Situation

Erik's DeliCafé lost roughly 179,000 guest orders over three years, an 8.1% decline in traffic, while Google and Meta ads ran the entire time. The prior agency reported a 27.12× return on ad spend through Google Analytics. The POS told a different story.

Approach

Kayla Dillon, fractional CMO at Torched Collective, initially stopped all advertising -- and nothing changed. brought SignalFlare and nCentiv together. SignalFlare mapped Erik's actual customers to the sepcific census block groups where they live and work. nCentive ran the same Google & Meta channels as the prior agency, with similar creative, against only those neighborhoods. The channels didn't change. The targeting did.

Results

Across 13 stores over four weeks, ~$8K in ad spend produced +$72K in incremental revenue and ~3,300 additional guest orders. A 9.04× return on ad spend, verified at the register, not estimated by attribution software. Every store was above its pre-period baseline in every week of the program. Two test groups launched two weeks apart returned nearly identical ROAS (9.32× and 7.92×), which is the clearest evidence in the data that this is repeatable methodology, not a single lucky campaign.