martijns_gordijnen_paleis ยท ai search ยท retail
How a fake curtain shop outranked IKEA and HEMA inside ChatGPT
Martijn's Gordijnen Paleis is a curtain shop that never existed. No store, no stock, no curtains. We built the brand from scratch with linkbuilding and brand mentions, and watched it climb above IKEA and HEMA in ChatGPT's answers. The experiment got picked up by De Telegraaf, Emerce and HCC AI, and started a national conversation about how AI search actually works.
// CASE STUDY
OUTRANK IKEA
INSIDE CHATGPT
AI SEARCH EXPERIMENT
Martijn van der Pas (founder @ Backlink) on why AI search rewards brand presence, not keyword stuffing.
Executive summary
- Ranked a fictional brand above IKEA and HEMA inside ChatGPT
- Built the entire presence with linkbuilding and brand mentions, no real store, no real stock
- Picked up by De Telegraaf, Emerce and HCC AI
- Hundreds of shares on LinkedIn, thousands of views from marketers, developers and SEO specialists
- Proved that AI search results can be earned the same way trust gets built on the open web
// backed by 500+ active brands
// team photo
// about the case
Martijn's Gordijnen Paleis is a curtain shop that never existed
We invented a Dutch curtains brand. No physical store, no warehouse, no products. Just a name, a story, and a small set of mentions across the open web. The whole point was to test a question we kept hearing from clients: can you really influence what ChatGPT recommends?
The answer turned out to be yes, and faster than anyone expected. Within weeks, asking ChatGPT for a Dutch curtain shop returned Martijn's Gordijnen Paleis above household names like IKEA and HEMA. The experiment got national coverage and sparked a real conversation about how easy it is to shape AI answers.
// HOW WE DID IT
See the whole process and the results
// ChatGPT response screenshot
AI search rewards mentions, not keyword density
Every chatbot answer gets generated on the fly from the sources the model has indexed. There is no fixed ranking page like in Google. The model pulls from what it has seen, weighs it, and writes a recommendation.
That means brand presence across trusted sources is the new ranking signal. Show up in the right places often enough and the model starts pulling from you. We built that presence for a brand that did not exist, and the model still cited it.
Linkbuilding and brand mentions, nothing else
We placed mentions of Martijn's Gordijnen Paleis across the open web. Editorial mentions, list-style posts, contextual references. The exact same playbook we run for real clients, just pointed at a fake brand.
No technical SEO. No on-page tricks. No website hacks. The brand existed only as a name across other people's content, and that was enough for the model to start recommending it. The case proves the strategy. The difference with real clients is simple: their brands do exist, and their products are worth recommending.
// mention timeline
Backlink vs legacy agencies in a nutshell
Legacy agencies still sell links for Google's ten blue links. We build presence for the engines people actually ask questions to.