Natural gas is an important source of energy. In some cities, up to 70% of its demand is created by households using simple heating systems and boilers that are not equipped with sophisticated tools to analyze gas properties. Earlier studies as well as measurements performed within this work indicate that gas fuel quality can vary by 9–12% in short periods of time (hours) and up to 24% daily. This is due to the influence of weather conditions, gas composition, and variable demand in local areas of the low-pressure gas distribution network. While the usage of lower-quality gas is related to higher heating costs, an artificial neural network based heuristic for gas quality estimation and two-step heating strategy are proposed to enable the control of a simple boiler so that heating mostly occurs when the gas quality is very good. Simulations performed on real-life data suggest that a basic version of the proposed system can achieve, on average, a 4.3% improvement in selecting heating time slots with higher-quality fuel.