PurposeEnvironmental rating ecolabels (ERE) are developing rapidly across Europe, intending to guide sustainable purchasing choices through performance ratings derived from life cycle assessment (LCA). Integrating robust data quality methods in the LCA modelling stage is essential for accurate and comparable ratings, but challenging as ERE must be feasible to generate with reasonable effort for broad adoption. The use of secondary life cycle inventory (LCI) datasets, alongside actual product information such as bills of materials, is inevitable. However, the selection process for secondary LCI datasets and its implications for environmental ratings have yet to be investigated.MethodsWe examine the context-dependent suitability of secondary LCI datasets (aka their representativeness), hypothesising that differing levels of representativeness could lead to differences in product ratings. In a case study involving 94 laundry detergent products, we investigate how different secondary dataset choices influence the resulting aggregated LCA results (single scores), relative product rankings, and environmental ratings. Four distinct data scenarios are defined, in which the production of 19 key ingredients is represented through datasets of varying representativeness (from most specific to most generic).ResultsOver 60% of the products studied obtained a different rating under at least one of the three lower representativeness (more generic) data scenarios explored as compared with the rating obtained when the highest representativeness (most specific) data scenario was used. Lower representativeness scenarios also led to wider distributions of product single scores, with scenarios which utilised LCI data representing the function of ingredients (e.g. alkalinity source) or generic LCI data (e.g. organic chemicals) showing the most significant deviations from the distributions under the specific data scenario. This highlights the importance of having clearly defined, consistent data quality management approaches for the LCA stage of ERE.ConclusionsProduct information (e.g. formulation and packaging specifications) from primary sources should be combined with representative LCI datasets from third party databases to obtain reliable product ratings. We demonstrate that the representativeness of secondary datasets can be evaluated through three contextual indicators of data quality (technological, geographical, and temporal representativeness) to help guide the data selection process aligned to defined data quality thresholds. Based on our findings, practical recommendations are provided for data quality management in ERE. Scheme developers are encouraged to embed these in their methodologies to underpin transparent, robust labelling for product comparison.