Ad ROAS is a bullshit measure that compares revenue with cost. It fails because we all understand that advertising has diminishing returns otherwise our ad budgets would be infinite. We need a measure that reflects the value of the next dollar spent and hence know when to stop spending more when incremental/marginal returns go negative. The first step in the journey to measuring ad Incrementality is to turn off ads and see what sells anyway. You need to monitor sales rates and ranks to see absolute and relative effects vs competitors. When/if you get jumpy turn Ads back on and you'll understand their effect (or otherwise) and get another data point. A second step is to understand branded vs generic search terms' effects. With branded search your customer is evidently predisposed to your brand already and you will likely get the sale anyway so I'd suggest that the incrementality is less on branded search. Turn off branded terms and see what happens. If you compare CVR on branded vs unbranded search, you'll find it is 2-3x more. Hence adjust down the reported ad Revenue on branded search by a factor of 2-3 to reflect incrementality. A third step is to see whether paid ads change your overall share of shelf. If you are already #1 in organic SERP results on important terms and have other highly visible products following right behind, then advertising is not going to be doing much to increase your SERP/shelf presence. Unfortunately Amazon is muddying the water here by reserving whole rows/particular positions for sponsored brand or product ads so others may creep in unless you defend.
