Automation Strategy: How to Build E-Commerce With a System

Most shops automate the way they shop: spontaneously. A tool here, a workflow there, depending on what is annoying right now. That saves hours. It changes nothing fundamental.

An automation strategy is something else. It answers three questions before any tool gets bought: in what order do we automate? What do we build ourselves, what do we buy? And how do we measure whether it works?

With this system we run 3 brands with 2 people in day-to-day operations. Across all brands we save 33 to 46 hours a week. Here is the strategy behind it.

The principle: automation first

At Flowhouse the sentence "we need someone for X" is banned. Instead we ask: can a machine do this?

The reason is the cost structure. Traditionally: double the revenue, roughly double the team, double the cost. With automation: double the revenue, the same system cost, slightly more server.

The principle is simple. It becomes a strategy through order, build-vs-buy rules and metrics.

The order: where you start

Automations are not all worth the same. The right order follows two criteria: how many hours does the process eat, and how rule-based is it? High and high means: first.

Stage 1: order processing. The most rule-based process in the entire shop. Order comes in, inform the fulfilment provider, push tracking back, notify the customer. At our brand mate this used to cost 4 hours a day. Today: 15 minutes. The setup is described in detail in the article on automated order processing.

Stage 2: support triage and standard enquiries. Around 40 percent of all support enquiries are WISMO, meaning "where is my order". Fully automatable, because the answer sits in your systems. In total, AI answers around 65 percent of our enquiries automatically. The path there is in the guide to customer service automation.

Stage 3: reporting and bookkeeping. Recurring, rule-based, error-prone by hand. At MUSTAX we took reporting from 2 days to 2 hours. For the finance side: automating bookkeeping.

Stage 4: marketing and content. Only now. Marketing matters, and here AI works alongside people instead of running on its own. More on that in the article about marketing automation.

Why this order? The early stages finance the later ones. The hours from stage 1 and 2 are the time you need for stage 3 and 4.

A note on the temptation to start with stage 4: marketing automation is the most visible one and feels the most strategic. But it fills a funnel whose far end still works by hand. Only once orders and support run automatically can your shop take more demand without quality slipping.

Build vs. buy: the decision rule

The second strategic question: build it yourself or buy a tool? Our rule in five test questions:

1. Does a good finished solution exist? Real research, not 5 minutes of googling.

2. What does it cost at your scale in 12 months? Not today.

3. What is it missing? Must-haves or nice-to-haves?

4. How long does your own solution take?

5. Who maintains your own solution long term?

Buy when the problem is standard: shop system, email, accounting software. You gain nothing by building those.

Build when the use case is specific and the SaaS costs grow with you. Our self-hosted n8n server costs a fixed amount per month, whether a hundred or ten thousand workflows run on it. Comparable SaaS tools for three brands would cost a multiple of that, and more every month.

The flip side: self-built solutions need someone who understands and maintains them. Without technical know-how on the team, "build" is a risk rather than a strategy. In that case, buy and accept the limits of the tools.

The metrics: how you know it works

Without metrics, every strategy is a feeling. We measure four numbers:

Important: measure before you build. The baseline is half the measurement.

The limits: what we deliberately do not automate

Automation first does not mean automation only. Four areas always stay with people at our end:

The reason: these things need context, judgement and empathy. AI can assist, it cannot decide.

And one more limit of the strategy itself: it has upfront cost. In the first weeks you build systems instead of making revenue. Anyone in an acute cashflow crisis should sell first and automate afterwards. An automation strategy is an investment. It will not pull you out of a hole this month.

Second limit: automated systems need care. APIs change, carriers switch interfaces, models get deprecated. Budget a few hours of maintenance per month. Ignore that and you will one day wake up to a stopped conveyor belt and notice it through an angry customer.

The difference from "just buying tools"

To close, the core of why a strategy is more than a toolbox:

If you want to know where your shop stands on this staircase: get in touch. We look at your processes together and tell you what pays off and what does not.