Scale profitably, not just loudly.
From storefront to scale — we build, optimise and advertise D2C brands for profitable, compounding growth.
Common challenges
- Rising ad costs
- Low repeat rate
- Cart abandonment
How we solve them
- Performance ads
- CRO & funnels
- Retention marketing
Recommended services for e-commerce & d2c
Go deeper on e-commerce & d2c
What we actually do differently for e-commerce & d2c, service by service.
Ready to grow your e-commerce & d2c brand?
Let's talk about where the growth is — and how we'll get you there.
Indian D2C has a specific arithmetic problem that brands in the US and UK do not face at anything like the same scale: cash on delivery. Depending on category, price point and geography, COD still accounts for roughly a third to two-thirds of orders for most Indian D2C brands, and COD orders are returned to origin at rates that commonly run 15–35% against 3–8% for prepaid. A returned-to-origin order costs you forward shipping, reverse shipping, packaging, handling and often a damaged unit — while the platform that generated it has already counted the purchase as a conversion and charged you for it. Every ROAS figure computed on gross order value is therefore inflated by an amount you cannot see in the ad account.
The second reality is channel conflict. Amazon and Flipkart give you demand you would otherwise pay to create, and they take a commission plus fulfilment fees, own the customer relationship, and expose your price to buy-box dynamics you do not control. Your own site gives you margin, first-party data and retention, and requires you to buy every visitor. These two channels have genuinely conflicting economics, and running them off one blended P&L conceals which one is actually funding the business. The brands that scale profitably usually run them as separate businesses with separate SKUs, separate pricing logic and separate targets.
Third, acquisition costs have been rising for years and creative is the only durable lever left. Post-ATT signal loss pushed Meta's optimisation towards broad targeting with creative doing the segmentation, and Google's Performance Max moved budget allocation inside a box you cannot fully inspect. In that environment the advertiser's controllable inputs shrink to three: the quality and quantity of creative, the accuracy of the conversion values you send back, and the margin structure of what you are selling. Everything else is dashboard maintenance.
How Indian D2C customers actually buy
Discovery is overwhelmingly social and increasingly video-led. Instagram Reels, YouTube Shorts and creator content do the work that a category page used to do — the customer does not search for your product type, they encounter yours. This matters for measurement: the ad that created the demand is frequently not the ad that gets the last click, and last-click reporting will quietly defund the campaigns doing the actual work.
Verification happens off your property, and you cannot prevent it. A shopper who sees your product on Instagram will check whether it is on Amazon, compare the price, and read the reviews there even if they intend to buy from your site — Amazon functions as a review database for Indian shoppers regardless of where the transaction lands. If your marketplace listing is priced lower, poorly photographed or carries three one-star reviews, that is your D2C conversion rate problem too.
Then the objection stack: sizing, fabric or ingredient specifics, delivery timeline to their pincode, returns policy. These are answered on the product page or they are answered by leaving. At mid ticket sizes Indian shoppers are markedly more delivery-date sensitive than price sensitive — a visible, pincode-accurate delivery estimate moves conversion more than another five per cent off.
Checkout is where the money leaks. Forced account creation, a five-field address form on a phone, no UPI option, and a coupon field that sends people off to hunt for a code all cost real orders. UPI has become the default payment expectation, and a checkout that treats it as a secondary option alongside cards is fighting its own customers. COD, meanwhile, is usually not a payment preference — it is a trust hedge. Customers choose it because they are unsure the product will arrive as described, which means the fix is credibility and delivery confidence, not simply switching COD off.
Post-purchase is where D2C economics are actually decided. First orders are frequently acquired at or below break-even; the margin is in the second and third. WhatsApp has become the highest-engagement retention channel in India by a wide margin — order updates, replenishment nudges and re-order links get read in a way that email does not. Brands that treat retention as a CRM afterthought end up permanently dependent on paid acquisition and permanently squeezed as its cost rises.
What each channel is actually for
Not every channel does the same job in e-commerce & d2c. Here's how we use each one.
Meta Ads
Meta is the primary demand-creation engine for Indian D2C and functions as a creative-testing operation more than a media-buying one. Once tracking is sound and the account structure is simple, the ceiling on spend is set almost entirely by how many genuinely distinct creative angles you can produce and validate each month.
- Run the Conversions API alongside the pixel with proper event deduplication — browser-only tracking loses a material share of signal and systematically under-reports the campaigns doing top-of-funnel work
- Send order value net of expected discounts and expected RTO by category, not gross order value, so bidding learns towards profitable cohorts
- Consolidate ad sets and let broad targeting plus creative do the segmentation; over-segmented audiences fragment learning and inflate frequency
- Maintain a creative pipeline with a fixed weekly cadence of new hooks — the same three ads running all quarter is the single most common reason a scaling account stalls
Google Ads & Shopping
Google captures the demand Meta and creators create, plus the category-level searches you did not generate. Shopping and Performance Max are efficient at finding cheap conversions and entirely indifferent to whether those conversions make money, which makes feed values and exclusions the whole job.
- Feed margin-adjusted values through Merchant Center custom labels and conversion value rules so hero SKUs and thin-margin SKUs stop competing for the same budget
- Segment Performance Max by margin tier rather than by product category, and keep brand search out of PMax so it cannot claim credit for demand you already own
- Fix the Merchant Center feed before touching bids — GTINs, correct categorisation, accurate availability and complete attributes affect delivery more than bid strategy does
- Run a defensive brand campaign; in India resellers and marketplace listings bid on D2C brand terms routinely, and losing that click costs you the whole margin
SEO & technical health
Most Indian D2C stores are broken for search before anyone writes a blog post. Faceted navigation on Shopify and WooCommerce generates thousands of near-duplicate URLs, category pages carry no unique content, and crawl budget goes to parameter combinations nobody will ever search. Fixing that usually beats content investment for the first quarter.
- Contain faceted navigation — canonicalise or noindex parameter combinations, and keep only the filter permutations with genuine search demand indexable
- Build category pages that can rank in their own right, with unique copy, internal linking and buying guidance above the product grid
- Implement Product, Offer, AggregateRating and Review schema correctly, and keep availability and price accurate — stale structured data suppresses rich results
- Report organic revenue and organic new customers, never organic sessions; discount-code and blog traffic can triple sessions while contributing nothing
Website & checkout conversion
For a D2C brand the storefront is the P&L. Load time and checkout friction move revenue in ways that another month of ad spend rarely matches, and unlike media, the gain compounds and persists after the budget stops.
- Treat UPI as the primary payment method, not an option below cards, and keep the checkout to two steps with guest checkout available
- Show a pincode-accurate delivery estimate on the product page — delivery certainty converts better at mid ticket sizes than an equivalent discount
- Answer the objection stack in the product page structure: sizing with real measurements, ingredient or material detail, and the returns policy stated plainly rather than linked in the footer
- Measure Core Web Vitals from real Indian mobile field data rather than a desktop lab score; a page that passes in a lab and takes seven seconds on a mid-range Android is failing where it counts
Retention & lifecycle
Acquisition economics only work if the second order exists. Retention is where contribution margin is actually earned, and in India the channel mix for it looks different from the Western default — WhatsApp carries engagement that email cannot match, though email still owns the long nurture.
- Build replenishment flows on real consumption cycles per SKU, not on a generic 30-day timer
- Use WhatsApp for transactional and time-sensitive messaging within the template rules, and keep promotional volume low enough that customers do not block the sender
- Run prepaid-incentive flows on COD orders — a small discount or free shipping offered at checkout to convert COD to prepaid frequently pays for itself several times over in avoided RTO
- Segment on contribution margin and purchase frequency rather than RFM scores alone; a high-frequency customer buying only discounted thin-margin SKUs is not your best cohort
The mistakes we see most often
Every one of these is something we've inherited from a previous agency in e-commerce & d2c.
Optimising to ROAS computed on gross revenue
Gross revenue ignores COD returns, shipping, payment gateway fees, discounts and cost of goods. A 4x ROAS on a category with a 30% RTO rate and 35% gross margin is a loss-making campaign that every report will describe as a success, and the platform will helpfully find you more of exactly that traffic.
Send margin-adjusted conversion values back to Meta and Google, discounted by category-level expected return rate. Report contribution margin per rupee of spend alongside ROAS, and set targets against the former. This is a data-plumbing project, not a bidding tweak, and it is usually the highest-value thirty days of work in a D2C account.
Treating COD as a fixed cost of doing business in India
It is treated as immutable because customers demand it, but COD share is substantially influenceable. Brands with strong review depth, clear delivery estimates and visible return policies consistently run higher prepaid mixes than competitors selling the same category, because COD is a trust hedge rather than a payment preference.
Attack it on three fronts: prepaid incentives at checkout, address and phone verification before dispatch to filter fake orders, and pincode-level RTO analysis so you can restrict COD where the data justifies it. Moving prepaid share up ten points typically does more for profitability than any bidding change.
Letting Performance Max absorb brand search
PMax will happily serve on your own brand terms, book those conversions as its own, and report a spectacular ROAS built almost entirely from demand you already generated. Budget then flows towards the campaign that is doing the least incremental work.
