E-commerce & D2C Growth Agency

We grow profit, not vanity ROAS

Most e-commerce brands don’t fail on traffic; they fail on economics. A big ROAS number on the dashboard can hide orders that lose money after product, shipping, payment and returns, so we start from contribution margin and CAC payback, not screenshots. We scale only what is already profitable and repeatable, pair acquisition with retention (the second and third order are where D2C actually makes money), and model quick commerce and marketplaces on true margin before pouring in budget. The goal isn’t the biggest revenue line you can’t sustain; it’s efficient, compounding growth the unit economics can carry.

Beyond ROAS: grow profit, not just revenue

A great ROAS can still hide a business losing money per order. We instrument contribution margin, LTV by cohort and blended CAC, so budget scales only where the unit economics work, read Beyond ROAS, and see our AI & Data Analytics.

Win the new channels, quick commerce included

Quick commerce is now a core D2C channel (ad spend surged ~202% in a year). We run the digital shelf and on-platform ads on Blinkit, Zepto and Instamart alongside Meta and Google, see our quick-commerce playbook and Performance Marketing.

A store that converts and a retention engine that compounds

Fast, high-converting Shopify or headless storefronts (Web & Product Engineering) plus WhatsApp-led lifecycle and win-backs (MarTech & Automation), because repeat rate is the quiet engine of profitable D2C.

Creative velocity, measured

Brands that test 15–20 creatives a week consistently beat those running the same three for months. We pair that velocity with attribution that ties every test to contribution margin, not just clicks.

What profitable D2C growth actually costs

The honest answer to ‘what should I spend?’ in D2C is: whatever keeps CAC below what a customer is worth, and that depends entirely on your margin and repeat rate, not a benchmark from another brand. A product with a 70% margin and strong repeat purchase can afford a much higher acquisition cost than a thin-margin, one-and-done product. So we start from the unit economics: contribution margin after all true costs, the LTV a cohort actually delivers, and the CAC payback window you can fund. Only then does a spend number mean anything.

Where the money works hardest shifts as you scale. Early on, creative and offer beat everything, testing angles cheaply until something clicks. As you grow, the leverage moves to retention (the second and third order is where D2C makes its money) and to efficiency: tighter creative testing, a faster store, and quick-commerce or marketplace channels modelled on true margin before you scale them. The brands that last aren’t the ones with the biggest ad budgets; they’re the ones whose economics let them keep spending. Model your own numbers on the ROAS calculator, and read Beyond ROAS for the metrics that actually predict profit.

We start with a diagnosis, not a campaign plan

Almost every e-commerce brand that calls us describes the same symptom. Sales are flat or falling, ad costs are up, and the current agency keeps sending reports full of green arrows. The symptom is real. The cause is almost never the one being reported on. So before we touch an ad account we work out where the money is going, which is a different question from where the clicks are going.

There are only a handful of places a store leaks. Contribution margin is too thin to support paid acquisition at any price, which means the problem is pricing or cost of goods or shipping, not media. Traffic is fine but the product pages do not convert, which is a store problem. Conversion is fine but nobody comes back, which is a retention and product problem. Or acquisition is genuinely mispriced because the account structure is wasteful and the targeting is buying people who were going to purchase anyway. Each of those looks identical from the outside. Each needs a completely different intervention, and running the wrong one for six months is how brands lose a year.

The diagnosis takes two to three weeks and produces plain answers. What each SKU earns after every deduction. Where the funnel drops, measured on real device conditions rather than a desktop in an office. What the cohort curve looks like. Which channels are claiming credit for orders that another channel also claims. We would rather tell you in week three that your ad spend is not the problem than take a media retainer for a year and manage the symptom.

The channels we run, and how we decide the mix

Paid search and shopping, paid social, marketplace advertising where you sell on marketplaces, email and WhatsApp retention, and the organic and technical SEO work that carries category and product pages. Not all of them for every brand. The mix is decided by two things: where your category’s demand already exists, and what your contribution margin can afford.

If people already search for your product by name or category, search and shopping come first, because you are harvesting demand that exists rather than manufacturing it. That is cheaper, it converts faster, and the feed work behind it is the highest-return technical job in the account. If your product is something nobody knows to look for, a new format, a new ingredient, a category that did not exist three years ago, then paid social carries the load and search only picks up later as brand awareness creates search volume. Getting that order wrong is expensive. Brands with genuine search demand who spend everything on social discovery are paying to create interest that Google was already serving them for a fraction.

The feed deserves specific attention because it is invisible until you look. Product titles that match how people search rather than how your catalogue system names things. Correct categorisation. Sizes, colours and identifiers populated. Availability syncing accurately so you are not advertising items you cannot dispatch. Images that meet requirements without being silently downranked. A neglected feed will quietly cap what shopping campaigns can do no matter how the bidding is set, and we have opened accounts where a third of the catalogue was disapproved and nobody in the building knew.

