Ecommerce operations: the four processes that decide your costs after launch

Running an online store does not end once the platform is live and payments are connected — that is the moment a second, longer phase begins: the daily work of handling orders, product data, your warehouse, and customer contact. Each of these four processes can be done by hand, automated, or outsourced to a specialist, and the choice between them is not a matter of taste — it comes down to numbers, order volume, team size, and how much time a given step actually takes. This article maps those ecommerce operations and points to where, in the rest of this section, you will find concrete answers.

Ecommerce operations — the four processes that start after launch

Whatever the industry or platform, after the first sale a store starts handling the same sequence of events: a customer places an order and pays, the order has to be picked in the warehouse and shipped, some orders come back as returns, and at every one of those moments the customer may need to reach your team. These are not four separate departments — they are one flow, where an error earlier on (missing stock data, say) shows up as a cost later (an order that cannot be fulfilled, a support ticket).

The first process is orders and product data — keeping prices, stock levels, and descriptions consistent across your store, your supplier, and any marketplaces you sell on. When this process drifts, the effect shows up two steps later: a customer buys something you do not have in stock, or sees a price at checkout that does not match what they saw before. The second is warehouse and shipping — physically picking, packing, and handing a parcel to a carrier, either yourself or through a specialist fulfillment provider. The third is customer contact — status questions, complaints, returns; a large share of these tickets is predictable before the customer ever sends them, if the first two processes behave consistently. The fourth, cutting across the other three, is measurement — knowing whether these processes actually work the way you assume they do, before you decide which one to fix.

Each of these four processes tends to break at a different point first. Orders and product data break on import — an overwritten mapping, the wrong character encoding, prices drifting out of sync after a supplier update. Warehouse and shipping break where volume outgrows resources — a team that coped with 50 orders a day does not cope with 500. Customer contact breaks when the first two processes do not feed it real-time data — support ends up saying "let me check and get back to you," because it genuinely does not know where the parcel is. Measurement breaks the most quietly: KPIs counted inconsistently (sometimes from sessions, sometimes from users; sometimes from revenue, sometimes from margin) create a false sense of control that is worse than having none.

What to measure, to know whether it is working

Before deciding what to automate or outsource, you need a handful of numbers that show how the store is actually doing today — not a gut-feel estimate, but numbers counted consistently, month after month, the same way each time. GMV (total order value) and AOV (average order value) describe the scale and shape of your sales; the number of transactions describes the operational volume you physically have to handle — and it is that volume that decides whether fulfillment or automation already make economic sense, or do not yet. Without these numbers, a decision to outsource the warehouse or buy an automation tool is a guess, however persuasive the salesperson on the other end sounds.

These numbers only mean something if you count them the way the report you are pulling them from counts them. A marketplace's GMV is an Alternative Performance Measure it defines itself, not an IFRS line item — it is not automatically the same calculation as the revenue figure your own store dashboard shows, and two marketplaces' GMV figures are not guaranteed to mean the same thing either. Google Analytics 4 no longer has a metric called "conversion rate" at all — only "session key event rate" and "user key event rate," one counted per session, the other per user. That name change is not cosmetic if you are comparing numbers from two sources and assuming they describe the same thing. Full definitions of these metrics, with the traps in calculating them, are in our article on ecommerce KPIs.

What to automate, and what to outsource

Not every process is worth handling the same way, and the choice is rarely just manual versus automated. Repeatable events governed by a clear rule — an order status change, sending an email or SMS, generating a shipping label, issuing an invoice — are a natural candidate for automation: you measure the return in hours of work saved, not in the promises of a sales pitch. Events that need human judgment, an unusual complaint, say, stay with a person regardless of how good the software is. How to choose what to automate first, and how to calculate that return, is in our article on ecommerce automation.

Integrations — connecting your store to ERP, WMS, and CRM — are a different kind of decision: not whether something should happen automatically, but which system is the source of truth for which kind of data (stock, price, document). Getting this wrong does not show up as a lack of automation — it shows up as data drifting apart between systems. We cover that architecture, and the order to roll it out in, in our article on ERP/WMS/CRM integration.

