Winning back lapsed buyers with Amazon DSP for 31% incremental revenue

On paper the account looked healthy: a consumable product, a 38% repeat rate and Sponsored campaigns running at a comfortable ACoS. The problem only showed up in cohort data — buyers from any given month kept reordering for about two cycles and then disappeared. Nothing was broken; they simply forgot, and a competitor's coupon was easier to find than last year's order.

Sponsored ads could not fix that. They reach shoppers who are already searching, and a lapsed buyer of a household consumable is by definition not searching. Every extra dollar in Sponsored Products bought more of the same in-market traffic at a worse rate, which is what the brand's “we've hit a ceiling” feeling actually was.

We split the job between two channels instead of asking one to do both.

• DSP for demand that isn't searching — audiences built on purchase recency in Amazon Marketing Cloud: buyers past their expected reorder window, and shoppers who viewed the product and never converted.
• Creative built for recognition, not discovery — the ads showed the product the shopper already owned, with the reorder as the single action.
• Sponsored ads left to close — PPC budget stayed on high-intent terms rather than chasing the awareness job DSP had taken over.
• Measurement on incrementality — DSP was judged on new-to-brand and reorder lift, not on last-click ROAS, so the two channels were not credited for the same sale.

Monthly revenue grew from $310k to $406k — a 31% increase that Sponsored ads alone had not been able to buy — at a DSP ROAS of 4.8x.

The number the brand cares about more is the repeat rate: 38% to 47%, which means the customers it had already paid to acquire are now worth roughly a quarter more each. Blended TACoS moved by a single point, so the growth was not bought by loosening efficiency — it came from a part of the funnel the account had never addressed.