“We will not build a loyalty platform for the AI era. We are building the loyalty platform of the AI era. That's the only standard worth shipping against.”
- •Understand why traditional loyalty point systems fail India's omnichannel retail shopper
- •Quantify the revenue gap caused by disconnected CRM and POS stacks in malls and brand retail
- •Map the five-layer architecture an AI loyalty agents platform needs to run seamlessly
- •Compare agentic AI loyalty against legacy rules-based CRM vendors operating in India
- •Evaluate Fundle.ai's live omnichannel metrics as a benchmark for your next deployment
India's organised retail sector crossed ₹8.1 lakh crore in FY24 and is on track to hit ₹12 lakh crore by FY27, yet the loyalty economics underneath that growth story remain structurally broken. The average Indian mall visitor transacts at 2.3 stores per visit but is enrolled in, at best, one loyalty programme — and that programme almost certainly cannot see what she bought in the other 2.3 stores. The result is a missed personalisation window that translates directly into missed repeat visits, missed cross-sell, and ultimately missed revenue. Retail CRM heads and mall marketing directors know this problem intimately: they sit on islands of first-party data — POS feeds from Wondersoft or POSist, e-commerce order data, app event streams — with no intelligent layer that unifies them into a single, actionable customer view.
The answer the industry has been reaching for is an AI loyalty agents platform: a system that does not merely collect points but actively orchestrates engagement across every channel a shopper uses, in real time, without a human having to write a campaign brief for every micro-segment. The distinction matters enormously. A rules-based loyalty engine — the kind sold by EasyRewardz or the older Capillary tier — can execute a birthday SMS. An AI loyalty agent can notice that a Tanishq buyer at Phoenix Marketcity Mumbai bought gold jewellery twice in twelve months, is about to lapse based on RFM signals, and has a high predicted affinity for home décor based on co-purchase patterns — and then autonomously trigger a personalised Diwali preview invite, a WhatsApp nudge, and a push notification, sequenced perfectly, without any campaign manager touching a keyboard. That is the operational gap this article addresses.
The urgency is real. India's top 10 mall operators — Phoenix Mills, DLF Malls, Prestige Estates, Nexus Malls, Virtuous Retail — collectively host over 4,000 brand stores and see north of 80 crore footfalls annually. Yet redemption rates on mall loyalty programmes hover between 8 and 14 percent, compared to 28–35 percent on best-in-class international programmes. The gap is not consumer apathy; it is platform inadequacy. Shoppers at Select CITYWALK in Saket, Delhi will happily engage if the offer is relevant and the friction to claim it is near zero. The problem is that the technology stack today — fragmented between POS vendors, WhatsApp BSPs, and point-redemption engines — cannot deliver relevance at scale.
Fundle was built specifically to close that gap. This article lays out the full picture: why the problem exists now, what a world-class AI loyalty agents platform looks like, how to evaluate vendors, and what metrics you should be demanding from any platform you deploy in 2024 and beyond.
India Omnichannel Loyalty: Benchmark Numbers Every CRM Head Should Know
Omnichannel Customer Journeys in Indian Retail
The Indian shopper does not think in channels. She discovers a Manyavar kurta on Instagram Reels, price-checks it on the brand's app, walks into the Phoenix Marketcity Pune store to touch the fabric, and completes the purchase on the brand website that night using a coupon from a WhatsApp group. That four-step journey touches four distinct data systems, and in the overwhelming majority of Indian retail organisations today, none of those four systems talk to each other in real time. The CRM head at a national fashion brand typically sees the in-store POS data in a nightly batch, the app event data in a separate analytics dashboard, and the e-commerce order data in yet another BI tool. The WhatsApp touchpoint is often not instrumented at all.
This fragmentation has measurable consequences. McKinsey's global retail research shows that brands with fully integrated omnichannel journeys achieve 10–15% higher revenue per customer annually. Translate that to an Indian context: a Lifestyle Stores customer with an average annual spend of ₹14,000 who shops both in-store and online is worth ₹1,400–2,100 more per year than her single-channel equivalent — but only if the brand can actually identify her across channels and serve her a unified experience. Most Indian brands cannot. The loyalty ID that Lifestyle or Pantaloons assigns at POS checkout is rarely the same identifier that fires when the same shopper adds to cart on the app. Identity resolution — matching a mobile number, a loyalty card, a device ID, and an email into a single customer record — is step zero of omnichannel loyalty, and it is still a largely unsolved problem for mid-tier Indian retail.
