Fundle.ai: AI Loyalty Agents Platform: Scaling Loyalty Across Indian Malls

Challenges of Scaling Loyalty Programs Across Locations

The single-mall loyalty program is already hard. The multi-mall loyalty program is an order-of-magnitude harder, and most operators underestimate this until they are already committed to an architecture that cannot scale. There are four structural failure modes that surface consistently when Indian mall groups try to centralise loyalty across properties.

First, tenant data fragmentation. A mid-size mall of 200 stores might have 40 different POS systems across its tenant mix — Manyavar on one stack, Tanishq on its own enterprise CRM, Cafe Coffee Day on Petpooja, a standalone kiosk brand on nothing at all. Each system defines a transaction differently. Each has its own customer ID schema. Building a unified shopper profile across these systems without an intelligent data layer is a six-month data-engineering project that needs to be repeated every time a new anchor tenant signs. Most CRM teams simply stop trying and settle for partial data.

Second, campaign localisation at scale. A shopper at Phoenix Marketcity Pune has a meaningfully different basket profile from one at Phoenix Marketcity Bangalore. The festival calendar differs. The anchor tenants differ. Average transaction values differ. A rule-based campaign engine that treats both properties identically — the default mode of platforms like EasyRewardz or a vanilla Capillary deployment — consistently underperforms against a system that adapts its offers, timing, and communication language to local context. The operational overhead of manually configuring separate campaigns for 20 or 50 properties is prohibitive for a team of five CRM managers.

Third, real-time decisioning gaps. The moment a shopper enters a mall, scans a QR code, or completes a transaction, there is a window of roughly 18-25 minutes during which a contextually relevant offer can change their next in-mall behaviour. Static campaign engines that run batch jobs at midnight cannot capitalise on this window. Most Indian mall operators are running loyalty interventions that reach the shopper six to twelve hours after the decision moment has passed — which is why redemption rates are structurally low.

Fourth, attribution and reporting chaos. When loyalty data sits in five different systems and campaign execution happens through a sixth, it becomes nearly impossible to answer the most basic business question: which tenant categories benefit most from loyalty-driven visits, and what is the incremental revenue generated per ₹1 of loyalty cost? Without clean attribution, mall operators cannot negotiate better joint marketing contributions from tenants or justify investment in the loyalty program to their CFO.

Features Needed for Multi-Mall Loyalty Automation

Not all loyalty platforms are equal when evaluated against the specific requirements of a multi-property Indian mall operator. The feature checklist that matters here is different from what a single-brand retailer like Lenskart or FabIndia would require. The following five capabilities separate a platform that can genuinely scale from one that will require a parallel IT project every time you open a new property.

Unified cross-property customer identity is the foundational requirement. A shopper who visits Select CITYWALK in Delhi and then travels to Phoenix Palladium in Mumbai must be recognised as the same individual, with their transaction history, tier status, and preferences intact. This requires a mobile-number-anchored identity graph with deduplication logic sophisticated enough to handle the realities of Indian data — multiple SIM cards, inconsistent name spellings across languages, and high wallet-app adoption alongside traditional loyalty card usage.

Agentic campaign decisioning — not rules, but goals. A rules engine says: if a shopper has not visited in 30 days, send a win-back SMS with 200 bonus points. An AI loyalty agent says: this shopper's historical pattern suggests she visits on weekends, her preferred category is ethnic wear (Pantaloons, Lifestyle, and boutique stores), she is currently in the silver tier and is ₹2,400 spend away from gold, and the next three weekends include two festive occasions — therefore the optimal intervention is a WhatsApp message on Thursday evening with a category-specific offer timed to the upcoming occasion, not a generic blast. The decisioning logic is continuous, contextual, and self-improving. This is the core of what makes an AI loyalty agents platform fundamentally different from legacy CRM automation.

Native multi-tenant data ingestion with pre-built POS connectors matters enormously in the Indian context. A platform that requires bespoke API development for every new tenant POS is a platform that will always have coverage gaps. Pre-built connectors to Petpooja, POSist, GoFrugal, Wondersoft, and the major apparel ERP stacks reduce time-to-data from months to days. It also removes the dependency on the IT team of individual tenants, which in a mall setting is a significant operational constraint.

Multilingual and multi-channel communication orchestration is non-negotiable. A Tier-2 mall in Coimbatore operates in Tamil. A property in Lucknow needs Hindi-first communication. A cosmopolitan audience in Bandra expects English. The platform must handle language personalisation at the member level, not the property level — because individual shoppers in any given mall will have different language preferences. Channel mix must be similarly adaptive: WhatsApp first for high-engagement members, push notifications for app users, SMS as a fallback, and in-mall digital screens for in-the-moment triggers.

Portfolio-level analytics with property-level drill-down closes the reporting gap. The Group CMO needs a single view of NPS, active member count, average revenue per loyal member, and tier migration velocity across all properties. The Mall Marketing Manager at each property needs the same metrics for their specific location, plus tenant-level attribution. Both views must come from a single data model, not from parallel dashboards that require manual reconciliation.

Case Study: 123+ Malls Using Fundle's AI Loyalty Solutions

Fundle supports 123+ malls with scalable AI-driven loyalty automation for diverse consumer segments. This is not a vanity metric — it represents the largest single deployment of an AI loyalty agents platform across Indian mall real estate and carries significant implications for what the platform has been stress-tested against.

