Fundle.ai: Dynamic Coupons in Loyalty Programs Transform Customer Experience

Customer Experience as a Competitive Advantage

Indian retail is entering a phase where location and product range are no longer sufficient moats. A shopper in Bangalore's Koramangala has access to Select CITYWALK-quality brands through quick-commerce, D2C websites, and a five-minute cab ride to three competing malls. The only durable advantage left is how a brand or mall makes the customer feel across every interaction — and the coupon or reward moment is one of the highest-stakes touchpoints in that journey.

Customer experience in loyalty programs is not soft or abstract. It is measurable through Net Promoter Score, repeat-visit frequency, share-of-wallet, and lifetime value. Tanishq's Encircle program is instructive: it uses purchase history to inform both the offer and the communication channel, resulting in one of the highest programme-active-member ratios in Indian jewellery retail. Apollo Pharmacy's loyalty tier structure similarly uses purchase cadence to personalise health-category coupons, which keeps pharmacist-recommended SKUs in the basket even when OTC generics are cheaper online. These are not coincidences — they are the downstream result of treating the coupon as a personalised signal rather than a bulk discount event.

The competitive set is beginning to catch up. Capillary Technologies and EasyRewardz have both added rule-based personalisation layers to their offer engines. MoEngage and WebEngage offer behavioural triggers that can push coupons based on app events. Xeno and Almonds.ai are carving out mid-market niches with campaign automation. But rule-based personalisation hits a ceiling quickly: rules written by a marketing manager in January do not account for the micro-segmentation nuance that an AI model discovers by March. A rule says 'send a 15% off ethnic wear coupon to members who bought kurtas in the last 90 days before Navratri.' An AI model says 'this specific member's price sensitivity has decreased 23% since her last purchase, her basket skews premium, and she responds to WhatsApp at 7 p.m. on weekdays — send a ₹500 flat discount on the silk range, not a percentage off, via WhatsApp at 7:05 p.m. on the Thursday before Navratri.' The granularity is not cosmetic; it translates directly into conversion rate and margin recovery.

For a mall CMO managing 150 to 300 brand tenants, the stakes are even higher. Every tenant's coupon program competes for the same shopper's attention in the same physical and digital space. A poorly timed, irrelevant coupon from one tenant creates noise that degrades the shopper's overall perception of the mall's loyalty programme. Centralising offer intelligence — so that the mall's AI layer arbitrates which tenant offer gets priority for which shopper at which moment — is not a luxury feature. It is table stakes for any mall that wants to grow its loyalty program NPS above 50.

Personalization and Timeliness in Dynamic Coupons

Personalisation and timeliness are not the same variable, but they are deeply interdependent. A perfectly personalised offer delivered 48 hours after the shopping intent has evaporated is just a delayed irrelevancy. Conversely, a real-time trigger on an offer that does not match the member's current category intent wastes the recency advantage. The best dynamic coupon programs in global retail — and the handful of Indian deployments that are beginning to outperform — get both dimensions right simultaneously.

Timeliness in the Indian mall context maps to three distinct trigger windows. The pre-visit window — roughly 24 to 72 hours before a projected mall visit based on historical footfall patterns — is ideal for high-consideration category coupons (jewellery, electronics, premium apparel). The in-visit window — triggered by geofence entry or beacon ping — is optimal for impulse categories: F&B, footwear, accessories. The post-visit window — within four to six hours of exit — is the highest-converting window for reactivation offers on categories the shopper browsed but did not purchase. Cafe Coffee Day, for instance, sees its highest coupon conversion rates when a beverage offer goes out within 90 minutes of a member leaving a food-court-adjacent anchor store without visiting the F&B zone.

Personalisation at the offer-mechanics level goes beyond selecting the right category. It includes: discount modality (percentage off versus flat rupee discount versus bonus points), minimum spend threshold calibration (a ₹999 minimum for a member whose average basket is ₹1,400 is a stretch goal; the same threshold for a member averaging ₹600 is a barrier, not an incentive), channel selection (transactional SMS for Tier 2 city members; WhatsApp for metro millennials; in-app push for high-frequency app users), and expiry window (a 48-hour expiry creates urgency for high-frequency buyers; a 7-day window is necessary for members whose visit cadence is monthly). Each of these variables should be set by the AI model individually per member cohort, not uniformly across the programme.

