Flow-based market coupling provides zonal day-ahead markets with appropriate signals of possible real-time congestions by incorporating information on local load and generation patterns. It relies on predictive parameters, notably the base case and generation shift keys. Also it only monitors part of the grid, through selecting critical network elements. In consequence, it naturally falls short of a nodal pricing-based cost-optimal solution. Based on a test network with three flow-based market coupling zones, we show that the results of flowbased market coupling converge to nodal pricing solutions with an increasing amount of re-configured market zones. We identify if re-configured market zones can help to improve the selection of critical network elements and lead to cost reductions even in the original market zone setting. We find that around 90% of the cost reductions from a market zone re-configuration can be maintained when the critical network elements, obtained from the re-configured market zones, are used for the original market zones. This is a strong indication that, both in reality as well as model-based research of flow-based market coupling, the selection of critical network elements should be based on expected congestion patterns. To find these congestions, we conduct a nodal pricebased market zone re-configuration that helps to identify lines with different congestion signals. This approach can constitute a helpful addition to static and assumption-based selection criteria for critical network elements, such as the often-used zone-to-zone power transfer distribution factors that strongly rely on assumptions like generation shift keys.