This study, drawing on insights from the Appraisal framework, the parameter-based approach to evaluation and corpus linguistics, investigates the evaluative language used in customer review texts. The primary goal of this investigation is to develop a framework of evaluation that can be used to account adequately for evaluative expressions in customer review texts, and the ultimate goal is to support the argument that the modelling and theorising of evaluation is context-specific. Based on the investigation into a corpus compiled of review texts retrieved from www.amazon.co.uk, this study proposes a data-driven, parameter-based and appraisal-informed framework of evaluation which comprises four parameters—Quality, Satisfactoriness, Recommendability and Worthiness. Since these parameters are not thought-up, but are generalised from real data, it is arguable that the proposed framework of evaluation is certainly valid and thus can be used to describe and analyse evaluative language used in this particular context. This in turn indicates that the description and theorising of evaluation is indeed highly dependent on the discourse type that is under examination.