The upstream requirement is first-party data depth. Brands that have SKU-level transaction history, category browse data from their app, and channel-response attribution are in a position to personalise at this level of detail. Those that only have transaction totals and phone numbers are not — and no AI model can compensate for data poverty. This is why Fundle's onboarding process begins with a data audit that maps what signals are available from the brand's POS (POSist, Petpooja, GoFrugal, Wondersoft are common integrations in Indian mall ecosystems) before a single campaign is designed. Garbage-in-garbage-out applies with particular force to dynamic coupon engines because the model's output is only as intelligent as its training signal.

Fundle's AI and Machine Learning Methods

Fundle AI Platform operates on a three-layer intelligence architecture purpose-built for the complexity of Indian mall and multi-brand retail environments. The first layer is the data unification engine, which ingests transaction streams from POS systems (POSist, GoFrugal, Wondersoft, Petpooja), CRM records, app behaviour events, geofence signals, and — where available — third-party demographic enrichment. This layer resolves member identity across channels (a shopper might transact under three different phone numbers across three years), creating a unified profile that is the foundation for all downstream AI scoring.

The second layer is the predictive modelling suite within Fundle AI Agents. This comprises four primary models running in parallel: a propensity-to-purchase model that scores each member's likelihood to buy in a given category within the next seven days; a price-sensitivity model that estimates the minimum discount required to shift a purchase decision for each member cohort; a churn-risk model that flags members whose visit frequency or transaction value is declining relative to their historical baseline; and a channel-response model that predicts which communication channel will yield the highest open and conversion rate for a given member on a given day of week and time of day. Outputs from all four models are combined by a ranking algorithm — part of the Fundle Agentic AI layer — that selects the single best offer to serve each eligible member at each trigger moment.

The third layer is the Fundle AI Workflow engine, which handles execution: coupon generation, channel dispatch, redemption tracking, and closed-loop attribution back to the source model. Attribution in Indian retail is genuinely hard because a shopper might receive a WhatsApp coupon, visit the store three days later, and redeem at a POS terminal that routes through a different system. Fundle AI Workflow maintains the attribution chain across this fragmented path by tagging each coupon with a unique identifier that persists from dispatch to POS redemption, enabling true incrementality measurement rather than last-touch proxies.

Fundle's dynamic coupon platform positively impacts customer NPS and retention rates — this is not a theoretical claim. In practice, when members receive offers that feel individually crafted, their perception of the brand or mall's programme quality improves materially. NPS improvement of 8 to 14 points has been observed in loyalty programmes that transition from generic blast campaigns to Fundle-powered dynamic offer flows within the first 90 days of deployment. Retention — measured as 90-day active rate among members who received at least one dynamic coupon — consistently runs 9 to 11 percentage points above the control group that received standard campaign offers during the same period.

Examples of Enhanced CX from Indian Retailers

The proof points from Indian retail are instructive precisely because they are drawn from the messy, fragmented, multi-format reality of the market — not from a sanitised Western SaaS case study.

Consider a mid-sized fashion brand operating 60 stores across Tier 1 and Tier 2 cities, with an average transaction value of ₹2,200 and a loyalty member base of 8 lakh members. Pre-Fundle, the brand ran four coupon campaigns a month: a season-opener, a mid-month push, a weekend special, and an end-of-month clearance blast. All four went to all opted-in members. Redemption hovered at 21 percent. Average redeemed basket was ₹2,400 — barely above the non-coupon average, suggesting the discount was subsidising purchases that would have happened anyway. Post-deployment of Fundle Brand Loyalty's dynamic offer engine, the brand moved to event-triggered, AI-scored offers. Members flagged as churn-risk received a time-bound flat ₹300 off on their next visit. High-frequency members with growing basket sizes received a points-multiplier offer (no margin compression). Price-sensitive mid-tier members received a 12% off coupon with a ₹1,800 minimum. Redemption climbed to 34 percent within 60 days. More critically, redeemed basket value rose to ₹3,100 — a 29 percent uplift — because the minimum spend thresholds were calibrated above each cohort's historical average.

In the mall context, a Select CITYWALK-type operator running a centralised loyalty programme faces an additional coordination challenge: tenant offers must not cannibalise each other or create redemption fatigue. Fundle Mall Loyalty's offer arbitration layer solves this by setting per-member daily offer limits, prioritising tenant offers based on the member's predicted category intent on that visit, and suppressing duplicate-category offers within a 24-hour window. A shopper who already redeemed a fashion coupon that day will not receive another fashion offer from a competing tenant; instead, the system will surface an F&B or entertainment offer to extend dwell time — which drives footfall value for the mall without margin compression for the fashion tenant.

