Calibrated weights are created to (a) reduce the nonresponse bias; (b) reduce the coverage error; and (c) make the weighted estimates from the sample consistent with the target population in terms of certain key variables. This technical report details our calibration analysis of singlelocation retailers for the Retailer Survey on the Cost of Payment Methods. We first compare two types of calibration approaches, consisting of (1) traditional calibration, in which calibration is implemented after explicit nonresponse modelling, and (2) nonresponse-embedded calibration, where the nonresponse correction is automatically built in (Särndal and Lundström, 2005). After carefully selecting auxiliary variables, we find minor differences between these two methods. We also examine the effects of trimming, sample size, smoothing and influential units on the calibrated weights, and show that our calibration is robust in view of these considerations.