In this paper, we investigate how to provide good transmission quality in LoRa networks by considering all parameters such as device heterogeneity, network topology, deployment density, and traffic intensity. We propose three approaches to adaptation of LoRa parameters to different scenarios. Enhanced Link Adaptation (ELA) improves the classical Adaptive Data Rate (ADR) algorithm by considering network load. The ELA scheme selects the configuration according to the network parameters (e.g., network congestion) and the radio environment (e.g., link budget), and optimizes single gateway capacity by taking into account current load and capacity of each Spreading Factor (SF) allocation. The second approach exploits network densification by increasing the number of gateways. It maximizes the overall capacity of the network for different SF configurations. Finally, the joint Multi-Gateway ELA (MG j ELA) strategy improves transmission quality by exploiting SF orthogonality, spatial reuse of communications, and optimized gateway capacity.