An AI-Driven Framework for Assessing Website Trustworthiness Through Code Quality Metrics and Multi-Criteria Decision Analysis

Abstract

The trustworthiness of websites has become one of the most important factors of the search engine optimization (SEO) performance but there is no common and data-supported methodology which combines code quality assessment with multi-criteria decision analysis (MCDM). This paper proposes an AI-based framework that links these two domains together by (i) operationalizing website trustworthiness and identifying a set of quantifiable code quality and web performance metrics from SimilarWeb, Ahrefs, Moz, SEMrush, and Google PageSpeed Insights, and (ii) evaluating and ranking websites against the aspiration level through a hybrid MCDM model that combines Decision-Making Trial and Evaluation Laboratory (DEMATEL), DEMATEL-based Analytic Network Process (DANP), and VlseKriterijumska Optimizacija I Kompromisno Resenje (VIKOR). The proposed framework is shown to produce trust-orderings that are highly correlated (at r = 0.91) with existing trustworthiness measures in an empirical study of fifteen representative websites in three content domains: health information, e-commerce, and financial services. The DEMATEL analysis shows that the dimensions with the greatest range of outward influence on the network of trust are domain authority and HTTPS compliance, while the dimensions with the greatest gap for improvement are page load time and citation integrity, according to the VIKOR gap analysis. It can be adapted to AI-driven SEO pipelines and offers website owners and SEO experts measurable, criteria-based website improvement suggestions.