Everyone Measures CAC, But Who’s Counting CFC?

Devangshu Dutta

June 30, 2025

In every strategy meeting today, one metric is invariably mentioned: Customer Acquisition Cost (CAC). Whether you’re a well-funded corporate retailer, or raising your first angel round, or a well-established digital duopolist brand scaling Series C, CAC is one of the key performance metrics. “Real” spend that is neatly broken down by channel, optimised by funnel tweaks, scrutinised to the last rupee or dollar.

But there’s a metric we almost never hear about that could be costing brands far more in the long run.

Let’s call it Customer Forfeiture Cost (CFC), the residual lifetime value that is lost when a customer walks away from your business not because of price, competition, or even shifting needs, but because of a “burn”: a delivery missed or messed up, a refund that took weeks, an arrogant customer service call, or a product that failed spectacularly against the promise. In other words, when your brand hurts someone enough to make them walk away. Probably for ever.

It’s a paradox: brands are pumping thousands of crores into acquiring users, but they’re bleeding value at the other end. Yet, while CAC is a line item in every financial statement, CFC is invisible in management dashboards. CEOs don’t announce, “We’ve cut our forfeiture cost by 20% this quarter.”

Yet. every CXO knows it exists. The NPS scores, the social media complaints, the “never again” comments in reviews, the sinking feeling when repeat purchase rates fall.

Why CFC Matters More Than Ever

In every business, during the early stages each sale is a victory. Whether it was the retail chains that grew in the 1990s and early-2000s or the digital upstarts that came up through 2010s and 2020s, scale has been the mantra, and investors have poured money into scaling through the growing consumption of India 1 and India 2 customers.

Today customer acquisition isn’t cheap. The same person who clicked impulsively in 2020 now thinks twice before confirming payment. In this landscape, retention isn’t optional, it’s existential.

Every lost customer isn’t just a refund processed, or a cart abandoned. It’s the long tail of future repeat purchases that will never happen, negative word of mouth and brand distrust in the customer’s circle of influence, and increased future CAC due to declining organic reach.

Way back in 1967, management consultant Peter Drucker wrote in his book “The Effective Executive”: “What gets measured, gets managed”.

Today your CAC may be Rs. 500-1,000. If the average customer life time value (LTV) is Rs. 10,000, and a single burn causes churn after just one order worth Rs. 2,000, your CFC is Rs. 8,000, and that doesn’t even include reputational spillover.

Why We Don’t Measure It

Yes, CFC is hard to quantify. It’s not as easily attributable as ad spends. There’s usually no neat model telling you why someone never returned, because tech stacks aren’t typically designed to track emotional exits. And let’s face it, introspection about broken relationships is uncomfortable, even for management teams.

But that doesn’t mean it’s not real. If a customer leaves because your delivery executive messed up, or because your app crashed during checkout twice in a row, that’s on you, not the market. And in a business climate where sustainable growth is the mantra, LTV is king.

Ignoring CFC is like watching your roof leak and blaming the rain.

Toward a New Discipline

Brands and retailers must start measuring CFC, the value lost when customers disengage due to friction, mistrust, or neglect, and then start working on reducing it. This can be done by:

  • Tracking negative exits: Build feedback loops for poor customer satisfaction scores, refund requests, support escalations, and analyse their downstream effect on churn.
  • Building burn indicators: Assign internal scores to incidents where customers express betrayal or frustration, and combine qualitative feedback (customer calls, social posts) with purchase history to gauge how and when you lost someone.
  • Incentivising retention, not just acquisition: Perhaps most important, align teams across functions, not just marketing, to reduce friction and foster delight. Your logistics, tech, and customer service teams are as responsible for growth as your ad agency.

The Competitive Edge We’re Not Using

In a crowded space where everyone’s vying for eyeballs, trust is the true moat. Customers don’t expect perfection – they do expect accountability, authenticity, and recovery when things go wrong.

Brands that understand and act on Customer Forfeiture Costs will quietly start building a powerful edge: deeper brand loyalty, lower CAC over time thanks to referrals and repeats and greater lifetime value per user.

In other words, real, compounding value.

As the Indian brand ecosystem matures, Customer Forfeiture Cost needs to be as visible and valued as CAC. Acquisition is the invitation; experience is the relationship. Relationships, once broken, are expensive to rebuild; if they can be rebuilt at all.

In the end, growth isn’t just about who comes in. It’s about who stays, and why.

