Trials run on products are typically done in university or research herds, and they may not accurately reflect how a product will work in commercial conditions. On-farm trials can more accurately reflect how a product will work with a commercial stocking density, herd health status, and general environment. Before starting an on-farm trial it should be decided which questions to answer, and determine if there will be financial benefit to the trial. Trial design is typically uses completely randomized design (CRD) or randomized complete block design (RCBD) to assign treatments to experimental units, and the replications per treatment need to be the same. CRD is the simpler option, but RCBD avoids confounding variables like starting weight or parity of sow being used in results. The number of replications depends on what level of difference needs to be detected, and the variability in the value used to measure results. A detailed trial protocol should be written up before starting the trial. Once data collection is completed, the data should be checked, summarized, and can be analyzed through programs like excel. It is often useful to analyze data in economic terms. Once the trial is complete, a more informed decision can be made for including a new product into production.