Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Evaluating the Use of LLMs for Automated DOM-Level Resolution of Web Performance Issues

Gideon Peters, SayedHassan Khatoonabadi, Emad Shihab

TL;DR

This work investigates nine state-of-the-art LLMs for automated DOM-level resolution of web performance issues. By extracting DOM trees from 15 popular pages, generating Lighthouse audits, and iteratively applying LLM-driven modifications, the study isolates DOM-level effects from server-side factors. It finds that while LLMs reliably fix SEO and accessibility issues, performance-critical edits yield mixed results and frequently degrade visual stability, with some models even regressing latency. The results highlight the potential of AI-assisted DOM optimization but emphasize the need for hybrid workflows with rigorous post-hoc validation and careful model selection to ensure safe, reliable production deployment. The study also provides a reproducible benchmark and detailed insights into DOM modification patterns to guide future research and tooling.

Abstract

Users demand fast, seamless webpage experiences, yet developers often struggle to meet these expectations within tight constraints. Performance optimization, while critical, is a time-consuming and often manual process. One of the most complex tasks in this domain is modifying the Document Object Model (DOM), which is why this study focuses on it. Recent advances in Large Language Models (LLMs) offer a promising avenue to automate this complex task, potentially transforming how developers address web performance issues. This study evaluates the effectiveness of nine state-of-the-art LLMs for automated web performance issue resolution. For this purpose, we first extracted the DOM trees of 15 popular webpages (e.g., Facebook), and then we used Lighthouse to retrieve their performance audit reports. Subsequently, we passed the extracted DOM trees and corresponding audits to each model for resolution. Our study considers 7 unique audit categories, revealing that LLMs universally excel at SEO & Accessibility issues. However, their efficacy in performance-critical DOM manipulations is mixed. While high-performing models like GPT-4.1 delivered significant reductions in areas like Initial Load, Interactivity, and Network Optimization (e.g., 46.52% to 48.68% audit incidence reductions), others, such as GPT-4o-mini, notably underperformed, consistently. A further analysis of these modifications showed a predominant additive strategy and frequent positional changes, alongside regressions particularly impacting Visual Stability.

Evaluating the Use of LLMs for Automated DOM-Level Resolution of Web Performance Issues

TL;DR

This work investigates nine state-of-the-art LLMs for automated DOM-level resolution of web performance issues. By extracting DOM trees from 15 popular pages, generating Lighthouse audits, and iteratively applying LLM-driven modifications, the study isolates DOM-level effects from server-side factors. It finds that while LLMs reliably fix SEO and accessibility issues, performance-critical edits yield mixed results and frequently degrade visual stability, with some models even regressing latency. The results highlight the potential of AI-assisted DOM optimization but emphasize the need for hybrid workflows with rigorous post-hoc validation and careful model selection to ensure safe, reliable production deployment. The study also provides a reproducible benchmark and detailed insights into DOM modification patterns to guide future research and tooling.

Abstract

Users demand fast, seamless webpage experiences, yet developers often struggle to meet these expectations within tight constraints. Performance optimization, while critical, is a time-consuming and often manual process. One of the most complex tasks in this domain is modifying the Document Object Model (DOM), which is why this study focuses on it. Recent advances in Large Language Models (LLMs) offer a promising avenue to automate this complex task, potentially transforming how developers address web performance issues. This study evaluates the effectiveness of nine state-of-the-art LLMs for automated web performance issue resolution. For this purpose, we first extracted the DOM trees of 15 popular webpages (e.g., Facebook), and then we used Lighthouse to retrieve their performance audit reports. Subsequently, we passed the extracted DOM trees and corresponding audits to each model for resolution. Our study considers 7 unique audit categories, revealing that LLMs universally excel at SEO & Accessibility issues. However, their efficacy in performance-critical DOM manipulations is mixed. While high-performing models like GPT-4.1 delivered significant reductions in areas like Initial Load, Interactivity, and Network Optimization (e.g., 46.52% to 48.68% audit incidence reductions), others, such as GPT-4o-mini, notably underperformed, consistently. A further analysis of these modifications showed a predominant additive strategy and frequent positional changes, alongside regressions particularly impacting Visual Stability.
Paper Structure (23 sections, 2 equations, 2 figures, 4 tables)

This paper contains 23 sections, 2 equations, 2 figures, 4 tables.

Table of Contents

  1. Introduction
  2. Background
  3. DOM Trees
  4. Performance Audits
  5. Study Design
  6. Dataset
  7. LLM Selection
  8. DOM Chunking
  9. Performance Audits
  10. Environment Configurations
  11. Benchmarking & Evaluation Metric
  12. RQ1: How effective are LLMs in resolving web performance issues through automated DOM-level modifications?
  13. RQ2: What is the nature of changes made by LLMs for automated DOM-Level resolution of web performance issues?
  14. Discussion
  15. Proficiency in Semantic Understanding vs the Pervasive Challenge of Visual Stability
  16. ...and 8 more sections

Figures (2)

  • Figure 1: Example structure of a DOM tree
  • Figure 2: Overview of our experiment workflow