WhenaRetailChainStoppedTreatingItsOwnCustomersLikeStrangers
A multi-channel retail brand had loyal customers—but their systems had no idea. Online, in-store, and on mobile, the same person was a different record in a different database. We unified ten fragmented systems into one coherent customer experience. Lifetime value grew 45% in nine months.
Business Context & Telemetry
Our client was an established retail brand with 35 stores and a growing e-commerce presence. On paper, they were omnichannel. In reality, they were three separate businesses sharing a name. A customer who'd spent ₹80,000 in-store was treated like a first-time visitor online. A cart abandoned on mobile was gone on desktop. The brand had loyal customers; it just had no infrastructure to recognize them.
Established mid-to-large scale retail brand
240 people across retail, e-commerce, and tech
Pan-India, 35 stores across 12 cities
Web, iOS, Android, In-store POS, Internal Operations Dashboard
2009
“We talk about our customers like they're one person. But our systems treat them like four different strangers depending on where they're shopping. The gap between those two things was getting harder to explain.”
Chief Digital Officer
A brand that knew its customers. Systems that didn't.
Retail fragmentation builds quietly. Each channel gets its own system. No one coordinates them. Then one day, the CEO tries to look up a customer's full history and discovers it doesn't exist anywhere in one place. That's usually when someone calls us.
Ten systems, ten versions of the truth
Customer data was trapped in the e-commerce platform, POS system, CRM, loyalty database, and six other silos. The same person existed under different IDs, with different purchase histories. Unifying them wasn't just tech; it was an archaeology project.
The 'Start Over' penalty
A customer would build a cart online, get distracted, open the app later, and find it empty. Every channel transition was a dead end. Exit surveys had one consistent, frustrated theme: 'I had to start over.'
Loyalty that felt like a lie
Points earned in-store took days to appear in the app. Online redemptions didn't work at the physical POS. Customers who had invested in the program lost trust when it failed to work seamlessly.
Personalization that was actually just noise
Without a unified view, personalization was tone-deaf. The system would recommend a category a customer had bought dozens of times, or worse, promote an item they had just returned.
Store staff flying blind
When a loyal customer walked into a store, the associate had no context—no purchase history, no active wishlist. They wanted to provide great service but had nothing to work with but their own intuition.
They had tried a partial CRM sync, but it broke constantly and was eventually abandoned. They also built a cross-channel cart that worked between web and mobile, but not in-store. It proved the problem was far harder than they'd thought.
"The leadership knew their future growth depended on knowing their customers better than pure-play online retailers. They had years of data and a physical footprint—massive assets that were being completely wasted by their fragmented tech stack."
We didn't just unify the systems; we built a shared truth.
Every fragmented retail stack has a history. Before we wrote any code, we spent three weeks auditing the 10 systems to understand why they'd fractured in the first place. The problem wasn't the individual tools; it was the absence of a shared customer reality.
Discovery & Methods
We interviewed 24 multi-channel shoppers and shadowed in-store staff. We analyzed 2 years of behavioral data. The picture was clear: the brand had built each channel in isolation, for good reasons at the time. The customer was now paying the price for that history.
The problem wasn't bad systems. It was the lack of a shared brain.
The POS, the app, the CRM—each was doing its job. But none of them agreed on the most fundamental question: 'Who is this customer?' We needed to build a single, trusted, real-time answer to that question that every system could access.
Design Philosophy
Instead of picking one system as the 'master,' we built a dedicated customer data layer that sat *above* all existing systems. This meant no single team had to give up their trusted tool, and no catastrophic 'rip and replace' was needed. It was harder to build, but it was the only approach that would survive the organization's political reality.
Constraints Respected
- No POS Replacement: We had to integrate with the existing POS across 35 stores, not replace it.
- Future-Proof Integrations: The e-commerce platform was due for a major upgrade; our data layer had to survive it.
- Low Maintenance: The solution had to be self-monitoring, proactively alerting a stretched IT team to issues.
- Granular Consent: The architecture had to respect that a customer opting into email didn't automatically opt into cross-channel tracking.
One version of every customer—visible everywhere, updated instantly.
We built a five-layer unified commerce platform with a shared customer data backbone, giving every channel—from the e-commerce site to the in-store associate—the same, perfect memory.
Unified Customer Data Platform (CDP)
A real-time data backbone that ingests events from all 10 touchpoints, resolves them into a single customer profile, and makes that profile available to every system in milliseconds.
This is the foundation. A purchase in-store is visible on the app before the customer reaches their car. There is one source of truth, everywhere, always current.
Event-driven architecture using Apache Kafka. Probabilistic and deterministic matching solves complex identity resolution across all systems.Real-time event streaming backbone for all 10+ touchpoints
Purpose-built for sub-50ms read latency for live retail interactions
Sub-millisecond caching for profile reads during checkout and at the POS
Unified customer experience across web, iOS, and Android
Single API layer connecting all frontends to the unified data platform
Auto-scaling infrastructure to handle unpredictable retail peaks
The unified cart is invisible.