Mall operators face an even harder version of this problem. A Phoenix Marketcity mall cannot directly access the POS systems of the 200+ brands in its tenant mix. It knows footfall through gate scanners and car park RFID, it knows spend aggregates from co-branded credit card data partnerships, and it knows app engagement from its own mall app. But it does not know that the woman who spent ₹12,000 at Zara last Saturday also spent ₹8,000 at Sephora. Without that unified spend view, the mall's loyalty programme can only reward generic visits rather than high-value shopper behaviour — which is precisely why mall loyalty NPS scores in India average 23–28, well below the 45–55 range that global mall operators achieve.
The structural fix requires three things simultaneously: a unified identity layer that works across tenant POS systems without requiring every brand to rebuild their stack, a real-time event pipeline that ingests transactions as they happen rather than in overnight batches, and an intelligence layer that can act on those events without a human campaign manager in the loop. That is exactly the architecture an AI loyalty agents platform must deliver — and it is the architecture that the best Indian mall operators are now actively procuring.
The Indian Omnichannel Shopper Journey: Where Loyalty Breaks Down Today
How AI Loyalty Agents Enable Seamless Engagement
The phrase 'AI loyalty agents' gets used loosely in vendor decks, so it is worth being precise about what it means in practice. A genuine AI loyalty agent is an autonomous software entity that monitors a continuous stream of customer signals — transactions, app events, web behaviours, support interactions, geofence triggers — and takes goal-directed actions on behalf of the brand or mall operator without requiring a human to specify the action for each individual customer. The agent has access to tools: it can send a WhatsApp message, issue a reward, trigger a discount code, escalate to a human service rep, or update a customer segment. It has a goal — typically maximising a combination of repeat visit frequency and revenue per member — and it makes decisions about which tool to use, when, and for whom, based on a continuously updated model of each customer's behaviour and preferences.
Contrast this with what most Indian retail CRM teams actually operate today: a campaign calendar maintained in a spreadsheet, segments defined once a quarter in a BI tool, and campaigns fired via a WhatsApp Business API provider or an email ESP. This approach has a structural ceiling. A human campaign manager working with static segments can meaningfully differentiate, at most, 8–12 customer cohorts per campaign cycle. An AI loyalty agent operating on Fundle Agentic AI infrastructure can differentiate at the level of the individual customer — 1:1 personalisation across 1.33 crore members — and do so continuously, not quarterly.
The practical impact on engagement metrics is not marginal. When Apollo Pharmacy moved from batch-and-blast SMS campaigns to AI-triggered personalised health reminders tied to purchase history, repeat purchase rates in the loyalty cohort increased by 22% within six months. When FabIndia ran AI-personalised festive outreach that matched product recommendations to each member's historical category preferences, redemption rates on festive offers climbed from 11% to 29% — almost exactly the gap between Indian average and global best practice cited earlier. These outcomes are not achievable with rules-based engines because the decision logic required — which category, which price point, which channel, which timing — is too high-dimensional for humans to write as rules at scale.
Fundle AI Agents take this further by introducing what the industry calls 'agentic loops': the agent not only sends the communication but also monitors the response — opened, clicked, redeemed, ignored — and updates its model of that customer in real time. If a Cafe Coffee Day member consistently ignores morning push notifications but responds to WhatsApp messages sent between 7:30 and 8:00 PM, the agent learns that preference within three to four interaction cycles and permanently adjusts its channel and timing decisions for that member. No campaign manager ever has to write that rule. The system writes it for them, at the individual level, across every member in the programme.
AI Loyalty Agents Platform vs. Legacy Rules-Based CRM: Side-by-Side for Indian Retail
Cross-Channel Data Integration and Analytics
Data integration is where most Indian loyalty deployments die quietly. The vision — a single customer view across every touchpoint — founders on the reality that a mall like Select CITYWALK hosts tenants running at least six different POS platforms: Wondersoft, POSist, GoFrugal, Tally-based custom systems, international brand ERPs, and food court systems like Petpooja. Getting a unified transaction feed from this ecosystem is not a data science problem; it is a systems integration and commercial negotiation problem. Brands have little incentive to share their POS data with a mall operator, especially if the mall's loyalty programme competes with the brand's own CRM initiative.
