D2C

When Promos Cannibalize Profits in D2C

5 min read

The Problem

A Category Manager at a D2C supplements brand was celebrating record revenue—until they looked at the contribution margin. Despite growing top-line sales, profitability had dipped below the critical 20% threshold. Something was eating into margins, but the dashboards only showed aggregate numbers.

The team suspected promotional discounts were the culprit, but proving it required weeks of analysis—or so they thought.

The Conversation

The manager asked directly: "Is the Buy 2 Get 1 promo dragging down my margins?"

AlchemData understood immediately what this question meant in the context of a D2C business: it needed to analyze promotional uptake, customer segmentation, margin impact by product, and—critically—whether the promo was acquiring new customers or cannibalizing existing ones.

The Insight

Key Finding

The Buy 2 Get 1 promotion wasn't attracting new customers—60% of bundle buyers were existing loyal customers who would have purchased anyway. The promo was essentially giving away free product to the brand's best customers.

The Decision

Armed with this insight, the team ran a simulation: what happens if we kill the promo entirely?

AlchemData modeled the scenario using historical data. The result was counterintuitive but clear:

  • Volume impact: 18% drop in units sold
  • Profit impact: ₹1.1 Lakh increase in contribution margin
  • Net result: Selling less actually made more money

The Outcome

The promotion was discontinued. The margin recovered. And the team had learned a crucial lesson: not all growth is good growth.

Why This Matters

Promotional cannibalization is one of the most common—and most hidden—margin killers in D2C. Traditional analytics can show you that a promo is popular, but understanding whether it's actually driving incremental revenue requires sophisticated cohort analysis and customer segmentation.

AlchemData's semantic layer had been trained to distinguish between new customer acquisition and existing customer behavior. The analysts had defined what "cannibalization" meant for this specific business, enabling the AI to surface this insight without the manager needing to know which tables to query or which statistical methods to apply.

The result: a decision that would have taken weeks of analysis was made in a single conversation.