$($a$)$test for difference in means between test vs controls for both products $($watch the # of stores$),($b$)$provide a unit sales forecast with and without ads $($use the "sales potential" information and the "represented X$\%$of sales" information$),($c$)$estimate profits with and without ads for Arizona and California $($assume they are running the same ad that was used in the experiment, ie treat the development of the ads as a sunk cost, and the $3$M ad budget in this roll$-$out is the national total, i$.$e$.$you need to adjust it for Arizona and California combined based on population $($tracks with the retailer's national sales percentages$))$