A mature AI loyalty agents platform must solve this with a combination of technical connectors and commercial data-sharing frameworks. On the technical side, that means pre-certified API integrations with the major Indian POS and billing platforms — Wondersoft, POSist, GoFrugal, Petpooja — that mall operators can activate without requiring brands to rebuild anything. On the commercial side, it means a data governance model that gives each brand visibility into their own customer data within the unified platform while preventing raw transaction data from flowing to competitors. This is a non-trivial privacy and commercial design challenge, and it is one that generic CDP vendors like MoEngage or WebEngage — designed primarily for D2C e-commerce — have not solved for the mall context.
Once integrated, the analytics layer needs to go well beyond standard loyalty dashboards. The metrics that matter for omnichannel retail are not open rates and click-through rates; they are incremental visit frequency (how many additional visits per year does a loyalty member make compared to a matched non-member?), cross-tenant spend share (what percentage of a shopper's total wallet at a mall is captured by tenants within the programme?), and churn prediction accuracy (how precisely can the system identify members who are 60–90 days from lapsing, early enough to intervene profitably?). These metrics require a data model that connects transaction events across time and across tenants — which is exactly the model that Fundle Brand Loyalty and Fundle Mall Loyalty are built on.
Real-time analytics also changes how marketing teams operate. Instead of a monthly loyalty report that a CRM analyst builds in Excel, the mall marketing director needs a live dashboard showing today's active member count, today's redemption pipeline, and — critically — a ranked list of members who are in a high-churn-risk state right now and what the system is already doing about each one. That operational shift — from retrospective reporting to prospective action — is only possible when the analytics layer is natively connected to the execution layer, which is precisely the integration that Fundle AI Workflow delivers.
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: Deploying an AI Loyalty Agents Platform for Omnichannel Retail in India
Audit and Unify Your Identity Graph
Before any AI agent can act intelligently, it needs a clean identity layer. Map every customer identifier in your ecosystem — POS loyalty card numbers, mobile numbers, app device IDs, email addresses, UPI handles — and run a probabilistic matching exercise to collapse duplicates. A mid-size mall programme typically finds 20–30% identity duplication in its raw database. Eliminate that before go-live; false duplicate removal alone typically lifts your active member count accuracy by 15–18%.
Integrate POS and Digital Event Streams in Real Time
Connect your primary POS platforms — Wondersoft, POSist, GoFrugal, Petpooja for F&B — to your loyalty platform via certified API connectors, not overnight file drops. Configure your app event SDK and website pixel to fire loyalty-relevant events (product views, cart adds, checkout completions) into the same event bus. Target a data latency of under 60 seconds from transaction to loyalty platform ingestion; anything above five minutes makes real-time agent decisions impossible.
Define Agent Goals, Not Campaign Rules
This is the mindset shift that separates AI loyalty deployment from legacy CRM deployment. Instead of writing 'Send a birthday SMS with 10% off', define the agent's goal as: maximise 90-day repeat purchase probability for each member, subject to a maximum communication frequency of four touchpoints per week and a maximum discount budget of ₹150 per member per month. Let the agent decide which channel, which offer, and which timing. Start with a single agent goal per programme tier and expand as you validate performance.
Run Challenger Tests Against Your Baseline
Deploy the AI agent to 20% of your member base in a holdout experiment while maintaining your existing rules-based campaigns for the other 80%. Measure incremental visit frequency, spend per visit, and redemption rate over 60–90 days. Indian retail benchmarks suggest a well-configured AI loyalty agent produces 18–26% higher redemption rates and 12–19% higher repeat visit frequency versus rules-based baselines within the first 90 days of operation.
Scale, Monitor, and Expand Scope Progressively
Once challenger tests confirm lift, roll out to the full member base and expand the agent's tool set — add WhatsApp two-way conversations, in-store beacon triggers, co-branded credit card event feeds. Establish a weekly operating rhythm: review agent decision logs to understand why the system made specific choices, monitor for any bias in offer distribution across income or geography segments, and set quarterly goals for new agent capabilities. AI loyalty is not a set-and-forget deployment; it compounds value with each additional data source and each additional interaction cycle.
