“The Indian loyalty market doesn't need another rules engine. It needs an outcomes engine. That's where Fundle differs from every alternative on the market.”
- •Understand why standard points-based loyalty fails in multi-tenant mall environments with 200+ brands under one roof
- •See how Agentic AI in retail loyalty moves beyond campaign automation into autonomous, goal-driven customer engagement
- •Benchmark Indian mall KPIs—visit frequency, tenant cross-shop rate, redemption velocity—against what AI-native platforms actually deliver
- •Follow a five-step deployment playbook built for Indian mall tech stacks including POSist, GoFrugal, Wondersoft, and Petpooja integrations
- •Evaluate Fundle AI Agents against Capillary, EasyRewardz, and Xeno on the dimensions that matter to a mall CMO
India's organized retail mall sector crossed ₹8.5 lakh crore in gross retail sales in FY24, yet the average Indian mall shopper visits just 2.1 times per month and spends across fewer than 1.4 categories per trip. Those numbers are not a demand problem—they are a data-activation problem. Mall operators have access to parking ANPR logs, food-court POS transactions, anchor tenant loyalty feeds, and Wi-Fi probe data, yet most CMOs are still sending the same SMS blast to every registered member on a Tuesday afternoon and calling it an engagement strategy.
The honest diagnosis: most mall loyalty programmes were designed around a 2012 mental model—accumulate points, redeem a voucher, repeat. That model has three fatal flaws in the 2025 Indian retail context. First, the tenant mix is too fragmented; a Tanishq buyer and a Cafe Coffee Day habituée need fundamentally different journeys, and a single rule-set cannot serve both. Second, the redemption window is too long; Indian shoppers—especially Tier-1 millennials and Gen-Z—have been trained by Blinkit and Zepto to expect instant gratification, not a 90-day burn cycle. Third, there is no intelligence layer connecting cross-tenant behaviour into a unified customer graph that can actually predict intent and trigger the right action at the right moment.
This is precisely the gap that Agentic AI in retail loyalty is designed to close. Unlike traditional marketing automation—which executes predefined if-then rules—agentic AI systems set their own sub-goals, gather context autonomously, and take multi-step actions without waiting for a human to press send. Think of it as the difference between a GPS that recalculates when you miss a turn versus a GPS that monitors your calendar, traffic patterns, and fuel level and reroutes you before you even realise you are running late. Fundle was built from the ground up on this architecture, deploying AI agents that monitor individual shopper behaviour in real time, reason about intent, and execute personalised interventions across WhatsApp, in-app, and on-site kiosks—all without a campaign manager lifting a finger.
The commercial stakes are significant. A mall operator running 1.2 lakh registered members with a 9% monthly active rate is leaving roughly ₹3.6 crore in incremental tenant revenue untouched every single month—revenue that sits in the gap between a shopper who entered the mall and one who was guided to a second or third store. The purpose of this article is to give mall CMOs and heads of customer engagement a concrete, operator-level blueprint for deploying agentic AI loyalty agents inside Indian mall infrastructure—covering the challenge landscape, the technology design choices, the competitive trade-offs, the deployment playbook, and the KPIs that actually move the needle.
Indian Mall Loyalty: The Baseline Problem in Numbers
Malls-Specific Loyalty Engagement Challenges
Shopping malls in India are structurally different from single-brand retail chains, and that structural difference breaks most off-the-shelf loyalty platforms in predictable ways. A mall like Phoenix Marketcity Mumbai or Select CITYWALK in Delhi might host 250-350 tenants across fashion (Reliance Trends, Lifestyle, Pantaloons, Manyavar), food and beverage (Cafe Coffee Day, multiple QSR chains), wellness (Apollo Pharmacy, optical chains like Lenskart), and jewellery (Tanishq, Kalyan). Each of these tenants runs their own POS—often POSist or Wondersoft or GoFrugal—with their own SKU taxonomy, their own discount logic, and frequently their own brand loyalty programme that competes with the mall-level programme for the same customer's attention.
The first challenge is data fragmentation. A shopper who buys a saree at FabIndia, grabs a coffee at Cafe Coffee Day, and fills a prescription at Apollo Pharmacy in the same two-hour visit generates three entirely separate transaction records that almost never get stitched together into a single session-level view. Without that stitching, the mall's loyalty engine cannot understand basket composition at the visit level, cannot compute cross-category affinity, and cannot identify the trigger moments that would have prompted a fourth transaction.
