Start with a narrow test. Choose products where faster delivery could plausibly change the buying decision: urgent household items, replacement products, replenishment lines or products where competing offers already have a faster promise.
For each SKU, record the current conversion rate, contribution margin, average order value, return rate and sales by delivery area. Then model the extra per-unit SSD charge as a cost of the orders that actually use the service. Include storage, fulfilment, advertising, returns and discounts in the same calculation. Do not compare the proposed charge only with the product’s gross margin.
The decision rule is simple. The incremental contribution from additional orders must exceed the SSD cost and the operational risk of moving more inventory into a faster network. If the product sells well already, the paid placement may simply charge you for demand that would have arrived anyway. If the product is slow-moving, speed may not solve the real problem, which could be price, listing quality, reviews or weak demand.
Amazon’s own FBA material points sellers towards its revenue calculator when estimating fulfilment costs. Use the calculator and your actual account data, not a generic marketplace estimate.