Tools for Personalised Offers and Rewards in AI-Driven Loyalty
Personalisation in loyalty is not just about recommending the right product; it is about constructing the right incentive structure for each individual customer. A Reliance Trends shopper who is highly price-sensitive responds to a ₹200 cashback on a ₹999 purchase. A Tanishq buyer with an ATV of ₹45,000 is not moved by cashback at all; she responds to exclusive early access to a new collection or an invitation to a private jewellery styling event. Getting this distinction right at scale — across hundreds of thousands of members — is precisely what an AI loyalty agents platform does that a human campaign manager cannot.
The offer construction problem has three dimensions: the incentive type (cashback, points multiplier, exclusive access, free gift, experiential reward), the incentive value (calibrated to each member's price elasticity, not applied uniformly), and the offer timing (at the point of maximum receptivity, not at the point of maximum convenience for the marketing team's campaign calendar). Rules-based systems can handle, at most, two or three tiers of incentive differentiation. Fundle AI Agents operate on a continuous incentive optimisation model that adjusts all three dimensions simultaneously for each member based on their observed response to previous offers.
In practice, this produces offer economics that are materially better for the brand. When incentive values are calibrated to individual price elasticity rather than applied uniformly, the same promotional budget produces higher response rates because high-value customers who would have purchased anyway receive smaller incentives, while price-sensitive lapsing customers receive larger ones exactly when it matters. Indian retail operators running AI-optimised offer engines typically report 15–22% improvement in promotional ROI within two programme cycles — meaning the platform pays for itself in reduced discount cost alone, before incremental revenue is counted.
For mall operators, the personalised rewards toolkit extends to the tenant mix. A shopper with a demonstrated affinity for F&B over fashion should receive offers anchored in the food court and restaurant tenants; a shopper who visits primarily on weekends and concentrates spend in electronics should receive weekend multiplier campaigns on electronics. This tenant-level offer routing is only possible when the AI system has a unified view of cross-tenant spend — which loops back to the data integration architecture described earlier. The full personalisation stack, from identity resolution through to offer delivery and post-redemption learning, is what Fundle AI Platform delivers as an integrated system rather than a collection of point solutions that need to be manually connected.
- Real-time POS integration certified with Wondersoft, POSist, GoFrugal, and Petpooja — not batch file imports
- Identity resolution engine that probabilistically matches mobile, email, device ID, and loyalty card into a single member record without manual deduplication
- AI agent architecture that sets goals and autonomously selects channel, offer type, and timing — not just an A/B testing wrapper over a rules engine
- Pre-built WhatsApp Business API integration with two-way conversation capability for redemption, balance queries, and offer claims
- Cross-tenant spend visibility for mall operators with brand-level data governance controls that satisfy DPDP Act 2023 compliance requirements
- Holdout group testing framework built into the platform — not requiring a separate analytics team to construct experiments manually
- Live operational dashboard with churn-risk ranked member lists and agent decision logs — not just monthly campaign performance reports
“In Indian retail, the data has always been there — buried in POS receipts and app logs. The breakthrough is AI agents that act on it in real time, at the individual level, without a campaign manager in the loop.”
How Fundle solves this
Fundle was built from first principles to solve the omnichannel loyalty problem as it actually exists in Indian retail — not as it exists in a SaaS vendor's global product roadmap. The Fundle AI Platform is the only loyalty infrastructure in India architected around agentic AI from the ground up: not a legacy points engine with an AI module bolted on, but a system where AI agents are the primary execution layer and human campaign managers set goals, review performance, and adjust guardrails rather than writing individual campaign logic.
Fundle Mall Loyalty solves the tenant data problem that has blocked mall operators for a decade. Pre-built connectors for Wondersoft, POSist, GoFrugal, and Petpooja mean that a Phoenix Marketcity or a Nexus Mall can achieve a unified cross-tenant transaction feed without requiring individual brands to change their POS stack. The identity resolution engine resolves member identities across tenant systems probabilistically, achieving match rates above 85% in live deployments — compared to 40–55% for manual deduplication approaches. Fundle manages loyalty engagement across malls, online, and mobile channels for over 1.33 crore members, which means the platform's models are trained on Indian shopper behaviour at a scale no domestic competitor currently matches.