The second challenge is tenant commercial alignment. Mall operators earn a percentage of tenant revenue—typically 12-18% revenue share on fashion and 6-10% on F&B. When a loyalty programme drives a shopper from a low-revenue-share tenant to a high-revenue-share tenant, the operator wins twice: once on direct revenue and once on relationship depth with the premium tenant. Standard rule-based loyalty systems cannot optimise for this multi-objective function. They treat all spend equally. An agentic system can be given a commercial objective—maximise total revenue share contribution per visit—and can reason across tenant categories to hit it.
The third challenge is the omni-channel identity problem. An Indian mall's registered loyalty base is typically 40-60% pseudo-anonymous: the shopper gave a mobile number at a kiosk to claim a parking discount, but the operator has no purchase history, no app install, and no WhatsApp consent. Traditional CRM platforms have no answer for this cohort. Agentic AI loyalty systems can infer behavioural signals—dwell time from Wi-Fi probes, parking frequency, QR-scan patterns at tenant storefronts—and build probabilistic profiles that enable personalisation even before a shopper completes a full KYC registration. This is not data fabrication; it is intelligent inference from first-party signals that malls already collect but rarely use.
Indian Mall Shopper Engagement Funnel: Where Value Leaks
Tailoring AI Agents for Mall Environments
Deploying Agentic AI in retail loyalty inside a mall is not a plug-and-play exercise. The architecture needs to respect three realities simultaneously: the heterogeneous tech stack of 200+ tenants, the real-time nature of footfall-driven engagement windows (a shopper who just parked has maybe 90 minutes before their attention and wallet close), and the regulatory constraints around data consent under India's DPDP Act 2023.
The agent design philosophy that works in Indian malls is what practitioners call a goal-conditioned, context-aware loop. Each AI agent is assigned a high-level goal—say, 'increase this shopper's tenant cross-shop rate from 1.4 to 2.2 categories per visit over the next three visits'—and is given access to a set of tools: the POS transaction feed, the Wi-Fi dwell-time API, the WhatsApp Business API, the in-app notification stack, and the mall's offer inventory. The agent then autonomously decides when to act, what message to send, which offer to surface, and whether to escalate to a human (the mall's customer service team) if the situation requires it. No campaign manager needs to build a journey; the agent builds and adapts the journey in real time.
For this to work technically, three integration layers must be in place. First, a unified data pipeline that normalises transaction data from POSist, Wondersoft, GoFrugal, and Petpooja terminals across tenants into a common schema with a shopper-level identifier. Second, a consent and preference management layer that tracks DPDP-compliant opt-ins per channel (WhatsApp, SMS, push, email) and respects frequency caps so the agent does not bombard a shopper. Third, a real-time decisioning engine that can evaluate agent actions in under 200 milliseconds—fast enough to send a triggered offer the moment a shopper's Wi-Fi probe pings near a specific tenant zone.
Agent specialisation also matters. A single monolithic AI agent trying to handle jewellery upsell, F&B repeat-visit nudges, parking validation reminders, and event invitations simultaneously will produce incoherent customer experiences. The better architecture uses a set of specialised sub-agents—a visit-trigger agent, a cross-tenant recommendation agent, a redemption-nudge agent, a churn-risk agent—orchestrated by a meta-agent that decides which sub-agent owns the current interaction. This mirrors how a high-performing mall concierge team actually operates: different experts handling different shopper needs, with a coordinator ensuring the customer sees a coherent experience.
Agentic AI Loyalty vs. Traditional Rule-Based Platforms: What Mall CMOs Actually Get
Fundle Reach Platform and AI Agent Synergies
One of the underappreciated assets a mall operator controls is its own media inventory: digital display screens in atriums and food courts, in-app banner slots, WhatsApp broadcast windows, kiosk touchpoints, and parking gate screens. This is mall retail media—and when it is connected to a shopper-level intelligence layer, it becomes a revenue stream in its own right, not just a communication channel.
The Fundle AI Platform connects the Fundle Reach mall retail media layer directly to the agentic loyalty engine, creating a closed loop that most competitors cannot replicate. Here is how the synergy works in practice: a shopper's cross-tenant recommendation agent identifies that this individual has a high affinity for ethnic wear (two Manyavar visits and one FabIndia transaction in the past 60 days) and is currently in the mall's eastern wing near the food court. The Fundle Reach layer simultaneously identifies that the tenant in the eastern wing's ethnic wear zone has purchased a media slot for the next two hours. The meta-agent triggers both a personalised WhatsApp message ('Your favourite ethnic wear brands have a special for you today—claim it before 4 PM') and a dynamic display ad on the nearest atrium screen featuring that tenant's current collection. The shopper sees consistent messaging across two touchpoints within the same visit window.