(Written by Devangshu Dutta, Founder of Third Eyesight, this was published in Financial Express on 2 July 2025)

What Zepto’s New Data Analytics Tool Signals For The Quick Commerce Industry

admin

May 19, 2025

Aakriti Bansal, Medianama

May 19, 2025

Zepto has launched Zepto Atom, a paid analytics product for consumer brands. The tool offers live dashboards with minute-level updates, PIN-code level performance maps, and Zepto GPT, an in-house Natural Language Processing (NLP) assistant trained on internal data.

While Blinkit and Swiggy Instamart have not announced comparable offerings, Zepto is pitching Atom as a first-of-its-kind play in quick commerce data access.

The launch comes as Zepto gears up for a public offering. The company is in talks to sell $250 million in secondary shares to Indian investors to boost local ownership ahead of its IPO. With a $5 billion valuation and a presence in just 15 cities, Zepto is seeking new ways to expand both revenue and market influence.

A strategic product in the lead-up to IPO

Zepto’s push to monetise platform tools comes at a time when it is attempting to raise its domestic shareholder base to 50%, reportedly as part of regulatory preparation for a future IPO. CLSA, in its 2024 App-racadabra report, estimates Zepto holds 28% of India’s quick commerce market despite a limited presence, trailing Blinkit at 39%.

With Zepto Atom, the company appears to turn its data infrastructure into a service layer for brands. This raises questions about how user behaviour transforms into brand-facing insight.

Zepto’s Multi-Lever Margin Play

Zepto’s cost structure is divided into warehouse transport, dark store operations, last-mile delivery, and corporate overheads. According to CLSA’s App-racadabra report, the company has achieved measurable efficiency gains across each of these categories. For instance, long-haul warehouse transport costs fell from Rs 1.7 per order in March 2022 to Rs 0.8 in February 2024. Handling costs inside dark stores declined from Rs 11 per order in June 2023 to under Rs 10 by January 2024. Last-mile delivery expenses dropped 20% between December 2023 and February 2024, from Rs 50 to Rs 40 per order.

HDFC Securities highlights three key levers for e-commerce profitability: raising average order values via premium or bundled products, improving take rates through ads and private labels, and reducing last-mile costs through better routing. Zepto has pursued these through initiatives like Zepto Café, Relish (in-house food and meat brands), the Zepto Pass loyalty program, and now Zepto Atom—signaling a multi-pronged approach to expand margins beyond logistics.

Whether brands will act on the data that Atom delivers, remains an open question.

Granular offtake data is rarely made available to brands, whether it is by offline retailers or by online platforms; so far brands have been largely flying blind, especially when it comes to marketplaces. In that sense, Zepto’s Atom can be a huge enabler and gamechanger,” Devangshu Dutta, Founder, Third Eyesight, told MediaNama.

Not All Brands May Be Ready

Zepto Atom lets brands monitor impressions, conversions, share of voice, and customer retention in near real-time.

“While having access to real-time geographical and time-stamped sales data is potentially an absolute goldmine for any brand, how useful it is will depend much more on how ready or capable the brand is to use the analysis and make adjustments to its strategy,” said Dutta.

Brands can use Zepto GPT, the NLP assistant embedded in Atom, to query platform data conversationally—for instance, to identify under-indexed Stock Keeping Units (SKUs) in a specific PIN code or analyse what’s driving category sales. However, it remains unclear how brands interpret or act on these insights in practice.

The company has not disclosed Atom’s pricing model. It also hasn’t confirmed whether access will be open to all brands or restricted to high-volume partners. These details will likely determine adoption.

How Atom Fits into the Margin Strategy

Zepto Atom’s real-time sales metrics, SKU-level performance data, and customer retention patterns align closely with the margin levers identified by HDFC Securities. By providing granular insights, Atom enables brands to fine-tune pricing, reposition products, and run targeted campaigns, potentially increasing order values, improving take rates, and optimizing delivery routes. Such adjustments could boost volumes and conversions, benefiting Zepto through higher commissions and ad revenues.

“For Zepto it is certainly a differentiator and could be a driver for additional revenue not just in terms of the subscription fees that they would charge but the incremental impact it could make on the brand partners’ sales and, by extension, on Zepto’s own overall fees/revenues,” said Dutta.

Still, widespread adoption may depend on how well Zepto supports brand onboarding and data literacy. “It may make sense for Zepto to even assist brand-side personnel in understanding how best to use the new tools and also help them create tangible operational changes on their side using the insights.”

Search behaviour and profiling concerns remain unresolved

Earlier this month, Zepto used search behaviour to curate mood-specific product categories such as “Crampy” and “Hangry,” in response to searches related to premenstrual syndrome (PMS)—a recurring condition affecting many women before menstruation. Critics told MediaNama that this kind of emotional profiling could occur without user awareness or consent.