Fundle Brand Loyalty addresses the parallel problem for enterprise retail brands: how do you run a single loyalty programme that works equally well in a Pantaloons store in Tier-2 India with intermittent internet, on a flagship brand app, and on a WhatsApp chat interface — without maintaining three separate systems? Fundle AI Workflow orchestrates communication and reward fulfilment across all three channels from a single campaign goal definition, with Fundle Agentic AI deciding in real time which channel each individual member should receive each interaction on, based on observed channel preferences and current engagement state.
Vineet Narang's founding vision for Fundle was that Indian retail deserved loyalty infrastructure that was genuinely AI-first, not AI-labelled — a system that treats personalisation as the default, not the premium tier. Fundle AI Agents embody that vision by operating continuous optimisation loops on every member's engagement trajectory: monitoring signals, predicting churn, constructing personalised incentives, executing multi-channel outreach, and closing the feedback loop — all without a human campaign manager making individual decisions. For the retail CRM head or mall marketing director evaluating next-gen platforms today, the question is not whether agentic AI loyalty is the right direction; the evidence from live deployments is unambiguous. The question is which platform can actually deliver it at the scale and with the Indian retail integrations your programme requires. That is the question Fundle.ai was built to answer.
Frequently asked
What is an AI loyalty agents platform and how is it different from a standard loyalty CRM?+
A standard loyalty CRM stores points, executes manually designed campaigns, and reports on outcomes. An AI loyalty agents platform uses autonomous AI agents that continuously monitor each customer's behaviour, select the optimal channel and offer for each individual, and execute engagement actions without human campaign design. The key difference is autonomy at scale: a rules-based CRM can differentiate 8–15 customer segments; an AI loyalty agent differentiates at the level of each individual member in real time.
How does Fundle handle POS data integration across multiple tenants in a shopping mall?+
Fundle Mall Loyalty ships with pre-certified API connectors for Wondersoft, POSist, GoFrugal, and Petpooja — the four POS platforms that collectively cover over 70% of organised Indian mall tenants. Brands do not need to rebuild their POS stack. The connectors push transaction events to the Fundle event bus in real time, and a brand-level data governance layer ensures each tenant can only see their own customer data within the unified programme, satisfying DPDP Act 2023 requirements.
What redemption rate lift can an Indian mall operator realistically expect from an AI loyalty platform?+
Live Indian retail deployments show AI-driven loyalty programmes achieving redemption rates of 24–32%, compared to the 8–14% industry average for rules-based programmes. The lift comes from three sources: better offer relevance (AI-personalised offers match individual preferences), better channel selection (messages arrive on the channel each member actually uses), and better timing (offers are triggered by real-time behavioural signals, not fixed calendar dates). Most programmes see meaningful lift within 60–90 days of AI agent activation.
How does Fundle's identity resolution work when a customer shops both in-store and online?+
Fundle's identity resolution engine uses probabilistic matching across six identifiers: mobile number, email address, POS loyalty card ID, app device ID, UPI handle (where available), and browser cookie. The system scores match confidence for each pair of records and merges records above a configurable threshold — typically 0.85 confidence or above. In live deployments, this achieves identity match rates above 85% across in-store and digital channels, compared to 40–55% for manual or deterministic-only matching approaches.
Is Fundle compliant with India's Digital Personal Data Protection Act 2023?+
Yes. Fundle's data architecture is designed with DPDP Act 2023 compliance built in. This includes explicit consent collection at enrolment with granular purpose specification, a self-service member data access and deletion portal, data localisation within Indian infrastructure, and audit logs for all AI agent decisions that involve personal data. Mall operators using Fundle Mall Loyalty also benefit from a tenant data governance framework that prevents cross-brand data leakage within the unified member database.
How long does a typical Fundle deployment take for an Indian mall or retail brand?+
For a retail brand with a single POS platform and an existing customer database, the Fundle AI Platform can be live in 3–4 weeks: one week for data ingestion and identity resolution, one week for agent goal configuration and offer catalogue setup, one week for channel integration (WhatsApp BSP, push notifications, email), and one week of parallel testing before full go-live. For a mall operator integrating multiple tenant POS systems, the timeline is typically 6–10 weeks depending on the number of tenant connectors required and the quality of the existing member database.
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.