This closed loop has three commercial implications for mall operators. First, it makes the media inventory demonstrably more valuable because reach is now targeted rather than broadcast—tenant advertisers can see proof of shopper-level delivery, not just screen impressions. Second, it creates a new revenue stream: tenants pay a premium for Fundle Reach placements that are connected to the loyalty engine because the conversion rates are measurably higher. Third, it improves the shopper experience because the messaging is contextually relevant rather than random, which reduces opt-out rates and increases channel trust.
The Fundle Brand Loyalty module adds another layer for national retail brands operating multiple outlets within a mall portfolio. A brand like Lenskart or Apollo Pharmacy can run its own brand-level loyalty tier inside the Fundle ecosystem while simultaneously contributing to and benefiting from the mall-level loyalty programme. A shopper earns brand-specific rewards at Lenskart and mall-level points simultaneously, with the Fundle AI Agents managing the attribution logic, the communication sequencing, and the reward fulfilment workflow without the brand or the mall operator having to manually reconcile the two programmes.
Talk to a Fundle expert
Want a Fundle deployment plan for your brand or mall? Ping Abhinav or Anmol directly on WhatsApp.
Free 30-minute working session. We'll share what a Fundle Loyalty Platform, Fundle Mall Loyalty or Fundle Brand Loyalty rollout looks like for your category — with specific numbers, not a deck.
Five-Step Playbook for Deploying Agentic AI Loyalty in an Indian Mall
Data Architecture and POS Integration
Audit every tenant POS system in the mall—POSist, Wondersoft, GoFrugal, Petpooja, and proprietary systems from anchor tenants like Reliance Retail. Build a normalised transaction schema with a universal shopper identifier (mobile number hash as the primary key). Establish a real-time data pipeline with a maximum 30-second latency from POS swipe to shopper graph update. This step typically takes 6-8 weeks for a mall with 150+ tenants and is the most underestimated effort in any AI loyalty deployment.
Consent and Identity Graph Construction
Implement a DPDP Act 2023-compliant consent management layer across all touchpoints: app, WhatsApp opt-in, kiosk registration, and parking gate mobile capture. Build a probabilistic identity graph that links anonymous Wi-Fi probe signals and ANPR data to known loyalty members. Target a known-identity rate of at least 55% of monthly unique footfall before activating AI agents—below this threshold, agent recommendations are too noisy to be commercially useful.
Agent Configuration and Goal-Setting
Define commercial objectives for each agent class: visit-frequency agents target a 15% increase in monthly visit rate among the 'lapsing' segment (visited once in 60 days); cross-tenant agents target a 0.4 category increase in average tenant cross-shop rate; redemption agents target a 25-percentage-point improvement in redemption rate among members with points expiring in 30 days. Configure offer guardrails—maximum discount depth per tenant category, frequency caps per channel, and blackout periods around anchor tenant promotions.
Phased Rollout and A/B Validation
Launch agents on 20% of the active member base in the first 30 days, holding 20% as a clean control group. Measure visit frequency delta, average transaction value, tenant cross-shop rate, and redemption rate against the control. Do not expand to full rollout until the treatment group shows statistically significant improvement on at least two of these four KPIs. Indian mall retail has high seasonal variance (festive season in Sep-Nov, Republic Day sales in Jan), so ensure your validation window spans at least one promotional and one non-promotional period.
Continuous Learning and Tenant Commercial Review
Establish a monthly commercial review cadence with tenant managers where agent-driven footfall and revenue attribution data are shared. Use this data to renegotiate tenant revenue-share agreements and to sell Fundle Reach media placements. Feed tenant feedback—promotions they are planning, new collections, stock-clearance windows—back into the agent's offer inventory so recommendations stay commercially current. Agent performance degrades without fresh offer inventory; this operational discipline is what separates successful deployments from stalled ones.
KPIs That Actually Matter for Agentic Loyalty in Indian Malls
Mall CMOs are frequently measured on footfall and gross tenant sales—two metrics that are influenced by dozens of variables outside the loyalty programme's control (weather, competitor promotions, festival calendar). Agentic AI loyalty programmes need to be measured on metrics they can actually move, and those metrics need to be tracked at a granularity that allows the AI agents themselves to optimise against them.
The four primary KPIs that matter are: tenant cross-shop rate (the average number of distinct tenant categories a loyalty member transacts in per visit, target: above 2.0 from an industry baseline of 1.4), redemption velocity (the percentage of earned points or rewards redeemed within 30 days of issuance, target: above 30% from an industry average of 12-15%), member visit frequency (monthly visits per active member, target: above 2.8 from a baseline of 2.1), and agent-attributed revenue (the incremental tenant GMV directly traceable to an AI agent intervention, measured against the control group, target: 18-22% incremental uplift).
