Search Console and Google Analytics 4, while web performance was monitored using Chrome User Experience Report (CrUX) API for Time To First Byte (TTFB) and the Largest Contentful Paint (LCP) to ensure that Core Web Vitals (CWV) remain stable.
The developed artifact centralized and enhanced the structured data generation logic within a backend service. By using strongly-typed models, a @graph structure, and a multi-layered caching strategy, the solution injected the JSON-LD structured data directly into the initial HTML response, removing the JavaScript dependency. The results of the quasi-experiment showed that while organic visibility and AI referral traffic remained stable and aligned with seasonal behavior, Rich Results performance experienced a significant breakthrough. Review snippet impressions increased by 168.0%, and clicks by 245.5% on the Test site. This outcome confirms that the artifact provides reliable indexing of structured data, removing the reliance on incomplete fallback data from Google Merchant Center.
The evaluation of web performance proved that the caching strategy successfully maintained the existing web performance, as both TTFB and LCP remained stable, preventing a decrease in CWV scores. While the limited sample sizes of web performance and AI traffic volumes reduce their statistical significance, success of the artifact is confirmed by the clear SEO breakthrough and the architectural improvements. The centralized backend service improved the maintainability, type-safety, and scalability of the solution. In conclusion, this thesis confirmed the importance of server-side rendered structured data, as it managed to increase the indexing reliability, and future-proof the solution for AI-driven search engines.
This study follows the Design Science Research (DSR) methodology with the primary objective of resolving the visibility issue by designing, implementing, and evaluating a hybrid rendering architecture. To set a baseline, the legacy architecture was analyzed, revealing separated and hardcoded JSON-LD generation logic across different frontend components, causing maintainability issues and schema vulnerabilities. To evaluate the artifact while accounting for seasonal traffic behavior, a comparative quasi-experiment was conducted on two geographically separated sites of a Nordic e-commerce SPA. SEO performance was measured using Google Search Console and Google Analytics 4, while web performance was monitored using Chrome User Experience Report (CrUX) API for Time To First Byte (TTFB) and the Largest Contentful Paint (LCP) to ensure that Core Web Vitals (CWV) remain stable.
The developed artifact centralized and enhanced the structured data generation logic within a backend service. By using strongly-typed models, a @graph structure, and a multi-layered caching strategy, the solution injected the JSON-LD structured data directly into the initial HTML response, removing the JavaScript dependency. The results of the quasi-experiment showed that while organic visibility and AI referral traffic remained stable and aligned with seasonal behavior, Rich Results performance experienced a significant breakthrough. Review snippet impressions increased by 168.0%, and clicks by 245.5% on the Test site. This outcome confirms that the artifact provides reliable indexing of structured data, removing the reliance on incomplete fallback data from Google Merchant Center.
The evaluation of web performance proved that the caching strategy successfully maintained the existing web performance, as both TTFB and LCP remained stable, preventing a decrease in CWV scores. While the limited sample sizes of web performance and AI traffic volumes reduce their statistical significance, success of the artifact is confirmed by the clear SEO breakthrough and the architectural improvements. The centralized backend service improved the maintainability, type-safety, and scalability of the solution. In conclusion, this thesis confirmed the importance of server-side rendered structured data, as it managed to increase the indexing reliability, and future-proof the solution for AI-driven search engines.