PanelVerdict is a synthetic-panel tool for comparing two headlines when you do not have enough real traffic to run an A/B test.
It lets you choose whether the panel should judge which headline people would prefer or which they would click. You can also describe the intended audience, while the panel itself is sampled from OECD demographic data for the United States, Germany, and Japan.
The system shows you who is in the panel before voting begins, so you can review and adjust the audience. It then asks synthetic personas to choose between the two headlines and produces a verdict, uncertainty estimates, demographic splits, and the reasons behind the votes. An analyst can also answer questions about the results and methodology using a small retrieval-based knowledge base.
