E-commerce

Industry Overview

E-commerce runs on three things: products, orders, and people. Everything else is a variation of those.

A typical online retailer has a catalogue. Products have variants: size, colour, pack size, region. Each variant is its own stock line, even if it looks like one product to a customer. Digital content — photos, descriptions, pricing, weight — has to stay in sync everywhere the product appears.

Sales come from more than one place. The owned store is one channel. Marketplaces like Amazon, Etsy, or Flipkart are separate channels with their own listings and rules. Social shops and buy-now links add more. Every channel sends orders in its own format, with its own fees and delivery expectations.

Behind the scenes, teams keep the machine running. Merchandising manages the catalogue and pricing. Operations runs fulfillment: picking, packing, dispatch. Customer service answers order questions, handles returns, and resolves delivery issues. Finance tracks payments, fees, taxes, and refunds. In a small company, the same few people do all of this.

Fulfillment takes many shapes. A business might pick and pack from its own warehouse, ship from a supplier who dropships, or hand everything to a 3PL (third-party logistics) company. Shipping may go through multiple couriers. Every parcel needs a label, a tracking number, and a delivery event. In markets like India, return-to-origin (RTO) is a daily reality: packages that cannot be delivered come back, and the business absorbs the cost.

Returns are a business in themselves. Customers exchange sizes, change their minds, or receive damaged goods. Refunds hit the payment processor, and stock has to be inspected and put back. None of this is glamorous, but it is where money leaks out.

Money also moves in odd patterns. Customers pay at checkout, but marketplaces hold funds for days or weeks. Fees vary by channel. Promotions change margins. Some customers pay on delivery. Reconciling what was sold, what was paid, and what was refunded is a regular weekly chore.

Seasons dominate the calendar. Peak days — Black Friday, end-of-season sales, festival sales — can bring five to ten times normal volume. Staff, stock, and courier capacity all strain at once. Then the quiet months follow, and the business has to plan for both.

The people who run e-commerce spend their days juggling spreadsheets, inboxes, and channel dashboards. The work is constant, and the margin for error is thin.

Repeat customers matter more than they appear to. A returning customer costs less to acquire and often buys more. So the operation has to keep not just orders straight but people straight: who bought what, when, and whether they came back. Loyalty programmes and order history both depend on that record.

Common Operational Challenges

Inventory numbers that disagree

Each channel keeps its own stock count. Amazon says 12 units. The store says 9. The warehouse actually has 11. These numbers drift apart because sales, returns, and adjustments land on one channel but not the others. When nobody can trust the numbers, someone overrides them with a manual correction. That correction becomes the next wrong number.

Overselling follows. A product sells on three channels faster than the warehouse can pick, and the business ends up selling stock it does not have. The customer learns about it after payment, when they get a cancellation or a long delay. Every oversell costs trust and, on marketplaces, a penalty.

Under-selling is the quiet twin. Stock sits on a channel that is not moving while a channel that is selling runs out. That is lost revenue with no error message to show for it.

Orders scattered across the inbox

Orders arrive by web, marketplace, and phone. Many small teams manage them from notification emails and sales dashboards. To answer one customer question — "where is my order?" — a support person opens the marketplace dashboard, checks a courier site, and reads three emails. The answer takes minutes to assemble.

The same order gets shipped twice. Or cancelled after it shipped. Or refunded and then shipped anyway. These errors happen because the order's state lives in several places and nobody owns the whole picture.

Promotion mistakes are expensive

A discount applies to the wrong variant. A coupon code leaks beyond the intended audience. A "sitewide 20% off" actually includes items that were already at cost. On a marketplace, a promotion mistake can sit live for hours before anyone notices, taking hundreds of orders at the wrong price.

Fixing it is hard. Orders taken at the wrong price have to be honoured, cancelled, or re-priced. Each option annoys someone. Marketplaces can also claw back the difference or penalise the seller.