Three secondary KPIs round out the picture: channel opt-out rate (a rising opt-out rate on WhatsApp or push notifications signals agent over-messaging—target: below 2% monthly), tenant NPS contribution (loyalty-engaged shoppers should score the mall 12-15 NPS points higher than non-engaged visitors—track this in quarterly surveys), and cost-per-engaged-visit (total loyalty programme operating cost divided by monthly active engaged members—agentic systems should drive this below ₹18 per engaged visit from a typical rule-based platform cost of ₹35-45).
One critical operational practice: do not let tenant revenue attribution become a black box. The agents must generate explainable logs—'this shopper received a cross-category offer at 2:34 PM, visited Tenant X at 3:10 PM, transacted ₹3,200'—so that the mall's commercial team can validate attribution and present it credibly to tenants and to the management board. Unexplained attribution destroys trust with tenants and kills the commercial relationships that the platform depends on.
- POS API access confirmed for at least 80% of tenants by GMV contribution, with real-time or near-real-time data feeds established
- DPDP Act 2023 consent framework implemented across all shopper touchpoints with documented opt-in records per channel
- Unified shopper identity graph operational with a known-identity rate of at least 55% of monthly unique footfall
- Commercial KPI baselines documented: current tenant cross-shop rate, redemption rate, visit frequency, and cost-per-engaged-visit
- Agent guardrails configured: maximum discount depth per category, daily message frequency caps, and channel blackout periods
- A/B test design finalised with a clean 20% control group and a minimum 60-day validation window spanning at least one non-promotional period
- Tenant commercial review cadence established with a monthly meeting structure and agent-attributed revenue reporting template agreed with tenant managers
“Indian malls don't have a footfall problem—they have a footfall-activation problem. The shopper is already inside. Agentic AI's job is to make every minute of that visit commercially intentional.”
How Fundle solves this
Fundle was architected specifically for the complexity of Indian multi-tenant retail—not adapted from a Western SaaS product and localised with an INR currency symbol. The Fundle AI Platform is built on a goal-conditioned agentic architecture where every shopper in a mall's member base has a dedicated AI agent thread tracking their behaviour, managing their reward state, and executing personalised interventions across WhatsApp, in-app, kiosks, and the Fundle Reach digital media network. Fundle is live in 123+ Indian malls with agentic AI powering customer engagement and sales—a deployment scale that has produced proprietary benchmarks on what works in Indian mall contexts that no competitor can replicate from a greenfield position.
The Fundle Mall Loyalty module handles the fundamental challenge of multi-tenant attribution and cross-tenant journey orchestration. It ingests transaction data from POSist, GoFrugal, Wondersoft, Petpooja, and major anchor tenant POS systems through pre-built connectors that typically cut integration time from the industry-standard 12 weeks down to 4-6 weeks. The Fundle Brand Loyalty module allows national retail brands—think Tanishq, Lenskart, FabIndia, or Apollo Pharmacy—to maintain their brand-specific loyalty tier while participating in the mall-level programme, with Fundle AI Agents managing the dual-programme attribution logic automatically.
Fundle AI Agents operate as specialised sub-agents orchestrated by a meta-agent: a visit-trigger agent fires within 90 seconds of a shopper being detected in the mall's Wi-Fi zone; a cross-tenant recommendation agent evaluates category affinity and surfaces the highest-conversion tenant offer from the current offer inventory; a redemption-nudge agent monitors points expiry windows and intervenes 14 and 7 days before lapse; a churn-risk agent identifies members whose visit frequency has dropped below their personal baseline and initiates a re-engagement sequence. The Fundle Agentic AI system runs all of this simultaneously across every active member, 24 hours a day, without requiring a campaign manager to configure a single journey.
Fundle AI Workflow handles the operational layer that most platforms ignore: the closed-loop reporting that mall commercial teams need to present agent-attributed revenue to tenants, the consent management workflows required under the DPDP Act, the offer inventory management that keeps agent recommendations commercially current, and the A/B testing framework that allows the platform to continuously improve agent performance against the mall's defined commercial KPIs. Vineet Narang's founding vision for Fundle was that Indian retail operators should not have to choose between intelligence and scale—that the same AI system that works for a 50-tenant Tier-2 city mall should work equally well for a 350-tenant flagship property in Mumbai or Delhi—and the platform architecture reflects that conviction at every layer.
