E-shop automation: what pays off and what is just another add-on

E-shop automation is not a feature you switch on. It is a list of tasks somebody in your company currently does by hand, and a decision about which of them a system should take over.

That is why “how much does e-shop automation cost” has no answer until we know what exactly you do manually and how many times a day. This article is about writing that list yourself.

The rule that saves the most money

You can only automate a process you are able to describe step by step.

When five people give five different answers to “what happens after an order comes in”, there is nothing to automate yet. Chaos automates perfectly well; you just end up with chaos that runs faster and that nobody can stop any more.

The order is always the same: describe the process, simplify it, then automate it. That middle step usually deletes half of what you were about to pay someone to build.

What actually pays off

Sort your tasks by one number: minutes per day times days per year. Not by what sounds modern.

1. The order flow. Status changes, customer e-mails, invoices, carrier labels, the entry in accounting. This happens on every single order, so it almost always pays off and it goes first.

2. Stock and availability. Syncing with the warehouse, the ERP or the supplier. An unsold item is lost margin; a sold item that does not exist is a complaint, and that costs more.

3. Comparison-site feeds. Google Shopping and the local players. A hand-maintained feed is always a day behind your stock, and you pay the difference in cancelled orders.

4. Communication around the order. Confirmation, dispatch, delivery, and a review request a few days later. Reviews influence purchases more than advertising does, which makes that last one the most profitable automation of the four.

5. Returns and complaints. Not because they are frequent, but because they are the slowest and they anger people most when they stall.

6. Reporting. Only once somebody actually acts on it. A dashboard nobody opens is the most expensive automation of them all.

Deliberately not on the list: automated pricing and “AI that decides for you”. Those are extensions, and they make sense once the six items above run on their own.

When add-on modules become the problem

Most e-shops solve this with plugins. While you have two or three, that is the right and cheap answer. The trouble starts when this is true:

• you have five or more paid modules, each with its own monthly fee,
• two of them do the same thing, because the first one lacked one detail,
• none of them knows about the others and data moves by export and import,
• something breaks after every platform update and nobody knows what,
• “why was this order not booked” takes an hour of digging to answer.

At that point you have stopped paying for features and started paying for the fact that they do not talk to each other. That is the moment to calculate whether having those processes inside the platform is cheaper. When it is and when it is not is covered in a separate article on own versus rented platforms.

The new layer: a machine reads your shop too

Until now a product had to be understandable to a human and to Google. There is a third reader now: shopping assistants inside ChatGPT, Perplexity and Google’s AI answers. Buying through them is still a small share of the market, but it is growing faster than any other channel and nothing suggests that will reverse.

The difference from classic SEO is that an assistant has no time or appetite for your design. It takes facts. In practice that means:

• price and availability must be in the HTML, not painted in by JavaScript after load — most AI crawlers do not run scripts and will see nothing,
• Product and Offer structured data with price, currency, stock status and delivery time,
• a feed that matches stock in real time, not once a night,
• specifications as data, not as prose — “120 × 60 cm” in a table, not in the third paragraph of a marketing text,
• unambiguous identifiers (GTIN, manufacturer code) so an assistant can match you to the same product elsewhere.

None of this is new technology. It is tidy data, which you should have had anyway. It was simply possible to avoid it until now.

What automation will not fix

It will not save an e-shop with no demand. At ten orders a month the saved minutes are worth tens of euros a year and the project never pays for itself.

It will not replace decisions either. A system does what you told it, even when that is wrong, only ten thousand times and without asking.

And it will not cut headcount immediately. The first year usually frees up capacity that the company spends on things it never had time for. That is fine, but expect it and do not promise yourself payroll savings.

Where to start

For one week, write down what you do by hand in your e-shop and how long it takes. Measured, not estimated. At the end of the week you will have a list sorted by time, and the top row is your brief.

If it turns out the problem is not one task but the fact that your platform will not let you do it properly, see how we build e-shops.