Imprecise control of fluid flows in paper-based devices is a major challenge in pushing the innovations in this area towards societal implementation. Assays on paper tend to have low reaction yield and reproducibility issues that lead to poor sensitivity and detection limits. Understanding and addressing these issues is key to improving the performance of paper-based devices. In this work, we use colorimetric analysis to observe the mixing behaviour of molecules from two parallel flow streams in unobstructed (on unpatterned paper) and constricted flow (through the gap of a patterned hourglass structure). The model system used for characterization of mixing involved the reaction of Fe3+ with SCN- to form the coloured, soluble complex Fe(SCN)2+. At all tested concentrations (equal concentrations of 50.0 mM, 25.0 mM or 12.5 mM for KSCN and FeCl3 in each experiment), the reaction yield increases (higher colorimetric signal) and better mixing is obtained (lower relative standard deviation) as the gap of the flow constriction becomes smaller (4.69-0.32 mm). This indicates enhanced passive mixing of reagents. A transition window of gap widths exhibiting no mixing enhancement (about 2 mm) to gap widths exhibiting complete mixing (0.5 mm) is defined. The implementation of gap sizes that are smaller than 0.5 mm (below the transition window) for passive mixing is suggested as a good strategy to obtain complete mixing and reproducible reaction yields on paper. In addition, the hourglass structure was used to define the ratio of reagents to be mixed (2 : 1, 1 : 1 and 1 : 2 HCl-NaOH) by simply varying the width ratio of the input channels of the paper. This allows easy adaptation of the device to reaction stoichiometry.