Fulfillment and shipping friction

Pick lists come from a channel report, not from reality. A picker walks a warehouse with a printed list while stock moves underneath them. Wrong items get picked, right items get missed.

Couriers need labels in their format. Some regions have several couriers, each with its own file layout, pickup times, and delivery updates. Preparing parcels for a peak day means producing thousands of labels under time pressure, and any mistake holds up the whole dispatch.

Returns, refunds, and RTO

Returns arrive with no paperwork. The customer says "size too big," the parcel says nothing. A refund goes out before the item comes back, and the item never comes back. In some markets RTO parcels accumulate at depots, costing money twice: once to ship out, once to bring back.

Inspecting returned stock takes time. A returned item needs checking, cleaning, and re-listing — or writing off. Nobody tracks this well, so the true cost of returns stays invisible.

Support teams chase orders

The inbox fills with the same question in different forms. Where is my order? When will it arrive? Why was I charged twice? Each answer requires hunting through systems. There is no single record a support person can trust.

As the catalogue grows and promotions change, the questions get more specific and the answers get slower.

Profit is unclear

Fees, shipping, returns, and refunds nibble at every order. The selling price on the store is not what hits the bank. A product can look profitable at list price and lose money after marketplace commission, payment fees, and return rates.

Teams discover this too late, usually after a discount runs. Pricing decisions get made without the real cost per order.

Tax and invoicing across channels

Each channel handles tax differently, and invoicing means different things on different platforms. Sales tax or VAT, export rules, and marketplace deductions all have to be recorded and reported. Getting the numbers right at tax time means keeping per-order tax data from the start. Teams without it spend days reconstructing what they owe and where.

Peak seasons strain everything

Peak days multiply every problem above. Stock numbers drift faster. Support queues grow. Couriers slow down. The business that runs on manual processes in the quiet months cannot simply work harder in the busy ones.

How Software Helps

Start with the record. Software works by making one event the source of truth and reusing it everywhere. A stock adjustment, a sale, a scan — each happens once, and every system that needs to know learns about it. This mechanism sits behind most of the improvements below.

Inventory sync is the obvious win. When a sale happens on any channel, the stock number moves everywhere at once. Channels see the same number, or close to it. Overselling drops because the count is honest. Underselling drops because stock can be moved to the channel that sells.

The order view is the second win. Every order, from every channel, lands in one list with the same fields: items, customer, payment, status, tracking. A support person answers "where is my order?" from one screen. The order's whole history — paid, picked, shipped, delivered, returned — reads like a timeline. Mistakes like double shipping become visible because the state is one record, not five.

Fulfillment gets automation. When an order is confirmed, the system turns it into a pick list in warehouse order. A scan while picking confirms the right item and updates stock. Labels generate in the courier's format. Tracking numbers flow back to the order and out to the customer automatically. The scan becomes the record, and nobody retypes anything.

Shipping rules remove decisions. The system picks the courier by zone, weight, and cost, and prints the label without a human choosing. Batch label generation handles peak volume without midnight spreadsheets.

Returns become a workflow instead of an argument. A return request starts a record. The refund waits for the item. Inspection results update stock — back to sellable, or written off. RTO parcels get matched to their original order, so the double cost becomes visible. None of this is magic; it is a consistent record where there used to be none.

Pricing and promotions get rules. A promotion applies to the intended scope only: this catalogue, these variants, this customer group, this window. The system checks the price against cost before it goes live and flags anything that sells below margin. The "did we just give it away?" panic stops happening on the storefront.

Reporting replaces guesswork. Because every order, fee, refund, and return is one record, the numbers add up. True profit per product after fees and returns is a query, not a month of spreadsheet work. Decisions about pricing and stock rest on numbers the team already has.

Customer service gets faster for a simple reason: the answers are in one place. A support agent sees the order, the courier, and the payment history on the same screen. What used to take three minutes of digging takes thirty seconds of looking. That does not just save time; it means the customer gets a straight answer instead of a "we will check and get back to you."

