How Anveshan Achieved a 13X WhatsApp ROI with Netcore's AI Propensity Model

Anveshan is a fast-growing D2C brand focused on pure, traditionally processed foods and health nutrition products – ghee, honey, cold-pressed oils, and more. Built on a commitment to quality and transparency, the brand has cultivated a loyal customer base that responds strongly to direct communication channels.

WhatsApp sits at the centre of Anveshan’s marketing mix. It is the brand’s highest-converting channel – but also its most expensive to operate. Every message carries a direct cost, making it critical that sends reach users who are genuinely ready to engage and purchase, not just those who are on the list.

Industry Context

Anveshan operates in India’s fast-growing D2C health-and-nutrition category, where consumers are increasingly moving away from mass-market packaged staples toward brands built on traceable sourcing and minimal processing – ghee, honey, and cold-pressed oils among them. This segment competes less on price and more on trust, which makes direct, high-affinity channels like WhatsApp disproportionately important: it is where these brands can reinforce provenance and quality story with an already-engaged audience. As the category matures, the brands pulling ahead are the ones that combine that trust-led positioning with more precise, data-driven engagement rather than simply increasing send volumes.

Anveshan had built strong WhatsApp marketing momentum – high volumes, consistent engagement, and a 7X ROI that outperformed most channels. But the ceiling was real and visible, and the existing approach offered no mechanism to break through it.

The ROI Ceiling

• Seven times return on investment is a strong result by any benchmark – but Anveshan’s team could see that performance had stabilised and was not trending upward
• With traditional audience segmentation, all users on the list received the same message regardless of their current intent or readiness to purchase
• There was no way to distinguish a user who was about to buy from one who had simply not opted out – both received the same spend

The WhatsApp Cost Problem

• WhatsApp is one of the highest-converting channels available – but it is also one of the most expensive to run, with a direct cost attached to every message sent
• Broad sends to the full audience meant a significant portion of budget was spent on users who would not open, click, or convert
• Unlike email, where the marginal cost of an extra send is near-zero, WhatsApp makes every low-intent send a real and measurable cost to the business
• The challenge was not volume – it was precision: how to concentrate spend only where it would generate a return

Netcore deployed its AI-powered Propensity Model – a machine-learning system that scores every user in the audience on their predicted behaviour before a single message is sent. Rather than broadcasting to the full list, the model identifies a high-intent cohort: users with the highest combined probability of engaging and converting.

The Two Predictive Signals

The Two-Phase Rollout

The model was deployed in two deliberate phases to validate performance before scaling:

Phase 1 – Proof of Concept (March 12)

Campaigns #1315 and #1316 launched as an initial test. An unexpected Meta platform issue limited the total reach, but the targeting precision was so high that the constrained send still delivered strong immediate ROI – validating the model immediately and prompting the client to request a rapid second phase.

Phase 2 – Rapid Scale (March 13)

Campaigns #1326 and #1327 targeted a fresh batch of untapped users (Phase 1 audience excluded to ensure new reach). With the Meta issue resolved, the model scaled successfully – multiplying revenue exponentially and confirming that the Phase 1 result was not an anomaly but a repeatable outcome.

Four campaigns. 0.33% of annual send volume. Benchmarked against 1,117 legacy campaigns sent over the previous 12 months.

The Propensity Model delivered a 13X return on investment – a milestone figure that had never been achieved in Anveshan’s prior marketing history. Broken down by targeting signal:

For context: Anveshan’s legacy manual segmentation campaigns delivered a 7X ROI. The propensity model delivered approximately 1.8X higher return – while using a fraction of the send volume.

Full Performance Benchmarking vs Legacy

When benchmarked against 1,117 legacy BAU campaigns sent over the previous 12 months, the Propensity Model demonstrated significant improvement across every metric in the funnel:

The most significant delta is at the bottom of the funnel. The Propensity Model does not just improve engagement – it fundamentally changes the quality of users who convert, and how much they spend when they do.

This means Anveshan can now achieve more revenue with fewer sends – reducing cost while increasing return. The ceiling that existed under the legacy approach is not just broken. It has been replaced with a fundamentally better model.

The WhatsApp Cost Question – Answered

The revenue-per-send metric directly addresses the cost challenge that made the status quo unsustainable:

• Legacy strategy: 4.42 revenue units generated per message sent
• Propensity Model: 15.79 revenue units generated per message sent
• Result: the model is 3.5X more profitable on a per-message basis – transforming WhatsApp from a high-cost channel into a demonstrably high-efficiency one

Campaigns that stood out

Campaign #1326 – Predictive Click, Scaled: The blockbuster performer of the rollout. Highest click rate of the cohort at 7.12% and the single highest revenue contribution across all four campaigns. Demonstrated that the model scales without losing precision.

Campaign #1315 – Predictive Open, Initial: Achieved the highest open rate of the cohort at 67.20% – effectively re-engaging Anveshan’s most loyal brand followers. Even under Phase 1 constraints, it validated the targeting signal immediately.

Campaign #1316 – Predictive Click, Initial: Secured an elite conversion rate of 0.65% – the strongest conversion result in the cohort. Demonstrated the quality of the click-intent signal even in the proof-of-concept phase.

Campaign #1327 – Predictive Open, Scaled: Overcame a delivery challenge (56.10% delivery rate) to still generate significant revenue – demonstrating that even partial reach from a precision cohort outperforms broad reach from a generic list.

The March 2026 rollout established a new baseline for what Anveshan’s WhatsApp channel can deliver. The propensity model is not a one-time intervention – it is a repeatable, scalable framework that improves as more data is added. The roadmap ahead:

The Anveshan story is not about sending less. It’s about sending smarter.