In many places, gillnet fishing is considered a conservation threat for air-breathing marine species. Fish pots represent an alternative to gillnetting; however, due to their low catch efficiency pots are rarely taken up by commercial fisheries. To improve pot efficiency for Atlantic cod (Gadus morhua), we used a novel enclosure to observe cod interacting with pot entrances, and investigated several entrance design parameters, including funnel colour, entrance funnel presence, length and entrance form. We demonstrate that the key factor for entrance passage is to give cod an unobstructed view of the inside or outside when they try to enter or exit the pot, respectively. Funnel colour (colours tested: white, green and transparent) influences entrance passage rates, with significantly higher entrance passage rates for the transparent funnel. Funnel presence increases the entrance encounter rate by enlarging the outer opening of the entrance. It decreases exit rates by deflecting cod away from the inner entrance opening and by reducing the area in which the exit is perceptible to cod inside the pot. Increasing funnel length further reduces this area and may deter cod by the longer passage length. This is the first study to observe cod–pot interactions day and night using an infrared camera, revealing a pronounced diurnal pattern with few nocturnal entrance passages, suggesting that cod–pot interactions are primarily guided by vision. The findings underline the importance of funnels and reveals promising avenues for their further improvement, e.g., by using transparent fish retention devices. The new pen-based method is superior in several ways to conventional field-pot catch-rate comparisons: It allows identification of differences in catch efficiency and describes the underlying cod behavioural mechanism leading to these differences. Thus, it allows targeted, efficient and iterative cod-pot catch-efficiency enhancements.