Benchmark settings establish thresholds with which to gauge the statistics created by a data rule, rule set, or metric. They can be used to identify a minimum level of acceptability for the data or a tolerance for some level of exceptions
A benchmark is a quantitative quality standard that you define. When you set a benchmarks for a data rule, all statistics from the rule can be measured against the benchmark. Differences between the benchmark and the statistic against which it is measured is called a variance. On an ongoing basis for data monitoring, the variances from the established benchmark identify whether a data rule has passed or achieved its target, or if it has failed its target. You can apply benchmarks to any of the statistics resulting from data rule tests or a metric value computations. You typically apply benchmarks to data rules or metrics as part of a data quality monitoring process.
For data rules the benchmark criteria can be based on a count or percentage of records, and determines if your data meets, exceeds or fails benchmark criteria. A positive or negative variance to the benchmark value is also calculated. Using percentage statistics to normalize the trends, you can evaluate the performance of your data against standards and analyze the deviations over time.