Purchasing improves too, quietly. Once stock is one record, the system can watch sell-through and flag items to reorder before they run out. It can even rank which suppliers ship on time. Most sellers start reordering from a hunch; a small amount of history turns that into a habit based on numbers.

Now the honest part. Software fixes processes, not products. If the goods are bad, the copy is weak, or the courier is slow, no system repairs that. A good tool makes a bad operation faster at being bad. Start with a clear process, then automate it.

Software also cannot fix pricing judgement. A rule can block a below-cost promotion, but deciding what margin a product needs is a business decision. A tool only makes the decision visible.

And software adds nothing when the operation is one person and twenty orders a week. The overhead of maintaining integrations may cost more than the mistakes they prevent. Tools earn their keep at volume, across channels, or with complicated products — not before.

There is also a list of problems software cannot touch, and it is worth naming. A product with weak photos sells slowly on every system. A marketplace that penalises poor service does not care how good your order tool is. A one-channel business with steady demand may fix more by improving its packing than by buying anything. The honest answer is that software only works on the parts of the operation that run on data. The parts that run on judgement, taste, and hustle stay human.

Common Business Systems

E-commerce platform

The owned store. Shopify, BigCommerce, and Magento are the common choices. The platform manages the catalogue, checkout, and payments for the direct channel, and it usually keeps the official record of what the store sold. Most businesses should start here — it covers the fundamentals and has thousands of apps for everything else.

Marketplace seller tools

Amazon Seller Central, Etsy, Flipkart, and eBay seller tools manage listings, orders, and performance on each marketplace. They are where fees, penalties, and performance scores live. These tools are not designed to talk to each other, which is why the order and inventory problem exists in the first place.

Order management system (OMS)

An OMS gathers orders from every channel into one place, controls fulfillment, and tracks status. It is the backbone for multi-channel sellers. Shopify Plus stores get basic OMS features, and dedicated tools like Linnworks or ShipStation layer on top. For a mid-size seller, this is often the first real purchase after the platform.

Inventory management system (IMS)

Tracks stock across warehouses and channels, handles purchase orders, and sets reorder points. Some are standalone; many are built into an ERP or an OMS. The value is simple: one number that everyone trusts, pushed to every channel.

Shipping and fulfillment software

Tools like ShipStation, Shippo, and EasyPost generate labels, compare courier rates, and send tracking updates. 3PLs run their own portals where orders are pushed and statuses pulled back. These tools matter because couriers never talk to each other directly, and labels are no place for manual typing.

Payment gateways and processors

Stripe, PayPal, Razorpay, and local alternatives handle checkout payments and refunds. They settle funds on their own schedules and take their own fees. Reconciling the payment provider's numbers against the order system is a recurring job, and it is a common source of mystery money.

Accounting and ERP

QuickBooks, Xero, or NetSuite hold the ledger: sales, fees, tax, refunds. Larger operations run an ERP like SAP Business One or NetSuite that also holds purchasing and costing. The link between orders and the ledger is where many small businesses lose hours every week.

Customer service helpdesk

Zendesk, Gorgias, or Freshdesk bring enquiries together and show order context beside the ticket. Gorgias is e-commerce specific and pulls order history into the conversation. A good helpdesk cuts the "chase the order" time dramatically, which is where most support frustration comes from.

Returns management

Dedicated returns tools such as Loop and Returnly run the return request, print labels, and trigger refunds. They are a newer category and optional for small sellers. The core job — refund only when the item returns — can also be done inside an OMS.

Email and marketing

Klaviyo and similar tools drive abandoned cart emails and campaigns. They are not operations software, but they plug into the same order data. Abandoned cart recovery is the cheapest revenue most sellers leave on the table, and the tool is cheap to run.

