A carregar...
Publicação
Drivers and prediction of organic search engine CTR: Improving and understanding predictions through model interpretation
| Nome: | Descrição: | Tamanho: | Formato: | |
|---|---|---|---|---|
| 2.45 MB | Adobe PDF |
Autores
Orientador(es)
Resumo(s)
33% of web traffic in the $5.7 trillion e-commerce industry originates from organic search
engine results. Thus, website providers benefit from a holistic understanding of the drivers of
click-through rates (CTR) on organic searches to increase traffic. However, providers face a
knowledge gap, as existing literature focuses on position as the primary CTR influence,
disregarding other result page characteristics. To solve this problem, I use an extensive dataset
comprising organic Google result page information. I conduct an elaborate data analysis
highlighting the impact of four categories of result page characteristics before determining
suitable CTR prediction modeling techniques. I discover novel patterns impacting CTR for
each category and find tree-based models to outperform state-of-the-art deep-learning models.
Additionally, by interpreting the XGBoost model, I find potentials for model improvements,
quantify the relevance of result page characteristics, and discover further drivers of CTR.
Descrição
Palavras-chave
Organic click-through rate CTR prediction E-commerce SHAP