The artist Lars writes to his fans – and everyone feels: this email can only be for me. Built with data, AI and a great deal of prompt engineering. The result was dramatic.
Lars is a young German artist. He sells originals and art prints – and recently expanded his portfolio with sketches. On Instagram and TikTok he has almost a million followers combined.
Then his TikTok account was temporarily blocked. Overnight, a large part of his reach: gone. Together with Meister Lampe und Freunde he therefore built something that truly belongs to him – his own email list.
This list has only around 5,000 addresses. Even so, Lars sells more works through it than through all his social channels combined. The conversion rate by email is roughly 150× higher than on social – and that with a very young audience.
The basis for outstandingly successful email marketing is relevance.
The more relevant an email, the more successful. Prerequisite one is the right data from CRM, shop and ERP. For Lars it came from Shopify, Klaviyo, Pipedrive – and from a survey of his audience.
Prerequisite two is fitting content. This is exactly where 1:1 marketing has failed so far: not on the data, but on the effort. The smaller the segment, the bigger the work. → This is where AI changes everything.
For Lars, the AI generated a completely individual subject line, its own pre-header and its own email copy for each recipient – written like a short, personal letter from the artist. Based on purchase, click, demographic and emotional data.
The effect: recipients felt genuinely seen. "He must have written this email himself." That's exactly the lever.
Four types of data came together – from exactly the tools already sitting in modern e-commerce stacks.
Purchase history (Shopify)
Opens & clicks (Klaviyo)
Gender, age, language (Pipedrive)
Survey answers (Typeform)
Lars starts with Klaviyo: systematically capturing clicks and opens.
Development of the "Meister Lampe Hyper Personalization Engine" begins.
A four-digit number of recipients surveyed on how they experience art.Stage 1
Ongoing project: tests and extensive prompt engineering.
Campaign launch (EMEA).Stage 2
Originals, prints and new sketches – a highly emotional product. That's exactly what made the emotional data so valuable.
With ~5,000 addresses the list is small. That makes the statistical results less precise than with large lists – yet given the enormous increase we still consider them highly significant.
There's a bias: more data was available for more engaged recipients, which favours conversion anyway. So part of the uplift goes back to this higher engagement from the start, not to the hyper-personalised copy alone. The effect is real but can't be quantified exactly.
Even so: hyper-personalisation led to a roughly 10× click rate (20.3% vs. 1.95%) – largely attributable to the use of AI. And sales happened exclusively in Segment 1.
Perfect Email Marketing for the AI Era · 2nd edition
How AI hyper-personalisation works in detail – and how to use it for your own email marketing – you'll read step by step in the book.