Zepto’s privacy policy states that it collects lifestyle, health, and behavioural data for personalisation and internal analysis. However, the company does not explain whether it stores inferred data, shares it with brands, or applies it to pricing and promotions.

Whether Atom makes any of this data visible to brands remains unclear.

Why This Matters

Zepto Atom signals a shift in how quick commerce platforms are looking to generate value—not just from delivery, but from the data their ecosystems produce. With tools like real-time dashboards and search-linked behavioural insights, Zepto is turning user interactions into assets for brand partnerships.

The move raises larger questions about where platform growth is coming from. Is the business of quick commerce becoming the business of behavioural data? As brands gain new visibility through Atom, the balance between consumer experience and commercial analytics becomes harder to separate.

MediaNama has reached out to Zepto with these questions:

What specific types of consumer behaviour and purchase data are made available to brands through Atom?
Does Zepto Atom include inferred metrics such as user intent, repeat behaviour, or emotional tagging in its brand-facing dashboard?
Are brands shown real-time access to individual-level trends, or only aggregated cohort-level insights?
Are users informed that their platform activity may be used to generate commercial insights for brands?
Can users opt out of this data being shared with third parties via Atom?

As of publication, Zepto has not responded. We will update the story when we receive a response.

(Published in Medianama)

One Ring That Rules Them All

Devangshu Dutta

January 10, 2017

In this piece I’ll just focus on one aspect of technology – artificial intelligence or AI – that is likely to shape many aspects of the retail business and the consumer’s experience over the coming years.

To be able to see the scope of its potential all-pervasive impact we need to go beyond our expectations of humanoid robots. We also need to understand that artificial intelligence works on a cycle of several mutually supportive elements that enable learning and adaptation. The terms “big data” and “analytics” have been bandied about a lot, but have had limited impact so far in the retail business because it usually only touches the first two, at most three, of the necessary elements.

Elements in Operationalizing Big Data and AI

“Big data” models still depend on individuals in the business taking decisions and acting based on what is recommended or suggested by the analytics outputs, and these tend to be weak links which break the learning-adaptation chain. Of course, each of these elements can also have AI built in, for refinement over time.

Certainly retailers with a digital (web or mobile) presence are in a better position to use and benefit from AI, but that is no excuse for others to “roll over and die”. I’ll list just a few aspects of the business already being impacted and others that are likely to be in the future.

  1. Know the customer: The most obvious building block is the collection of customer data and teasing out patterns from it. This has been around so long that it is surprising what a small fraction of retailers have an effective customer database. While we live in a world that is increasingly drowning in information, most retailers continue to collect and look at very few data points, and are essentially institutionally “blind” about the customers they are serving.
    However, with digital transactions increasing, and compute and analytical capability steadily become less expensive and more flexible via the cloud, information streams from not only the retailers’ own transactions but multiple sources can be tied together to achieve an ever-better view of the customer’s behaviour.
  2. Prediction and Response: Not only do we expect “intelligence” to identify, categorise and analyse information streaming in from the world better, but to be able to anticipate what might happen and also to respond appropriately.
    Predictive analytics have been around in the retail world for more than a decade, but are still used by remarkably few retailers. At the most basic level, this can take the form of unidirectional reminders and prompts which help to drive sales. Remember the anecdote of Target (USA) sending maternity promotions based on analytics to a young lady whose family was unaware of her pregnancy?
    However, even automated service bots are becoming more common online, that can interact with customers who have queries or problems to address, and will get steadily more sophisticated with time. We are already having conversations with Siri, Google, Alexa and Cortana – why not with the retail store?
  3. Visual and descriptive recognition: We can describe to another human being a shirt or dress that we want or call for something to match an existing garment. Now imagine doing the same with a virtual sales assistant which, powered by image recognition and deep learning, brings forward the appropriate suggestions. Wouldn’t that reduce shopping time and the frustration that goes with the fruitless trawling through hundreds of items?
  4. Augmented and virtual reality: Retailers and brands are already taking tiny steps in this area which I described in another piece a year ago (“Retail Integrated”) so I won’t repeat myself. Augmented reality, supported by AI, can help retail retain its power as an immersive and experiential activity, rather than becoming purely transaction-driven.

On the consumer-side, AI can deliver a far higher degree of personalisation of the experience than has been feasible in the last few decades. While I’ve described different aspects, now see them as layers one built on the other, and imagine the shopping experience you might have as a consumer. If the scenario seems as if it might be from a sci-fi movie, just give it a few years. After all, moving staircases and remote viewing were also fantasy once.

On the business end it potentially offers both flexibility and efficiency, rather than one at the cost of the other. But we’ll have to tackle that area in a separate piece.

(Also published in the Business Standard.)