K5 Theses on Retail

Stefan Wenzel's annual K5 theses on Agentic Commerce, AI agents, platform economy and the future of retail.

K5 2026

01. If you want to play, you structure data. If you want to be part of the decision, you build a new muscle group: Machine Equity.

One billion ChatGPT users per week, 16 billion Google searches per day, 20–50% of them inside the AI Overview — and for "best of" queries even 80%. On average, 6 of 60 billion listings make it through. In that world, structuring data cleanly is no longer enough. To stay relevant, you have to become part of the machine's decision. The machine counterpart to brand equity is Machine Equity — the sum of signals by which an AI system understands, trusts and selects a provider over its alternatives.

K5 2026 — three theses on the future of commerce in the era of AI agents, answer engines and demand-led commerce.

02. Sales-led commerce is over. The real disruption of retail comes in the slipstream of AI: demand-led commerce.

Contribution margin I is sliding, EBIT halves, CAC is up 70%, retention stalls at 30% — and the cost of capital is climbing. Fashion sits 168 days in stock before it sells, with 30–40% markdowns, 50–60% returns and 20–30% overhangs. 70–140 billion USD of deadstock per year. The error is structural: sales-led commerce with its "first-party fata morgana" is no longer sensible and no longer financeable — channel irrelevant. The next Amazon is an AI-first hybrid of Zara/Shein, TikTok and Uber: demand-led, asset-light, with its own discovery layers and 30% of the legacy cost base.

03. 10% of your people build 100% of your future, not technology. Find the 10%.

88% of companies use AI in at least one business function. But only 6% see real impact and financial returns above 5% EBIT. 20% of companies generate 74% of the total AI value. We do not have a tool problem. The limit sits in front of the screen. A digital company ≠ a digital-native company — it is data, systems, processes, meetings, decision flows. AI is a multiplier, not an equalizer. To leave the pilot purgatory, identify the 10% of people who actually move things forward — and give them room, mandate and speed.

Video of the K5 2026 theses on this page

Machine Equity · Dependency-Volume Continuum · K5 2026 keynote: Bot Betta Have My Money?

K5 2025

01. Our shops are becoming the vinyl record of customer access.

02. The white-elephant gift game is over: quality, relevance and uniqueness win – with humans and with AI.

03. Fixed costs will drop by 50% over the next 18 months – because they can or because they must.

Video of the K5 2025 theses on this page

K5 2024

01. Relevance and visibility. The rest can go.

02. Effectiveness as competitive advantage: effectiveness is the new efficiency.

03. Whoever loses access to the customer, loses.

Video of the K5 2024 theses on this page

K5 2023

01. Instead of discounts: manage prices and raise willingness to pay.

02. Individualization where willingness to pay is influenced. Customisation > Customersation.

03. The lower the share of 1P data in revenue, the less efficient the format.

Video of the K5 2023 theses on this page

K5 2022

01. Degree of variabilisation is Darwin's best friend.

02. Not a ball pit – the "Corporate OS" is mission-critical.

03. Brand/IP and operational excellence is everything. Period.

04. Tech is commoditised – what matters are value-creating applications.

05. In the slipstream of "pile high, sell cheap" there is plenty of clean air.

Video of the K5 2022 theses on this page

Book Agentic Commerce · AI companion to the book

STEFAN WENZEL