Across this network, the median mall property has seen enrollment as a percentage of monthly footfall rise from 14% to 31% within the first six months of deployment. The mechanism is not a flashier sign-up form — it is the Fundle AI Agents triggering enrollment prompts at the precise moment a shopper completes a transaction at an anchor tenant, using a conversational WhatsApp flow that takes under 90 seconds to complete and immediately delivers a tangible first-visit reward. The timing and personalisation of the enrollment ask, determined autonomously by the AI agent based on transaction size, tenant category, and time-of-day, produces dramatically higher conversion than a static QR code at the mall entrance.

In the fashion and lifestyle category — where anchor tenants like Reliance Trends, Lifestyle, and Pantaloons drive the majority of loyalty transactions — Fundle's Agentic AI has demonstrated a 26% improvement in cross-tenant visit frequency. Specifically, members who had historically transacted only in fashion retail were being triggered with contextually relevant offers for F&B and entertainment categories at the moment their fashion transaction crossed a threshold amount. This cross-category activation is the fundamental economics of mall loyalty: a member who shops across four categories is worth approximately 3.8x the margin contribution of a single-category shopper, and the AI agent is the mechanism that engineers this behaviour at scale.

For mall operators managing properties in tier-2 markets — Nashik, Mysuru, Raipur, Surat — the platform's multilingual capability and WhatsApp-first communication strategy have proven particularly impactful. In these markets, app download rates for mall loyalty applications are low (typically 8-12% of enrolled members). The Fundle AI Workflow operates natively over WhatsApp, which has near-universal penetration in these cities and requires no app installation. This single architectural decision accounts for a significant portion of the platform's outperformance in non-metro deployments, where legacy platforms that depend on a branded app as the primary engagement channel consistently underperform.

Managing Data Privacy and Consistency at Scale

Scaling a loyalty program across 20, 50, or 120 mall properties in India introduces data governance obligations that many operators have not yet fully mapped against the Digital Personal Data Protection Act, 2023 (DPDP Act). The DPDP Act creates explicit consent obligations for collection and processing of personal data, introduces the concept of a Data Fiduciary responsible for managing user data, and carries penalties of up to ₹250 crore per violation. For a mall group with millions of loyalty members across multiple legal entities, this is a material compliance risk — not a theoretical one.

The consent architecture is the first and most operationally sensitive requirement. Every loyalty program enrollment must capture explicit, informed consent for each specified purpose of data use: points tracking, personalised marketing, cross-tenant data sharing, and third-party analytics. Consent must be granular — a member can agree to points tracking but opt out of personalised marketing — and the platform must be capable of honoring those granular preferences in real time. A blanket checkbox at enrollment is no longer sufficient. Platforms built pre-DPDP, including several legacy loyalty vendors in the Indian market, require significant rearchitecting to meet this standard.

Data consistency at scale is a different but equally important challenge. When a loyalty member updates her phone number, that update must propagate across every property's data layer within a defined SLA — typically under five minutes — to avoid the experience degradation of being unrecognised at a different property. When a member exercises her right to data erasure under DPDP, the erasure must be complete and auditable across all systems in the portfolio. These are not problems that a distributed, property-by-property CRM architecture can solve without significant custom engineering.

Fundle's AI Platform addresses this through a centralised data residency model with in-India hosting, a consent management module that is DPDP-aligned by design, and a real-time member profile synchronisation layer that maintains consistency across the full property network. The platform's audit trail for consent events, data access, and erasure requests is exportable in a format compatible with regulatory inspection requirements. For mall operators facing their first DPDP compliance cycle, this removes one of the most significant implementation risks of deploying a multi-property loyalty program.

How Fundle solves this

The Fundle AI Platform was architected around one organising principle: that loyalty at scale in India requires a system that makes autonomous, intelligent decisions at the member level — not a system that makes it easier for a CRM team to configure more campaigns. This distinction is the distance between incremental improvement and structural transformation of loyalty economics.

Fundle Mall Loyalty handles the full complexity of a multi-property deployment out of the box — cross-property identity, unified points currency, tenant-level attribution, and a portfolio analytics layer that gives Group CMOs and individual Mall Marketing Directors exactly the data granularity they need without separate BI tooling. The platform's pre-built POS connectors mean that a new property can be onboarded in days rather than months, which matters enormously when a mall group is opening three new properties in a financial year and the CRM team has not grown proportionally.

Fundle Brand Loyalty extends the same intelligence to anchor tenant programs — brands like Apollo Pharmacy, Tanishq, or Manyavar who want to run a loyalty program that is both their own branded experience and integrated into the mall's coalition currency. The integration is seamless from the member's perspective: a single wallet, a single WhatsApp thread, but with the brand's visual identity and personalisation logic applied when the interaction is brand-initiated. This solves one of the most persistent tensions in mall loyalty: the anchor tenant who wants control over their own loyalty proposition but does not want to fragment the shopper's experience.

Fundle AI Agents and Fundle Agentic AI represent the decisioning core of the platform. Each agent operates on a defined business goal — retention, tier upgrade, cross-category activation, win-back — and uses a continuous learning loop to improve its intervention strategy. The agents operate across WhatsApp, push notifications, in-mall screens, and SMS, with channel selection determined dynamically based on each member's historical response patterns. The Fundle AI Workflow orchestrates these agents across the full member lifecycle, from enrollment through to lapsed member re-engagement, without requiring manual campaign management. Vineet Narang's vision for Fundle has always been that the platform should do the work that 50 CRM executives would struggle to do manually — not just automate the tasks they already do, but make decisions they never had the data or bandwidth to make at all. For a mall group running 20 properties with a five-person CRM team, that vision is not aspirational — it is the operational reality the platform delivers today.