AI Foundations for E-Commerce

AI Foundations for E-Commerce The two-day seminar for everyone who wants to use AI in retail and not stop at the demo: you learn what AI can do today, which data you need for it and how to get from an idea to a pilot that proves something. By the end you know which use case is first for you.

AI Foundations for E-Commerce at a glance

• Format: 2 days on site + self-learning track
• Level: basic
• Participants: max. 16 people
• Language: German (English on request)
• Certificate: AEI certificate of attendance
• Investment: €149.00 plus VAT

A lot is promised about AI in e-commerce and little is measured. Personalisation, forecasts, chatbots, price optimisation: all of it sounds sensible, and almost every project starts with a technology instead of a problem. Six months later nobody can say whether it worked.

The cause is rarely the model. It is that nobody decided in advance which decision should improve, which data exists for it and how success would be measured. Without those three answers every AI project is an experiment with an unclear outcome.

This seminar works from the decision to the technology. You rank use cases by benefit and effort, check the data situation and write a pilot that shows something in weeks instead of nothing in quarters. What has to stay with the team is part of the plan.

What you take away

• An overview of use cases ordered by benefit and effort.
• A check of which data really exists for your case.
• A one-page pilot brief with goal, metric and stop criterion.
• An understanding of when AI errs and how to notice it.
• A split of which tasks must stay with people.

Your learning objectives

• You distinguish AI, automation and rule sets cleanly.
• You assess use cases by benefit, data situation and effort.
• You recognise whether your data is sufficient for a case.
• You formulate a pilot with a measurable expectation.
• You name limits, risks and responsibilities before a project starts.

The levers at a glance

First the decision, then the model. An AI system is only useful when it improves a decision — which products are shown, which order looks suspicious, which request gets which reply. Without that decision there is nothing to improve.

Data quality beats data volume. Two thousand well-maintained articles beat twenty thousand with empty attributes. Whoever starts with poor data measures the state of the data, not the ability of the model.

A pilot needs a stop criterion. A project that cannot fail fails later and more expensively. Whoever decides in advance when to stop can measure honestly.

What the team keeps decides the benefit. If nobody can check the result it stays a black box. Checkability is part of the system, not an add-on.

Shadow AI is already in your company. People are using chat tools for customer replies and product texts right now, often without a policy. A realistic plan starts by finding out what is already happening instead of pretending the decision is still open.

Who is this seminar for?

For people who have to introduce or assess AI in retail.

• E-commerce and shop managers with first AI ideas and an unclear benefit.
• Marketing and range owners planning personalisation.
• Project leads who have to set up and justify a pilot.
• Managing directors who want to know what is realistic.

Technical background is not required. The terms are explained during the seminar; the work is a management task with a data angle.

The seminar is not suitable for data science teams wanting to train models. We work on selecting, justifying and steering initiatives.

Investment and dates

Participation costs €149.00 plus VAT, including the self-learning track, materials and certificate of attendance. There are six dates per year in Berlin, Hamburg, Cologne, Munich and live online, plus an in-house variant for teams. You will find all dates in the list on this page.

You do not need system access for the exercises. Anyone who brings their own idea works on it; everyone else picks one of the provided cases.

After the seminar

You keep the materials, a list of use cases with an assessment and the brief for your pilot. With that you can prepare a decision next week instead of a presentation.

The list of use cases is deliberately short. Twelve fields with a score beat a market overview with sixty, because you can only act on what you can compare.

Two weeks after the seminar you receive a short follow-up: your pilot brief and the question whether the stop criterion is really measurable. That is exactly where most initiatives fail.

For teams the seminar is available as an in-house event with your own cases and data. The procedure stays the same, the examples come from your shop.

Programme: 2 days

Day 1 – What AI can do in retail

• Typical use cases
• Separating the terms cleanly
• Which data is needed
• Quality before quantity
• From use case to test
• What must stay with the team

Day 2 – First steps: from use case to pilot

• Using chatbots and assistants sensibly
• People and machines working together
• When AI errs: hallucinations and bias
• Transparency and responsibility

Both days run from 09:00 to 16:15 with breaks in the morning, at midday and in the afternoon. Short self-checks sit between the blocks so you can test where you stand before moving on.

Each day has three parts: an introduction to the pattern, an exercise on your own case and a review in the group. At the end of day one you pick your use case; on day two it becomes a pilot brief.

The seminar includes a self-learning track. It contains the patterns from both days as short units with examples and one self-check each. You can work through it before the date to clarify terms, or afterwards to repeat what stuck and what did not.

Our approach

We work on your case, not on a landscape map of AI. At the start of each exercise you decide which decision should improve and how you would measure that; at the end stands a brief you can stand behind.

The exercises are short and repeat in variations. That is deliberate: an eye for feasible initiatives does not come from listening but from repeatedly shrinking ideas until they are testable.

Every block ends with a sentence you could show a colleague: which decision, which data, and which number proves that it worked.

We work in pairs, in small groups and in plenary, with changing partners. Every participant gets feedback on their brief, not just on their idea. Group level and participant count are deliberately limited: with 16 people no anonymous lecture is possible.