Manyavar's wedding-season campaign cycle is another reference point: the brand's high-consideration purchase rhythm (customers buy 1 to 2 times per major life event) makes recency-based triggers largely useless. What works instead is life-stage modelling — identifying members approaching a projected wedding or festive occasion based on browsing signals and peer-group purchase patterns — combined with a premium-experience offer (complimentary alteration, priority appointment) rather than a discount coupon. This is the kind of nuance that AI-driven dynamic offer engines handle naturally and that rule-based systems cannot replicate.

Measuring Customer Satisfaction and Loyalty Gains

Measurement frameworks for dynamic coupon programs must go beyond redemption rate, which is a top-of-funnel proxy that flatters programme activity without capturing economic value. A mature KPI stack for dynamic coupons in loyalty programs should operate across three time horizons.

In the immediate horizon (0 to 30 days post-campaign), track: coupon open rate by channel and cohort, redemption rate versus control group, incremental basket value (redeemed basket minus historical average basket for that cohort), and offer-to-visit conversion rate (for pre-visit triggered coupons). These metrics validate that the AI model's offer selection is improving on the baseline. A redemption rate that is high but basket uplift is zero suggests the discount threshold is set too low — the offer is being redeemed but not stretching behaviour.

In the medium horizon (30 to 90 days), track: repeat-visit frequency change among members who redeemed at least one dynamic coupon, share-of-wallet shift (are members spending more of their retail budget at this mall or brand versus alternatives?), and churn-risk cohort movement (what percentage of flagged churn-risk members were reactivated by dynamic offers?). These metrics capture whether the programme is building genuine behavioural loyalty or just manufacturing transactional spikes.

In the long horizon (90-plus days), the headline metric is customer lifetime value delta — the difference in projected LTV between members enrolled in the dynamic coupon programme and a matched control group. Supporting metrics include NPS movement (surveyed at 90-day intervals), tier upgrade rate (are mid-tier members moving to premium tiers, indicating deeper engagement?), and referral-driven acquisition rate (loyal members who received relevant offers are significantly more likely to refer peers).

For mall operators, an additional metric layer covers tenant economics: which tenant categories show the highest basket uplift from dynamic coupon activity, which categories show cannibalistic redemption patterns, and what is the mall's overall dwell-time change on days when the dynamic offer engine is active versus campaign-dark days. Dwell time is the mall's most monetisable KPI because it correlates directly with total tenant sales density, F&B capture rate, and entertainment spend — all of which flow into variable rent structures that underpin the mall's revenue model.

How Fundle solves this

Fundle was designed from the ground up for the specific constraints of Indian mall and multi-brand retail: fragmented POS ecosystems, mixed digital literacy across member bases, multi-tenant coordination complexity, and the need to deliver AI-grade personalisation without requiring each brand or mall to hire a data-science team. Every product in the Fundle ecosystem addresses a distinct layer of the dynamic coupon problem.

Fundle Loyalty is the core programme infrastructure: tier management, points engine, member onboarding, and campaign management. It integrates natively with the POS and CRM systems most commonly found in Indian mall ecosystems and provides the data foundation that all AI models require. Fundle Mall Loyalty extends this with the multi-tenant offer arbitration layer described earlier — ensuring that a mall's central loyalty programme can intelligently coordinate hundreds of tenant offers without creating a cacophony of discount notifications that degrades member experience.

Fundle Brand Loyalty serves individual retail brands — whether they operate inside malls or across standalone stores — with the same AI-driven offer personalisation at the brand level. Lenskart-style omnichannel brands, Reliance Trends-format value fashion operators, and Lifestyle-tier department stores each have different member economics and offer-mechanics requirements; Fundle Brand Loyalty's configuration layer accommodates all three without requiring bespoke development.

Fundle AI Agents power the predictive modelling suite: propensity scoring, price sensitivity modelling, churn prediction, and channel optimisation. These agents run continuously on live transaction data, recalibrating their outputs as member behaviour shifts across seasons, economic cycles, and competitive events. Fundle Agentic AI is the orchestration layer that combines agent outputs into a single ranked offer recommendation per member per trigger event — removing the human bottleneck that slows down rule-based platforms. And Fundle AI Workflow handles the execution and attribution chain that closes the loop from offer dispatch to POS redemption, giving mall CMOs and retail marketing heads the incrementality data they need to defend programme investment in quarterly reviews.

Vineet Narang's founding vision for Fundle was straightforward: every Indian shopper deserves a loyalty programme that understands them as an individual, not as a demographic bucket. Dynamic coupons are the most direct expression of that vision — the moment where the programme's intelligence becomes tangible in the shopper's hand. The Fundle AI Platform makes that moment repeatable at scale, across millions of members and hundreds of brands, without the margin bleed that comes from indiscriminate discounting.