Warehouse and fulfilment tools (for bigger sellers)

When a business runs its own warehouse, warehouse management systems (WMS) control locations, pick paths, and stock counts. They tell a picker where each item sits and confirm each pick by scan. Most sellers grow into a WMS later; a warehouse with a few thousand SKUs can run on an OMS plus discipline. When the picking errors or the stock counts start costing money, that is the signal a WMS is due.

Channel-specific apps

Marketplaces and platforms maintain their own app stores. Amazon has its own reporting and repricing tools; Shopify has thousands of apps for shipping, reviews, and loyalty. These sit in a grey zone: some replace a custom build, some create the need for one. The practical rule is to use them while they fit, and treat them as replaceable pieces rather than the backbone.

What ties it together

No single vendor sells all of these. The platform, the OMS, the IMS, the couriers, and the helpdesk each come from a different supplier, and none of them share data by default. That is why integration exists as a category at all. A typical mid-size seller lives in the gaps between these systems, and the gaps are where custom work earns its keep.

Typical Workflow

1. A customer buys on a channel — the store, a marketplace, or a social link.
2. The order enters the channel's system with items, address, and payment.
3. The order appears in the business's order view, alongside orders from every other channel.
4. Stock is reserved and the order turns into a pick list.
5. A picker collects the items; a scan confirms each one and updates stock.
6. The parcel is packed, labelled, and handed to the courier.
7. A tracking number flows back to the order and to the customer.
8. Delivery updates follow until the parcel is signed for.
9. If something goes wrong — return, refund, RTO — the order record carries it.

Today, most teams run this with a mix of tools. The platform handles the store. The marketplace tools handle their own orders. Spreadsheets hold stock. The courier portal makes labels. The inbox holds customer questions. It works until it does not: peak day, a promotion, a product that sells fast.

Software's job is to compress the middle. When orders from every channel land in one view, steps 3 through 8 take minutes instead of hours. The record follows the parcel, and the parcel follows the record. A customer question becomes a lookup instead of a hunt.

The workflow does not get more complex with software. It gets fewer copies of the truth. That is the whole point.

Two honest caveats belong here. First, the workflow only works if the data entering it is honest — a stock count is useless if nobody counts the returns coming back in. Second, the biggest gains land on the steps that happen repeatedly. A business that ships forty orders a day gets more from step 6 automation than one that ships four. Volume decides where software pays.

Why Custom Software

Custom software earns its place when the shape of the business does not fit the standard tools.

Unusual catalogues are one case. Custom products, configurable products, or catalogues with complex variants do not fit a platform's defaults. The same is true for products sold by weight, by quote, or in bundles that change over time.

B2B and retail hybrids are another. Wholesale pricing, credit accounts, purchase orders, and sales teams do not work like consumer checkout. Mixing both in one system, with one inventory and one set of records, is where standard tools strain.

Complex pricing models matter too. Contract pricing, customer-specific rates, and discount structures beyond a rule or two push past what a platform or app handles. Marketplaces with unusual fee structures also reward a small custom calculation layer.

Deep integration needs arise when the e-commerce system must talk to a legacy ERP, a custom warehouse, or an internal system that has no standard connector. Sometimes the gap is narrow — a custom integration between two good tools beats a custom platform any day.

Then the honest half. Most businesses do not need custom software. A new store, a single channel, or a standard catalogue works fine on Shopify or BigCommerce with apps. Even multi-channel sellers usually start with an OMS and standard integrations, not a build.

Custom software is expensive and permanent. It needs maintenance, documentation, and ownership. It only pays off when it solves a problem the standard tools genuinely do not, at a scale where the cost makes sense. If a solid off-the-shelf stack covers the need, use it. Custom is a decision, not a milestone.

A useful test before any build: write down the exact task the standard tools cannot do, and the money that task loses each month. If the answer is clear and the number is real, custom is worth discussing. If the answer is vague, the problem is probably the process, not the software. Most conversations about custom e-commerce software start vague and end with a cheaper fix.