The differential abundance analysis method can improve the precision of stellar chemical abundances. The method compares the equivalent widths of a certain line in a star with the same line in a star considered to be a standard representative of its class, using high resolution and high signal to noise ratio spectra. The method has achieved great results by reducing the measurement errors to unprecedentedly low levels. However, to date, there has not been a consistent analysis on the actual improvements of this method when compared to a classical analysis in metal-poor stars. Here we present a comparison between the errors of a classical stellar analysis and a differential analysis among low-metallicity stars.