Kellanov cuts 30 minutes per store visit with AI shelf scanner, gains 1.8% sales lift
Kellanov deployed image recognition across modern retail stores to automate shelf checks that once required manual counting. The system scans product placement, inventory levels, and competitor positioning in real time, cutting store visit time by 30 minutes while improving on-shelf availability tracking.
A 0.9% increase in on-shelf availability translated to a 1.8% sales uplift during the pilot phase.
The company operates across three channels in India-general trade, modern retail, and digital commerce-each with different AI applications. The shelf-scanning project succeeded because it didn't disrupt existing workflows, according to Suchindra Khaidem, Head of Data & Analytics for Kellanov South Asia.
Real-time nudges are reshaping field sales
In general trade, where sales reps historically relied on judgment, Kellanov built a machine learning engine that suggests which products to push. The system analyzes transaction history to recommend must-sell, cross-sell, and focus SKUs in real time.
The pilot showed a 2 percentage-point sales increase in top outlets, plus 2% assortment expansion in urban areas and 5% in smaller towns. The company scaled the pilot nationally after seeing those results.
Khaidem said the engine "delivers consistent execution at scale and enables quicker onboarding by reducing dependence on individual judgement."
Trade spend is moving from guesswork to prediction
Trade promotions consume some of the largest budgets in FMCG. Kellanov built predictive models that guide promotion spending at the brand and SKU level instead of relying on after-the-fact analysis.
The shift freed up 20% of the team's effort and allowed the company to redeploy 15% of spending across its 20 top SKUs. Khaidem said the models "prioritise promotions that deliver sustainable value rather than short-term volume spikes."
During a product launch, the company used machine learning and transliteration to create 6,000 personalized retailer video ads that reached over 1 million consumers.
Factory floors are getting connected, slowly
Kellanov started deploying IoT sensors across two manufacturing sites-Taloja and Sri City-to capture real-time machine data. The work is foundational: connecting equipment, surfacing data through dashboards, and building visibility for factory and supply chain teams.
Khaidem said the factory transformation is "inherently more gradual" than front-end projects, but the company is "steadily scaling deployments and embedding it into decision-making cycles."
Trust and focus matter more than technology
Scaling AI pilots is harder than building them. Khaidem identified three requirements: cultural alignment, strong data foundations, and scalable architecture.
Trust is the starting point. "We build trust in AI outputs through close collaboration with business teams early on, which helps drive adoption and manage change," he said.
The company also prioritizes a narrow set of high-impact use cases. Some of these solutions are now expanding beyond India to global markets.
Khaidem's final point cut to the core: "Scaling AI is as much about mindset and continuous learning as it is about technology."
Kellanov's approach reflects a broader shift in FMCG. The industry still runs on scale, but competitive advantage increasingly comes from precision-knowing what to sell, where, and when. AI for Sales and AI Data Analysis are becoming operational capabilities, not experiments.
Your membership also unlocks:
