What share of revenue an Indian online store should spend on marketing, and why the ratio is an output

A marketing spend of 15 per cent of revenue is comfortable for one Indian store and fatal for another selling the same thing at the same price, and the difference is contribution margin and how often a customer comes back. So the useful question is not what percentage other people spend. It is what percentage your own unit economics can carry, which is a calculation you can do this afternoon from your own order data and your own cost sheet. Any published industry figure is an average across businesses with different margins, different repeat rates and different return rates, and acting on it is how a growing store runs out of cash while hitting its revenue target.

This page gives the calculation, then the four Indian cost lines that change the answer and are routinely left out of it.

The calculation, and the order to do it in

• Start with net revenue, not gross. Take orders placed, subtract cancellations, subtract returns, subtract any order that came back undelivered. That is the revenue your marketing actually produced.
• Compute contribution per order. Net selling price, minus cost of goods, minus shipping you paid, minus payment or platform charges, minus packaging, minus the cost of handling the returns and undelivered orders in that cohort. Leave out rent, salaries and software for this step.
• Count orders per customer per year from twelve months of your own data. Distinct customers, total orders, divide.
• Multiply. Contribution per order times orders per customer per year is the gross contribution one acquired customer gives you in a year.
• Decide what share of that you are willing to spend. Spend all of it and you grow with no profit. Spend half and you have a business that funds itself. That decision, not a benchmark, sets your acquisition budget.
• Convert to a percentage at the end. Annual marketing spend divided by annual net revenue. Now the number means something, because you built it from your own costs.

A worked example, with your own numbers to substitute

The figures here are an illustration and not a finding about anybody's store. A product sells at Rs 1,200. Cost of goods Rs 500, shipping Rs 90, payment and platform charges Rs 40, packaging Rs 20. That is Rs 550 of contribution on a delivered order. If 12 per cent of orders come back undelivered or are returned and each one costs Rs 150 in forward and reverse shipping, the cohort loses about Rs 84 per order placed, bringing contribution to roughly Rs 466. If the average customer orders 1.6 times a year, one acquired customer contributes about Rs 745 in the year. Spending Rs 370 to acquire that customer leaves about half, and expressed against revenue that is a defensible ratio. Spending Rs 700 looks like a 58 per cent payback and is a business that grows and never pays anybody.

Run the same arithmetic with your own four inputs and the ratio falls out. If it comes out uncomfortably low, the lever is almost never the marketing. It is the cost of goods, the shipping deal, the return rate or the second order.

The four Indian lines that change the answer and get left out

• Undelivered cash on delivery orders. This is the single biggest hidden cost in Indian ecommerce and it appears nowhere in a revenue figure. An order that goes out, is refused and comes back costs you forward shipping, reverse shipping and the handling, and produces no revenue at all. Measure the rate by payment method, because prepaid and cash on delivery behave completely differently, and discounting prepaid is often cheaper than absorbing the returns.
• Marketplace commission and fulfilment fees. If part of your revenue comes through a marketplace, those orders have a different contribution per order from your own site, so a blended ratio hides which channel is paying. Keep two columns.
• Returns in apparel and footwear. Size driven returns in these categories are structural rather than a service failure, and the only fix that moves the number is better size information on the product page.
• The tax treatment of your own costs. What is recoverable and what is not changes the real cost of several lines above. Your accountant settles this, and it is worth asking before you fix the budget rather than after.

Three numbers to track instead of the percentage

The third one is the one to put in front of whoever funds the business. A payback of 1.8 orders in a category where customers buy 1.6 times a year is a business that needs more cash every month it grows. The same payback where customers buy four times a year is healthy. That is why one published percentage cannot serve both, and our page on what customer acquisition cost means and the six metrics it gets confused with is worth reading before you quote a figure to anybody.

What to do when the ratio is too high

• Fix the second order before cutting the first. A customer who buys twice halves your effective acquisition cost without any change to the advertising, and the cheapest channel for that is the one you already own.
• Attack the return and undelivered rate. Better size and specification detail, prepaid incentives, and address confirmation before dispatch. Each point of improvement goes straight to contribution.
• Separate the channels before you cut any of them. Blended cost per acquisition hides the one channel that is profitable, and cutting the budget proportionally usually cuts that one too.
• Raise the average order value. A bundle or a size upgrade adds contribution without adding acquisition cost, which moves the ratio faster than any advertising change.
• Check which strategy your repeat rate actually puts you in. Our page on the number that chooses between the three ecommerce strategies uses purchases per customer per year to settle whether you are an acquisition business or a retention one, and spending against the wrong one is the most expensive version of this mistake.

Questions Indian store owners ask about marketing spend

Is there any useful rule of thumb at all

The useful one is a limit rather than a target. Never spend more to acquire a customer than the contribution you expect from them inside twelve months, and know which of those two numbers you are least sure about. Everything else comes from your own cost sheet.

Should the marketing percentage include the agency fee

Yes, as its own line, with media spend separate. A single blended figure makes it impossible to see whether a change came from the fee or from the media, and both move for different reasons.

Does the ratio change as the store grows

It usually falls, because returning customers become a larger share of revenue and they cost almost nothing to reach. If it is not falling as you grow, the repeat rate is flat and that is the problem to work on.

How do I treat a festive season spike

As its own period with its own ratio, and compare it against the same festive window last year rather than against the month before it. Indian demand is seasonal enough that a month on month comparison across a festival is not a comparison.

What about brand spend that produces no immediate order

Budget it as a separate line with its own expectation, and judge it on branded search volume and direct traffic over quarters rather than on orders this week. Mixing it into the acquisition ratio makes both numbers unreadable.

My marketplace sales are most of my revenue. Does any of this apply

It applies per channel. Compute contribution per order separately for marketplace and for your own site, because the commission, the fulfilment fee and the return handling are different, and the decision you are making is where the next rupee goes.

The spend line that has no cost per click

One route to an Indian online store's product now carries no media cost and appears in no ratio. A shopper describes what they want to ChatGPT or Gemini, and the answer names products, quotes a price and sometimes recommends where to buy. Whether your product is in that answer depends on what your own product pages say in text, which our page on the declarations an Indian store already prints works through declaration by declaration. When we put 198 India buying questions to both engines on 22 October 2026, with the country written into the wording of each one, 91.1 per cent of the 1,606 sources behind those answers were pages on a company's own website, and no marketplace catalogue page appeared among them at all. AI Knows Us measures which way that answer currently goes for a specific brand, which is a channel with a setup cost and no cost per order.