'E-Commerce Process Optimization: The Method From 12 Years in Consulting'

Process optimization in e-commerce means finding your recurring workflows, measuring them, prioritising them, and then improving or automating them. In that order. Reverse the order and you optimise the wrong thing.

I was at PwC from 2006 to 2018. From junior consultant to senior manager, running efficiency projects for DAX corporations and mid-sized companies. Today I build e-commerce brands. The consulting method works just as well in a shop doing 100 orders a day. You just have to strip it down hard.

Here is the method in four steps. After that: what from the corporate world does NOT belong in e-commerce.

Step 1: Find your processes

Sounds trivial. It is the step almost everyone skips. Most founders optimise whatever process is annoying them loudest that week, while the expensive one keeps running untouched.

Here is how to do it systematically. Take an hour and answer three questions:

The result is a list of 15 to 30 processes. Ours started with items like "enter tracking numbers in Shopify" and "answer WISMO emails". WISMO stands for "where is my order", and that one question makes up around 40 percent of all support requests. For an overview of which processes are candidates, see our guide to e-commerce processes.

Step 2: Measure before you optimise

The most important sentence from my consulting years: you cannot improve what you do not measure.

Most founders "know" a process takes too long. They do not know how long. Without a baseline there is no improvement, only the feeling of improvement.

In consulting we spent weeks on process mapping for this. In e-commerce a stopwatch does the job:

An example from our own brand mate: order processing cost 4 hours a day, measured with a stopwatch rather than estimated from memory. After automation: 15 minutes. Without the first number we could never have believed the second one.

Second example, our brand MUSTAX: monthly reporting took 2 days. Today it takes 2 hours. Before we measured it, that was also just a vague "reporting is a pain".

Step 3: Prioritise with Pareto and the bottleneck

Now you have a list with hours attached. Two rules from consulting that always hold:

Pareto: 20 percent of your processes eat 80 percent of the time. Your top 3 time sinks are almost always order processing, support and reporting. Start there and ignore the rest for now.

Bottleneck first: every system has a bottleneck. Optimising anything else does little while the bottleneck stays in place. Ours was order processing for a long time. If we had optimised marketing first, we would only have pushed more orders into the same traffic jam.

On top of that, run one calculation per process: what does the improvement cost, what does it save? If an automation takes 40 hours to build and saves 2 hours a week, break-even sits at 20 weeks. That is fine. At 80 weeks it is not. The full calculation is in our article on the cost of automation.

Step 4: Automate, then iterate

Only now does the technology come in, never earlier. An automated bad process is a fast bad process.

The order per process:

1. Write the process down. Trigger, steps, decisions, outcome. One Google Doc, not 50 pages.

2. Automate the rule-based parts. Everything that is "if X, then Y": workflows in n8n or a similar tool.

3. Give the judgement parts to AI, with limits. At our brands AI answers around 65 percent of support requests automatically. The rest goes to humans. How that is built is in our customer service guide. For when you need a workflow, a chatbot or an AI agent, see our comparison of the three approaches.

4. Ship the 80 percent solution. The first version is never perfect. Use it, watch it, improve it.

The result across our three brands: 33 to 46 hours saved per week. With a team of 2 people running day-to-day operations.

What does NOT work from the corporate world

This is part of the method too. Four things I had to actively unlearn from my PwC years:

Perfect planning before execution. In consulting we planned for months before anyone lifted a finger. In e-commerce: do it, then improve it. The perfect automation on the first attempt does not exist.

Optimization for its own sake. I have seen projects that cost 2 million euros to save 500,000. The maths worked out over the long run. Nobody measured the opportunity cost. That is why step 3 exists: do the maths, then build.

Copying best practices blindly. What works for a DAX corporation crushes a 5-person team. Every template has to be adapted to your context.

Consensus above all. In a corporation every decision goes through five meetings. As a founder: decide, do, adjust.

The limits of this method

Process optimization has a blind spot. It makes existing things better. It does not tell you whether the existing thing is right.

If nobody wants your product, the fastest fulfilment process in the world will not help. And some processes should not be optimised at all: difficult customer conversations, product decisions, anything with legal weight. There, being slow is a feature.

There is also the up-front cost. The measuring week feels unproductive. It is the most productive week of the quarter, but you only see that afterwards.

And the numbers do not stay stable. A process that costs 2 hours a week today costs double at double the order volume. Repeat the measurement every few months, or you end up optimising against an outdated map.

Your start this week

1. One hour: write your process list

2. One week: measure the top 5 with a stopwatch

3. One afternoon: do the maths and prioritise the top 3

4. Then: automate your biggest time sink

If you want support on step four: get in touch. We are happy to walk you through what our setup at nano, mate and MUSTAX actually looks like.