Stack advice tends to come in two unhelpful sizes. There is the beginner version, which is a list of apps and a Shopify plan. And there is the enterprise version, which assumes a solution architect, a data team and a budget with a comma in an unfamiliar place.
The brands in between get very little written for them. So here is the version for a specific, common moment: an Indian D2C brand crossing roughly ₹1 crore a month, selling on its own store plus one or two marketplaces, shipping from one or two locations, with a team where nobody's full-time job is systems.
Two notes before the list. First, costs below are published market figures, not our rate card, and Indian delivery economics usually land lower than global equivalents. Second, the order matters more than the list - buying the right things in the wrong sequence is how stacks become expensive without becoming useful.
The layers, in buying order
1. Platform and payments. You already have this, and it is the least interesting line in the budget. The question at this size is whether to move up a tier. Published guidance puts Shopify Plus at roughly ₹1.65 lakh a month in India and suggests it earns its keep above about ₹15-20 crore a year in GMV. At ₹12 crore a year you are near that border, which means the decision should be driven by a named requirement - checkout extensibility, B2B pricing, multi-storefront, higher API limits - and not by the milestone. If you cannot name the requirement, stay where you are and spend the difference on layer three.
2. Logistics and courier aggregation. Non-negotiable by this point, and usually already bought. What is often missing is serviceability-aware routing and a real view of courier performance by pincode. The brands that get value here are not the ones with the cheapest rate card, they are the ones who can tell which courier fails in which region and act on it.
3. Order orchestration. The highest-return purchase at this stage, and the one most often postponed in favour of marketing tooling. This is the layer that holds the authoritative order state across store, marketplaces, warehouse and courier, turns routing from a human judgment into a rule, and treats exceptions as defined states rather than as somebody's afternoon. Whether you need it is a question with a concrete answer - the eleven signals are in when a Shopify brand actually needs an OMS, and the cost drivers are broken down in what a custom OMS actually costs.
4. Integration layer. Needed when several systems write to the same record. Published 2026 ranges put iPaaS-style Shopify and ERP integration at roughly $10,000 to $40,000 implementation with $500 to $3,000 a month platform fees for two to four connected systems, plus twenty to forty percent for the parts buyers underestimate. Many brands at ₹1 crore a month do not need this yet, and the signals that you do are in when to stop adding Shopify apps.
5. Support and communication. A shared inbox that covers WhatsApp, email and marketplace messages, connected to order state. The connection is the point. Support software that cannot see the order turns every query into a tab-switching exercise, which is where the response-time complaints come from.
6. Reporting, last. Deliberately last. A reporting layer built on records that disagree renders disagreement faster, as covered in why your ecommerce numbers never match. Buy it once the definitions exist.
What it actually costs
The useful way to hold the budget is in three buckets rather than as a list of subscriptions.
Software subscriptions are the smallest bucket and the one everyone optimises. Platform, courier aggregator, support desk, a handful of apps. Published data shows the average paid Shopify app costs somewhere around $30 to $60 a month and the typical store runs six to eight of them, so for most brands this bucket is real but not decisive.
Implementation and integration is the bucket that surprises people. Published 2026 TCO guidance repeatedly lands on the same finding: total cost of ownership runs at a multiple of the advertised subscription - commonly 2.5 to 4 times - once implementation, integrations, training, user growth and annual increases are counted. Maintenance on integration work runs around fifteen to twenty-five percent of implementation cost annually.
People is the bucket nobody puts in the stack conversation at all, and it is usually the largest. Every tool that is not genuinely self-running consumes attention from someone whose time has a price. A tool that saves four hours and costs six hours of coordination is a negative purchase that will never show up as one in the spreadsheet.
Compare across all three buckets or the comparison is decorative.
The four things to not buy yet
ERP. It solves finance, procurement, manufacturing and consolidation. Your pain at this size is order orchestration. Buying ERP for order problems is the most expensive common error at this revenue, and we separate the layers in Shopify Flow vs OMS vs ERP.
A PIM, unless catalogue complexity is actually your bottleneck. Many variants across several channels with differing attribute requirements justifies one. Two hundred stable SKUs on two channels does not. Published guidance warns migration and onboarding add twenty to forty percent to the first-year budget, which is a poor trade for a problem you do not have.
A customer data platform. At this scale the data problems are upstream - definitions, order state, returns - and a CDP built on them inherits every one of them at a higher subscription.
A second analytics tool to explain the first one. When two dashboards disagree, a third does not adjudicate. Write the definitions instead; it costs a day.
The sequencing rule that keeps stacks sane
Buy the thing that removes the most human decisions per order, then stop and run it for a quarter before buying the next thing.
That rule sounds conservative, and it is the single most useful constraint we see. Stack damage at this size is rarely caused by buying something bad. It is caused by buying three things in one quarter, none of which gets configured properly, all of which now write to the same order, and none of which anybody owns.
A brand that buys one layer, configures it properly, documents who owns which field, and only then looks at the next gap will end the year with a smaller stack and better operations than a brand that bought twice as much.
What changes at ₹25 lakh a month and at ₹5 crore a month
The layers do not change with revenue. The build-or-buy answer for each one does.
Below roughly ₹25 lakh a month, almost everything on this list should be a spreadsheet and a person. One channel, native platform fields, a shared sheet for stock, and manual reconciliation at month end. That sounds unsophisticated and it is correct. At that volume a tool costing ₹15,000 a month is consuming a meaningful share of margin to solve a problem that takes forty minutes a week. Buy the platform, buy payments, buy shipping aggregation, and stop.
Above roughly ₹5 crore a month the pressure inverts. Manual reconciliation stops being forty minutes and becomes a role. Channel count rises, each channel wants different product data, and the cost of a stock error is now a cancelled order on a marketplace with a performance score attached. This is where a proper product data layer and real order orchestration earn their keep, and where the integration layer stops being optional because the number of system pairs that must agree has grown faster than the headcount watching them.
The ₹1 crore a month band is awkward precisely because it sits between those two regimes. The volume is high enough that manual work hurts and low enough that most tooling is overpriced for what it removes. That is why buying order matters more here than at any other stage: you are choosing which single pain to spend money on while leaving the rest manual on purpose.
A test before any purchase
Before you sign anything, answer three questions in writing. Which decision does this remove from a person? Which system becomes authoritative for the fields it touches, and which system loses that authority? Who owns it in twelve months, by name?
If the first answer is vague, the tool is a preference rather than a need. If the second is unanswered, you have just added a writer to a record that already had one. If the third is "we will figure it out", published experience says it quietly degrades - which is also the core of the build versus buy decision covered in build vs buy an AI agent.
If you want this mapped against your actual operation - which layer is genuinely missing, what it would cost at your volume, and what to postpone - book an operations review. We come back with the sequence, the realistic budget across all three buckets, and the purchases we would tell you to skip this year.