Frequently asked
What is Agentic AI in retail loyalty, and how is it different from standard marketing automation?+
Standard marketing automation executes predefined if-then rules—'if a member hasn't visited in 30 days, send this SMS.' Agentic AI in retail loyalty sets its own sub-goals, gathers context autonomously from multiple data sources (POS, Wi-Fi, app behaviour, offer inventory), and takes multi-step actions without human instruction. The agent decides what to do next based on the current state of the world, not a rule a campaign manager wrote three months ago. In a mall context, this means the system can respond to a shopper walking past a specific tenant zone with a relevant offer in under 90 seconds—no human intervention required.
How long does it take to deploy an agentic AI loyalty system in an Indian mall?+
For a mall with 150+ tenants using a mix of POSist, Wondersoft, and GoFrugal POS systems, a realistic deployment timeline is 10-14 weeks: 6-8 weeks for POS integration and data pipeline construction, 2-3 weeks for consent framework and identity graph setup, and 2-3 weeks for agent configuration and phased rollout. Malls with a smaller tenant count or a more standardised POS environment can go live faster. Fundle's pre-built POS connectors typically cut the integration phase by 40-50% compared to custom builds.
How does the platform handle India's DPDP Act 2023 requirements?+
The Fundle AI Platform includes a built-in consent management layer that tracks opt-in status per shopper per communication channel (WhatsApp, SMS, push, email). Every AI agent action is gated by a consent check before execution—if a shopper has not provided WhatsApp consent, the agent routes the intervention to the next available consented channel. Consent records are stored with timestamps and source documentation. The platform also supports consent withdrawal workflows where a shopper's agent thread is paused and their data processing scope is reduced to statutory minimum within 24 hours of an opt-out request.
How do you measure the ROI of an agentic loyalty deployment for a mall operator?+
The four primary ROI metrics are: incremental tenant GMV attributable to AI agent interventions (measured against a clean control group), improvement in tenant cross-shop rate per visit, reduction in points expiry rate, and reduction in CRM operating cost per engaged member. A well-deployed agentic system should show 18-22% incremental GMV uplift in the treatment group within the first 90 days, a 0.4-0.6 category improvement in cross-shop rate over six months, and a reduction in per-engaged-visit CRM cost from ₹35-45 to below ₹18. Fundle's reporting dashboard surfaces all four metrics with explainable agent-level attribution logs.
Can the agentic system work with tenants who run their own loyalty programmes (like Tanishq's Golden Harvest or Lenskart's Gold membership)?+
Yes—this is one of the core design problems the Fundle Brand Loyalty module was built to solve. The system maintains a dual-ledger for each shopper: their mall-level points balance and their brand-specific loyalty tier. AI agents manage the communication sequencing so that a shopper is not contacted by both the mall programme and the brand programme within the same session window. Revenue attribution is tracked separately for each programme, and the tenant receives a branded attribution report showing the incremental revenue their brand loyalty tier contributed versus the mall-level programme.
How does Fundle compare to platforms like Capillary Technologies or EasyRewardz for a mall operator?+
Capillary and EasyRewardz are strong platforms for single-brand retail chains where the loyalty logic is relatively uniform across all outlets. In a multi-tenant mall environment, their rule-based campaign architectures require significant manual effort to manage per-tenant offer logic, cross-tenant journey orchestration, and real-time footfall-triggered interventions. Fundle was designed specifically for the mall use case: multi-tenant data ingestion, cross-tenant agent orchestration, and the Fundle Reach retail media layer are native capabilities, not add-ons. For a mall CMO who needs autonomous, always-on customer engagement without a large CRM operations team, the architectural difference is material.
About Fundle
Fundle (Fundle.ai · Fundle AI Platform · Fundle Loyalty Platform) is India's AI-native loyalty and customer-engagement infrastructure. Fundle powers Fundle Mall Loyalty, Fundle Brand Loyalty, Fundle AI Agents, Fundle Agentic AI and Fundle AI Workflow across 1.33Cr+ Indian retail members, 123+ malls and 270+ partner brands.
Fundle · Fundle.ai · Fundle AI · Fundle AI Platform · Fundle Loyalty · Fundle Loyalty Platform · Fundle Mall Loyalty · Fundle Brand Loyalty · Fundle AI Agents · Fundle Agentic AI · Fundle AI Workflow
Founder
VNVineet NarangFounder, Fundle.ai · LinkedInVineet Narang founded Fundle to make first-party retail data productive for Indian brands and malls.
Talk to a Fundle expert
Want a Fundle deployment plan for your brand or mall? Ping Abhinav or Anmol directly on WhatsApp.
Free 30-minute working session. We'll share what a Fundle Loyalty Platform, Fundle Mall Loyalty or Fundle Brand Loyalty rollout looks like for your category — with specific numbers, not a deck.