Retention runs alongside from month one, not as a phase two. It is the cheapest revenue in the business and it is usually the most neglected, sitting in a tool somebody set up during launch and never revisited. Cart recovery, browse abandonment, post-delivery follow-up, replenishment, winback. Built once. Tuned quarterly.

A worked example: what one order is really worth

Take a single product and follow the money. Selling price ₹1,500. Product cost ₹520. Packaging ₹35. Forward shipping ₹90. On a prepaid order that stays delivered, subtract a payment gateway charge of roughly two percent, about ₹30, and the contribution before any marketing is around ₹825. That is the ceiling on what you can pay to acquire that order and still be above water on the first sale.

Now add returns. Say one in six orders comes back, and a returned unit costs you the forward shipping, the return leg at ₹80, and the packaging, so ₹205 gone with no revenue against it. Across six orders you collect five contributions of ₹825, which is ₹4,125, minus ₹205 for the one that returned. That is ₹3,920 across six dispatched orders, or roughly ₹653 each. The number moved by more than a fifth and nothing about the product or the ad account changed.

Set your acquisition target against ₹653, not ₹825. Then decide how much of the first order you are willing to give away in exchange for a customer who buys again, which is a decision only your cohort table can inform. If 30 percent of buyers return within 90 days at similar margin, you can afford to acquire at close to break-even on order one. If almost nobody returns, every order has to pay for itself the day it ships, and your whole media strategy changes shape.

The mistakes that cost the most in this category

Judging channels on platform-reported ROAS. Meta and Google both count the same order, so adding their reported revenue produces a number larger than your actual sales. Decisions made on that inflated view push spend towards whichever platform claims most aggressively, not whichever platform is genuinely producing incremental orders. Use blended MER for the business, platform numbers for comparing ad sets inside a single account, and never mix the two in the same argument.

Ignoring returns until the quarterly review. Returns to origin and post-delivery returns both consume shipping and handling on units that produced nothing, and if you are not tracking them by product and by courier you are running campaigns against margins that do not exist. Second: a standing discount code. It looks like a conversion tool and it functions as a permanent price cut, applied hardest to the customers who would have paid full price. Third: chasing new customer acquisition while the repeat rate quietly decays, which is expensive, slow to notice, and usually visible in the cohort table months before it shows up in revenue.

Then there is the store itself. Brands spend six figures a month sending traffic to product pages that load slowly on the mid-range Android handsets most Indian shoppers actually use, that hide the delivery estimate until checkout, that state a returns policy in language written by a lawyer for a lawyer, and that offer three photographs for a product people need to inspect before spending three thousand rupees on it. Fixing that page costs a fraction of a month of media and lifts every campaign pointed at it, permanently. It is the least fashionable work in the discipline and reliably the best return.

How we measure, and what we refuse to report

Four numbers run the account. Blended MER, which is total revenue divided by total advertising spend, with no attribution argument possible. Contribution per order after product cost, shipping both ways, packaging, gateway charges and returns. Repeat purchase rate by cohort at 60, 90 and 180 days. And new customer acquisition cost, calculated against new customers only rather than all orders, because a rising blended number can hide the fact that you are buying fewer new people and living off returning ones.

Those four sit at the top of every report. Underneath them go the working numbers, the channel splits, the creative performance, the feed health, the page speed. What does not appear anywhere is impressions, reach, follower counts, engagement rate, or a ROAS figure quoted without the spend it was earned on. Those are the numbers agencies reach for when the real ones are not moving, and once you have seen a few decks you can spot the substitution instantly. The report should be readable by your finance lead without a translator, and if a metric cannot be connected to rupees in two steps it does not belong on page one.

Reporting is monthly written, with a call. Weekly, you get a short dashboard view of spend and MER so nothing runs away between meetings. Quarterly we go back to the economics model and rebuild it, because product costs change, shipping rates change, return behaviour changes with season and with whatever you added to the catalogue, and targets set against last quarter’s margins slowly stop being true. That rebuild is where most of the real decisions get made.

Who this suits, and who it does not

This works for brands doing meaningful monthly revenue who already have product-market fit and a store that ships reliably. It works when the founder or the commercial lead will sit in the economics conversation rather than delegating it, because half the decisions are pricing and product decisions wearing a marketing costume. It works when there is enough patience to spend the first three weeks on diagnosis. Not everyone has that patience. That is fine, but it is a bad match for us.

It does not suit brands looking for someone to run ads at a fixed ROAS target and nothing else. It does not suit a business whose unit economics cannot support paid acquisition at any efficiency, where the honest answer is to fix pricing or sourcing first and we will tell you that in the first call rather than the tenth month. And it does not suit anyone who wants a guaranteed revenue number, because nobody controls the auction, the season, or your competitor’s decision to discount in November.

We work with brands across Delhi NCR from Gurugram, and with Indian brands selling into the USA, which brings its own problems around shipping economics, duties, and a market where COD is not a factor but acquisition costs sit in a different band. If you sell in both, they are two businesses in one admin panel, and should be modelled separately.