The company wanted more customers booking grooming appointments at its salons. Instead of writing a new email campaign and testing it against a control group over several weeks, the retailer gave an artificial intelligence system a single objective and let it run.<\/p>\n
The AI system identified which of PetSmart\u2019s more than 80 million Treats Rewards<\/a> members were likely candidates for a grooming visit, then tested thousands of combinations of subject lines, offers, creative and send times against each one individually, continuously, rather than on a fixed schedule, Retail TouchPoints reported<\/a> Tuesday (Aug. 4).<\/p>\n The result was a 22% lift in incremental grooming salon bookings and a 13% lift in autoship transactions, according to a case study<\/a> published by Databricks<\/a>.<\/p>\n PetSmart built the system on Databricks, which holds the retailer\u2019s first-party customer data, and Hightouch<\/a>, which turns that data into live marketing actions.<\/p>\n The retailer\u2019s marketing team no longer selects subject lines, offers or send times directly. Instead, they define an objective, set guardrails around approved offers and creative, and let the AI system determine which specific message and timing is likely to work for each customer, according to the Retail TouchPoints report.<\/p>\n The shift from PetSmart\u2019s old approach illustrates why the change mattered operationally, not just financially.<\/p>\n \u201cFor years, our teams made thoughtful, data-driven decisions around the right offer, email content and timing for every campaign,\u201d said PetSmart Senior Vice President of Marketing Bradley Breuer<\/a>, per the report. \u201cBut those decisions were still largely calendar-based and limited by the number of variables a person can realistically evaluate at once.\u201d<\/p>\n For the salon-booking initiative specifically, PetSmart set a single clear objective to identify pet parents likely to book a grooming appointment, then determine the most effective way to prompt them, the report said. Within the guardrails the marketing team defined, the AI continuously tested thousands of combinations of subject lines, calls to action, content and timing to find what worked for each customer.<\/p>\n In the Databricks case study, Hightouch described the underlying mechanism as reinforcement learning applied to marketing decisions.<\/p>\n \u201cMarketing teams can set clear goals and provide an action space for potential offers or messages while the system figures out the optimal treatment for each customer,\u201d Hightouch co-CEO Tejas Manohar<\/a> said.<\/p>\n Loyalty Programs Are Becoming Allocation Engines for Marketing Spend<\/p>\n PetSmart\u2019s Treats Rewards program has scaled since a 2024 relaunch designed to reward its most loyal customers with tiered benefits, growing from 67 million to more than 75 million members in the year following the redesign, Modern Retail reported<\/a> in 2025. More than 90% of PetSmart\u2019s transactions now trace back to a Treats Rewards account, giving the retailer a first-party data foundation broad enough to make individualized AI decisioning worth the investment.<\/p>\n PetSmart\u2019s approach fits a pattern showing up across retail. Ulta Beauty<\/a> centralized customer data<\/a> from emails, loyalty programs and in-store purchases into unified profiles, then used AI to predict what a customer would do next and adjust recommendations in near real time, driving a 95% customer repurchase rate.<\/p>\n Avery Miller<\/a>, vice president and head of global loyalty for value-added services at Visa<\/a>, and Kipp Johnson<\/a>, senior director of AI solutions at Braze<\/a>, told PYMNTS<\/a> in June that personalization<\/a> has become table stakes in loyalty, with the real differentiator now resting on context, relevance and the ability to orchestrate individualized customer journeys across an entire relationship rather than a single campaign.<\/p>\n PetSmart\u2019s results suggest what that orchestration looks like in practice. A marketer\u2019s task shifts from choosing the message to choosing the rules an AI system follows while making that choice automatically, continuously and for each customer individually.<\/p>\n