Abstract Easiness of navigation within a website is an important factor for information seeking performance. Several cognitive models exist that simulate the web-navigation process and these models in turn can be useful in supporting information seeking behavior. In this chapter we first discuss previous work we did on further developing a cognitive model of web-navigation CoLiDeS (Comprehension-based Linked model of Deliberate Search) that takes information from pictures into consideration, next to information from hyperlinks. This model is called CoLiDeS + Pic. Just like its parent model CoLiDeS, it uses Latent Semantic Analysis to compute semantic similarity in order to measure the information scent of hyperlinks available on a page. Next, we propose a new model CoLiDeS ++ Pic that adds path adequacy (with information from both hyperlinks and pictures) and applies backtracking. We hypothesize that in this way the information seeking process can be better modeled when compared to the previous model CoLiDeS + Pic. This was verified by simulating the model on a mockup website and comparing the results with the previous CoLiDeS + Pic model. The results support our hypothesis. We also present briefly the results of an experiment with tool-support based on the new model CoLiDeS ++ Pic. The results prove that model-generated support is fostering information seeking performance and helps in search tasks. We further discuss the challenges and advantages of automating navigation support using the proposed model.