Two carts of the same value can generate different contributions. Products have different costs and margins; some combinations also require packaging or delivery. Reading only the average receipt can therefore hide important differences in the service economy.
I propose to observe frequent combinations, using categories simple enough to guide a decision. The aim is to understand which purchases are accompanied and which resources they require. A relationship between products does not, by itself, prove that proposing them together will increase the result.
In a hypothetical example, a main item is often purchased with an accessory. Making both available can facilitate the customer. A common package could, however, introduce an additional cost: the comparison must include it, along with the sales actually obtained.
The commercial test must be checked against alternatives and customer requests. I would avoid turning any pairing into an obligatory offer. The value of the analysis lies in designing an understandable proposal and recognizing the contribution of the combinations actually chosen.