Warehouse and shipping is a process you can hand over to an external company entirely past a certain scale — fulfillment: receiving, storing, picking, and shipping your stock through someone else's warehouse, for fees charged per order or unit and for the storage space you use. It buys you time and space, but you need to run the numbers on your actual volume, not on a hunch — we cover that separately in our article on fulfillment. If you are still estimating what running the store itself actually costs, before adding fulfillment or automation on top, run the numbers in our ecommerce running-cost calculator — you will see which costs stay fixed regardless of what you outsource, and which ones grow with volume.

The obligations you cannot outsource

Whichever processes you automate or outsource, two obligations stay on your side, because they attach to the simple fact of running a store that takes payments and holds customer data: Swiss data-protection law — the Federal Act on Data Protection (FADP), which applies to every store processing customer data, with the EU's GDPR on top if you offer goods to customers in the EU — and PCI DSS (card-payment data security — every store accepting card payments), even when a third-party processor handles the card and not your own server; the integration method then decides the scope of what applies to you, not whether anything does. NIS2, the EU's cybersecurity directive, binds EU member states, not Switzerland; a business based outside the EU is drawn into it only if it offers certain digital services in the Union, such as running an online marketplace — which a store selling its own products does not do. Switzerland has its own duty instead: since 1 April 2025 the Information Security Act (ISG, Art. 74a–74f) requires cyberattacks to be reported to the Federal Office for Cybersecurity (BACS), but only by the authorities and sectors listed in Art. 74b — energy, banks and insurers, hospitals, telecoms, transport, cloud and data-centre providers, suppliers of essential everyday goods and others. An ordinary online store is not among them — unless it supplies essential everyday goods on a scale where an outage would cause serious shortages (Art. 74b(1)(p)). The full reasoning, together with the scope of Swiss data-protection law and PCI DSS for a small store, is in our article on PCI DSS and data protection.

Map of the operations section

The operations section splits into eight articles, each covering one of the processes described above:

• Ecommerce KPIs: GMV, AOV, LTV — the handful of numbers you need to run a store, counted the same way the reports you pull them from count them.
• Supplier feed integration — how to sync your catalogue and stock with a supplier, in a format that matches how often things change.
• ERP/WMS/CRM integration — which part of your data has its own source of truth, what order to roll systems out in, and how e-invoicing fits.
• Ecommerce automation — what to automate first, and how to calculate the return in hours of work, not in a vendor's promises.
• Fulfillment — when it pays to hand your warehouse and shipping to an external provider, and how to tell whether the numbers work for your volume.
• PCI DSS and data protection — card-payment security, Swiss data protection, and why neither NIS2 nor the Swiss cyberattack reporting duty usually applies to a single store.
• Ecommerce customer service — how to cut down "where is my order" tickets before the customer ever sends one.
• Omnichannel — when it is worth linking your store's stock to a physical location, and when that is unnecessary complexity.

Where to start

If you are only now putting operations in order after launch, start with measurement, not automation — without reliable GMV, AOV, and transaction-count numbers, you do not know whether the process you want to automate or outsource even has the scale to justify it. Next, work out which data has a single source of truth (prices, stock, documents) — without that, every automation and every integration will just be putting out fires it started itself. Only then calculate what is actually worth automating, and what is worth outsourcing.

In practice, a sensible order looks like this:

• Calculate GMV, AOV, and transaction count for the last 3 months — that is the basis for every decision after this one.
• Write down which system is actually the source of truth today for prices, stock, and documents — not which one should be, but which one genuinely is.
• List the events that repeat daily and follow a clear rule — that is your shortlist of automation candidates.
• Check whether order volume and seasonality already justify a conversation about fulfillment, or not yet.
• Check that your privacy policy, marketing consent, and payment setup comply with Swiss data-protection law and PCI DSS — the one item on this list you cannot put off.

Each of these steps leads to one of the operations-section articles, where you will find a concrete way to calculate or implement it.