Run brand as a separate exact-match search campaign, add brand terms to the account-level negative list or brand exclusion list for PMax, and compare PMax performance before and after. Expect the reported number to fall and the actual business to improve — a distinction worth explaining to stakeholders before you make the change, not after.
Running the same P&L for marketplace and own-site
Amazon and Flipkart carry commission, fulfilment fees, storage and mandatory participation in platform discounting; your own site carries acquisition cost and payment fees. Blending them produces an average that describes neither, and typically hides that marketplace volume is subsidising a D2C channel that is not yet working, or the reverse.
Separate contribution-margin reporting per channel, with a deliberate SKU strategy — differentiated packs or variants for marketplace, so price comparison does not force you into a race you cannot win on your own site. Decide explicitly which channel is for acquisition and which is for margin.
Buying growth with discounts and calling it demand
A 25% launch discount will produce revenue, a healthy ROAS and a customer base that has been trained never to pay full price. The second-order effect shows up two quarters later as a repeat cohort that only converts on promotion, and margin that never recovers.
Use bundles, free shipping thresholds and product-value additions in preference to headline percentage discounts, and track repeat rate by acquisition-discount level. If discounted cohorts repeat materially worse, that is your answer, and it is a number you can compute this week from existing order data.
The numbers that actually matter
Contribution margin per rupee of ad spend
Revenue after cost of goods, shipping both ways, payment fees, discounts and expected returns, divided by media spend. It is the only acquisition metric that cannot be improved by selling unprofitably, which is exactly why it is uncomfortable and worth adopting.
Blended MER
Total revenue divided by total marketing spend across all channels. It survives attribution disputes, iOS signal loss and platform self-reporting, because it is computed from your bank account rather than from an ad dashboard. Track it alongside platform ROAS and treat divergence as a signal.
New-customer acquisition cost
Blended CAC flatters itself with repeat orders. Separating spend against genuinely new customers reveals whether you are growing or simply re-selling to the same cohort at increasing cost — the distinction that determines whether scaling budget is safe.
RTO rate, segmented by pincode, category and payment method
The unglamorous number that decides Indian D2C profitability. Segmented properly it becomes actionable — restrict COD in the worst pincodes, price the risk into thin-margin SKUs, and stop treating a national average as a fact of nature.
Prepaid share of orders
A single number that moves gross margin, working capital and RTO exposure simultaneously. It is also directly influenceable through checkout design and incentives, which makes it one of the few operational levers a marketing team genuinely controls.
Repeat purchase rate at 90 and 180 days
It tells you whether the business compounds or leaks. Two brands with identical CAC and identical first-order margin have completely different valuations if one repeats at 12% and the other at 35%, and no amount of ad optimisation closes that gap.
Vanity metrics to ignore
Sessions, add-to-cart rate in isolation, follower count, platform-reported ROAS taken at face value, and gross merchandise value. Each can be inflated by discounting, cheap traffic or double-counting across platforms — GMV in particular has been the favoured metric of every unprofitable Indian e-commerce story of the last decade.
Consumer protection, dark patterns and data rules
The Consumer Protection (E-Commerce) Rules, 2020 set the baseline: accurate seller identity and contact details, clear return, refund, exchange and warranty terms, a named grievance officer with published contact details and defined response timelines, and no manipulation of price to exploit consumer behaviour. These are not decorative footer requirements — they are the pages a Central Consumer Protection Authority enquiry will look at first.
The CCPA's 2023 guidelines on dark patterns are the ones that most directly collide with conventional CRO practice, and they name thirteen specified patterns. False urgency — countdown timers that reset, 'only 2 left' claims that are not tied to real stock — is on the list. So are basket sneaking (adding items or donations by default), drip pricing (revealing shipping or fees only at the final step), confirm shaming ('No thanks, I don't want to save money'), forced action, subscription traps and interface interference. A meaningful share of the tactics in circulation on Indian D2C storefronts sit somewhere in that list. Genuine scarcity displayed accurately is fine; manufactured scarcity is now a specified violation.
Product listings carry their own statutory declarations. The Legal Metrology (Packaged Commodities) Rules require e-commerce listings to display the declarations that appear on the package — MRP inclusive of all taxes, net quantity, country of origin, manufacturer or packer or importer details, consumer care contact and, where applicable, date of manufacture and best-before. Missing country of origin on marketplace and own-site listings is one of the most commonly cited lapses. Category-specific regimes stack on top: FSSAI licensing and labelling for food and nutraceuticals, BIS certification for a growing list of goods, and Drugs and Cosmetics rules for anything making a cosmetic or therapeutic claim.
