.png){kind=logo}
Pet-care e-commerce has risen sharply, with digital spending on pets outpacing traditional retail and reshaping customer expectations. Owners want reliability in everything from nutrition deliveries to insurance claims, and they judge brands not only by product selection but by the consistency of their digital support. In this environment, every missed interaction triggers downstream costs, operational strain and, most importantly, a loss of trust.
For Senior Data Architect Digvijay Waghela, who leads data-platform and customer-journey initiatives at Chewy after building large-scale data systems at major tech giants, the core problem is simple to state and hard to solve. Customer service teams sit on enormous volumes of telemetry, yet agents often see only fragments of the story.
“Support should not start when a customer calls,” Waghela says. “It should begin the moment a pattern in their journey tells you that something is about to go wrong.”
That principle now guides how he thinks about consumer technology more broadly, from pet-care platforms to gaming and streaming services. The lens is Chewy’s Customer Service Recommendation Engine and Customer Journey Data Lake, but the argument extends to any company that wants to treat support as a product, not a cost line.
Most large-scale consumer companies inherited architectures where browsing, orders, loyalty programs and support histories live in different systems. Dashboards look polished, but the underlying data never converges. In pet care, where shipments involve time-sensitive nutrition, medication and wellness products, that fragmentation translates directly into customer anxiety.
At Chewy, Waghela led the transition of customer-service analytics away from a constrained warehouse into an S3-based data lake with clear raw and refined states catalogued through AWS Glue and queried through Athena. Rebuilt Tableau dashboards drew directly from this fabric, shrinking latency and eliminating the inconsistencies that came from point-to-point reporting. This shift mattered at scale, where more than eighteen million customer contacts move through the system each year. Dashboard latency dropped by roughly 80-90%, giving every operational and customer-experience team a shared, real-time view of what pet parents were actually doing across channels
This shift created more than operational efficiency. By linking search behaviour, Autoship adjustments, CarePlus interactions, support reasons, returns and pet profiles, the platform gained a continuously updated narrative of each customer. “When data behaves like a fabric, the system can finally surface intent instead of noise,” Waghela says. “You stop asking what happened and start asking why it happened.”
The outcome is a foundation where machine learning, analytics and frontline service all speak a common language.
Once the architecture produced a unified journey view, the next challenge was applying that intelligence inside live service conversations. Context-aware support improves first-contact resolution and reduces unnecessary volume, but only when the system understands a customer’s full behavioural history, not isolated moments.
Waghela and his collaborators built a Customer Journey Feature Store in Glue and PySpark that turned raw interactions into structured signals-journey transitions, subscription rhythms, friction patterns and indicators of satisfaction or decline. These signals powered SageMaker models for contact propensity, recommendation and uplift.
Recommendations surfaced as a focused, ranked set of options with clear reasoning, not aggressive prompts. Autoship adjustments, relevant food suggestions, wellness choices and CarePlus guidance appeared only when supported by context. “Intelligence should reduce effort, not add friction,” Waghela says. “If a recommendation cannot justify itself in the moment, it does not belong in the interaction.”
The impact was immediate. Contact volume dropped by about 12%, from eighteen million to roughly fifteen point eight million annual interactions. Agent-assisted recommendations climbed from forty-two million dollars to sixty-two million dollars, a 45% uplift. Autoship and CarePlus adoption rose in double digits once suggestions matched each customer’s journey. With a clearer view of what mattered, agents eliminated unnecessary contacts and strengthened decisions that influence loyalty—spotting likely shipment issues, interpreting ambiguous behaviour or recognising when confusion, not intent, was driving the interaction.
Beyond customer service, Waghela’s approach reflects a broader shift toward platforms that optimise themselves. His UK-registered design, granted under number 6430209, describes a computing architecture that embeds AI directly in data-engineering layers so pipelines can tune workload sequencing, configuration and resource allocation autonomously.
Static tuning cannot keep pace with modern consumer platforms, where demand spikes, product changes and support surges occur too quickly for manual intervention. The patented design proposes a continuous feedback cycle in which telemetry informs how pipelines adapt, ensuring stability without delaying downstream experiences. Security, anomaly detection and governance are integrated into the processing path so that performance gains never compromise compliance or safety.
For Waghela, this philosophy extends naturally to customer-journey systems. The same design principles behind his patented architecture also shaped Chewy’s operational gains, where reduced contacts, faster agent decisions and stabilised data flows translated into multimillion-dollar savings across staffing and support overhead. “If your platform only runs well during predictable traffic, it is not engineered for real customers,” he says. “Platforms should notice when behaviour changes and adjust before the customer feels the impact.”
Consumer-technology companies from e-commerce to digital media face similar pressures. As systems take on more personalisation, more real-time inference and more complex support flows, resilience cannot rely on human vigilance alone. Architecture must internalise the responsibility to monitor, correct and adapt.
The pet-care industry is often underestimated in its technical complexity. It deals with time-sensitive fulfilment, regulated health-adjacent products, long-term subscription cycles and emotionally charged user expectations. That combination exposes every weakness in a platform’s data stack.
This is why Waghela’s work has relevance beyond Chewy. It shows that customer support is no longer an operational afterthought; it is an architectural discipline. Systems need a unified journey fabric, context-aware intelligence for frontline teams and adaptive infrastructure capable of tuning itself as customers evolve.
“Trust is built when the system recognises the customer’s path and responds with clarity,” Waghela says. “The architecture sets that tone long before a model or an agent speaks.”
As pet-care e-commerce grows alongside gaming, streaming and digital services, the advantage will belong to companies that treat each customer journey as an interpretable, actionable narrative. Platforms that can read, predict and adapt will define the next era of customer loyalty, while those that rely on fragmented systems will fall behind regardless of how polished their interfaces appear.
