Insights on e-commerce, web development and digital solutions
Here you find our articles about online shops, websites and web tech. We explain new rules, tools and good ways of working. You can filter the articles by topic. At the top of an article you find a short version in plain words. Ask us if something is still not clear.
A shop chatbot is only as current as its sources. How to maintain delivery times, return rules and promotions with validity dates, cite the source in every answer and prepare for the holiday season.
Which data an AI feature carries out of the shop, which operating models exist and when a model on your own hardware starts to pay off.
Ten thousand product texts in four languages overnight – technically possible, risky in business terms. This guide shows how to classify content by risk, steer the machine stage with a glossary and style rules, measure translation quality through error classes and serve the language versions cleanly – from hreflang to the update run.
Returns eat the margin in 2026. An AI return prediction estimates the return probability per cart from size, fit and behavioral signals and intervenes before checkout – lowering return rates based on data without choking revenue.
36% of shops hamper their users in lists and filters (Baymard). A dialogue-based AI product advisor leads customers to the right product in a few questions – built as a custom solution for assortments that need explaining.
Text and data mining is permitted unless a machine-readable reservation exists. How shops declare the TDM opt-out in three places – without accidentally blocking retrieval crawlers too.
A third of AI-using companies report higher costs than expected. That is rarely down to the price per token, but to context length, output length and call frequency. Seven levers and a worked example.
As soon as a shop gets an AI advisor, a new attack surface opens up: language models do not separate instruction from data. What that means in practice and which layers hold.
In 2026 WebGPU is enabled by default in Chrome and Edge, and in Safari and Firefox on some platforms. Small language models now run directly in the browser – for private shop assistants, semantic search and text snippets, offline and without API costs. We sort out the chances, limits and hybrid approaches.
Larger targets, terms explained, input help, clear focus, no time limits
You decide which optional data processing to allow.
Our anonymous reach measurement runs without cookies and without IP storage, including campaign parameters in links – we do not ask for consent for it because none is legally required.
Strictly required: language preference, core site functions and storing your cookie choice. Without these the site does not work. Settings you choose yourself (light/dark appearance, accessibility) are kept in your browser's local storage only once you choose them; "Reset" in the accessibility settings removes them.
Records your interactions (clicks, scrolling, inputs) pseudonymously so we can spot usability issues. Form field inputs are masked automatically before storage; your IP address is not saved with the recording. Served from our own domain, no external tool. Opt-in, revocable any time.