“Early prototypes had a 'Your cart is synced!' notification. Users found it distracting. We removed it. The cart is just always there, always current. The best features are the ones you don't even notice.”
In-store personalization is opt-in.
“Some customers love being recognized; others find it unsettling. We made this a clear, explicit setting in the loyalty program. When the benefit was explained clearly, 74% of customers opted in.”
Twenty weeks to untangle a decade of digital spaghetti.
Omnichannel projects fail at the edges where systems meet. We structured the entire build around those dangerous integration points, not around the individual components.
Delivery Timeline
Operational Log
Discovery & System Audit
Weeks 1–3Audited all 10 legacy systems and mapped the chaotic data models. Produced a detailed integration dependency map that became our shared bible for the project.
CDP & Identity Resolution
Weeks 4–7Built the customer data platform and the complex identity resolution engine. Ran old and new systems in parallel for two weeks to prove data parity before declaring the CDP the new source of truth.
Cart & Loyalty Engines
Weeks 8–12Built the cross-channel cart and unified loyalty system. Piloted the POS integration in 5 stores to validate real-time sync before scaling.
Personalisation & Full Rollout
Weeks 13–19Trained the personalization models on the newly unified data. Rolled out the Store Associate Panel across all 35 stores with on-site support.
Tuning & Handoff
Week 20Ran 3x peak load testing, optimized latency, and completed a structured knowledge transfer with the client's internal engineering team.
Team Topology
Deployed Roster
Collaboration
Working Rhythm
We ran a weekly integration review with the client's IT, Digital, and Retail Ops leads—all three, every week. We kept a live 'Traffic Light' dashboard showing the health of all 10 system integrations. This radical transparency turned the client's team from worried observers into active collaborators.
Course Corrections
Diagnostic Log
Identity resolution was a nightmare. The same loyal customer existed as 2-3 different people across 10 systems, using different emails, phone numbers, and guest checkouts.
We built a two-stage pipeline: deterministic matching for exact IDs, and a probabilistic model for likely matches. We built a simple UI for the client's data team to manually approve the 'fuzzy' matches. Human intuition closed the gap the algorithms couldn't.
One of the legacy inventory databases had no API and was running on ancient, fragile infrastructure. It held 6 years of vital transaction history.
We didn't risk a real-time integration. We reverse-engineered the schema and built a safe, read-only nightly batch extraction. It wasn't real-time, but it was enough to feed the personalization models without crashing the legacy system.
Store staff adoption of the new POS panel was uneven. Some managers embraced it; others saw it as 'more IT nonsense.'
We identified the highest-adopting stores and asked their managers to share their 'scripts' and best practices. We then shared these peer-to-peer success stories with the laggard stores. Adoption followed immediately.
Nine months later, the brand's best customers finally felt known.
The metrics moved strongly, but the real win was cultural. For the first time, the Net Promoter Score (NPS) for in-store and online experiences converged. Customers were finally describing both channels as if they were part of the same, single brand.
45%
Customer LTV increase
comparing 9-month post-launch and pre-launch cohorts
60%
driven by the seamless cross-channel cart continuity
30%
increase attributed to context-aware cross-sell recommendations
Qualitative Objectives Reached
- The loyalty program re-engagement rate—from members who had stopped using points—increased by 38% once the system became reliable.
- The store associate panel became an unexpected retention tool. Staff reported higher job satisfaction because they felt more empowered to provide genuinely personal service.
- The data team finally had a complete cross-channel view. They completely restructured their marketing attribution model, discovering that in-store visits were driving far more online conversion than they'd ever been able to measure.
"We've been collecting customer data for fifteen years. But until this project, we'd never actually been able to *use* it. Now when I look at a customer record, I see a person, not a partial transaction list. That sounds like a small thing. For us, it changed everything."
Chief Digital Officer
Omnichannel Retail Client
Insights Gained
Valuable lessons and strategic insights uncovered through this project that inform our future work and architectural decisions.
Identity Resolution is the unsexy foundation of everything.
Every client assumes their customer records are cleaner than they are. The messy, unglamorous work of merging duplicate identities is where most omnichannel projects quietly fail. We now treat it as its own dedicated project phase.
Retail tech adoption is a human problem wearing a technical costume.
The tech worked from day one. Whether staff used it depended entirely on how their manager framed it. The stores where managers treated it as a customer service tool saw huge adoption. The stores where it was 'a new IT thing' didn't.
The absence of friction is the most powerful feature.
The biggest driver of conversion wasn't a shiny new feature; it was the disappearance of a problem. Customers don't always notice what's fixed; they just know the experience feels better. Building that invisible reliability is what truly changes behavior.
Capabilities & Archive
Selling across multiple channels—and suspecting your customers are experiencing each one as if they've never met you before? That's the exact gap we close.
Services Leveraged
Your customers know they're loyal. Your systems should too.
Fragmented retail data is one of the most common—and most solvable—problems we see. Tell us what your stack looks like, and we'll give you an honest read on where the gaps are and what it would take to close them.
"No jargon about 'digital transformation'. Just a clear conversation about your systems."