Influencer marketing has its own disclosure regime. The CCPA's endorsement guidelines require clear, prominent, hard-to-miss disclosure of any material connection between a brand and an endorser, and the disclosure must sit where the consumer will actually see it rather than buried in a caption or a hashtag block. The liability extends to the brand, not only the creator. Finally, the Digital Personal Data Protection Act, 2023 governs the customer data your store accumulates — consent, stated purpose, retention limits and breach notification. Third-party pixels, chat widgets and analytics scripts all pass personal data to processors you are accountable for, which is worth auditing before it becomes a question someone else asks.
This is a practitioner's summary of how these rules affect campaign work, not legal advice. Get your own counsel to review anything you publish.
What the first 90 days look like
Fix the measurement
- Build the unit economics model — COGS, shipping both ways, payment fees, discount rate and category RTO — before touching any campaign
- Deploy Conversions API with deduplication and server-side tagging; audit consent and data flow while doing it
- Push margin-adjusted conversion values into Meta and Google rather than gross order value
- Audit product listings and checkout against dark-pattern guidelines and Legal Metrology declarations
- Establish baselines for blended MER, new-customer CAC, prepaid share and RTO by pincode
Rebuild the funnel
- Simplify account structure — consolidated ad sets, brand pulled out of Performance Max, margin-tier segmentation in Shopping
- Ship checkout changes: UPI first, two steps, guest checkout, prepaid incentive on COD orders
- Fix faceted navigation and canonicals, and rebuild the top category pages to rank
- Launch the creative pipeline with a fixed weekly production and testing cadence
- Add pincode-level delivery estimates and address verification before dispatch
Scale on margin
- Reallocate budget on contribution margin per rupee rather than platform-reported ROAS
- Separate marketplace and own-site P&Ls with a deliberate SKU and pricing strategy for each
- Build replenishment and win-back flows on WhatsApp and email against real consumption cycles
- Run structured incrementality checks — geo holdouts or spend-level tests — on the channels claiming the most credit
- Set the ongoing reporting rhythm around MER, nCAC, prepaid share and 90-day repeat rate
E-commerce & D2C questions
Because ROAS is computed on gross order value and your bank balance is not. Work backwards: subtract cost of goods, forward and reverse shipping, packaging, payment gateway fees, discounts and the orders that came back RTO. In a category with a 30% COD return rate and 35% gross margin, 4x reported ROAS can be comfortably loss-making. The fix is to feed margin-adjusted values back into the bidding so the platforms optimise towards the same number you are judged on, and to expect reported ROAS to drop when you do.
Both, run as separate businesses. Marketplaces give you demand and reviews you would otherwise buy, and Indian shoppers verify products there regardless of where they eventually purchase — an absent or poorly-reviewed marketplace listing damages your own site's conversion rate. What you should not do is run one price list and one P&L across both. Differentiated packs or variants for marketplace, separate contribution-margin targets, and an explicit decision about which channel is for acquisition and which is for margin.
Rarely, and never as a blanket change. For most Indian D2C categories COD removal costs more in lost orders than it saves in returns, particularly outside metros and for first-time customers. The better sequence is: analyse RTO by pincode and category, restrict COD only where the data clearly justifies it, add address and phone verification before dispatch, and run a prepaid incentive at checkout. Brands that do this typically move prepaid share up substantially without losing volume — which is a far better outcome than switching COD off and discovering how much of your demand depended on it.
Creative volume and creative variety, in that order. With broad targeting now the default and the algorithm doing the segmentation, the number of genuinely different angles you put into the auction is the binding constraint on how much you can spend profitably. That means a production process, not a burst of ads — a fixed weekly cadence of new hooks, formats and messages, with clear rules for what gets killed. Everything else, including whatever bidding trick is currently circulating, is marginal by comparison.
Only if repeat purchase rate justifies it, and for most brands it does not yet. Below roughly two orders per customer per year, a fast mobile site with good retargeting and WhatsApp retention beats an app on every measure including cost. The honest test is to build the retention flows first and look at 180-day repeat rate. If it clears the bar, the app should exist to serve one-tap reorder and genuine replenishment push — not to be a slower copy of your website with an install requirement in front of it.
Technical fixes — faceted navigation, canonicals, page speed, structured data — often show up in three to eight weeks because you are recovering existing crawl and index efficiency rather than building new demand. Category and content ranking typically takes six to twelve months in competitive D2C categories. Judge it on organic revenue and organic new customers, not sessions; it is easy to triple organic traffic with coupon-code and informational content that never buys anything.
Let's build your growth engine.
Book a free strategy call. We'll audit your funnel and show you exactly where the growth is.