Impact assessment methods are regularly updated to reflect the newest scientific advances, the inclusion of new mechanisms of cause-and-effect, newer data, etc. Ecoinvent uses the results published by method developers to calculate the impacts of the calculated life cycle inventories (LCI).
The implementation of impact assessment methods is not always straightforward and unambiguous. The method developers often calculate characterisation factors (CFs) with assumptions that are not compatible with the results of LCIs.
For example, some methods require precise knowledge of geographical location of an emission to select the appropriate CF.
Other methods ask for information about some properties of the emissions, like water quality. Some methods characterise for example dissipative use of water or materials, when this information is not directly present in the LCI.
Method developers often use different naming conventions than ecoinvent for chemicals, land use and material extraction elementary exchanges. In most cases, CAS number can help, but these are not always present, or for some substances like pesticides, commercial names are ambiguous, contain a mix of active substances and therefore report more than one CAS number.
Some methods do not carry CFs for a substance in all the sub compartments present in ecoinvent. For example, should a chemical emission to groundwater receive the same CF as the same emission to surface water, if the groundwater CF is not explicitly mentioned in the method?
Different actors (method developer, database managers, software developer) might have a different opinion about how to resolve ambiguities.
Finally, plain old human mistakes is bound to happen when managing and aggregating a large quantity of information coming from disparate sources.