It took 42 days to launch on the first marketplace. It took nine days to launch on the fourth.
It took 42 days to launch on the first marketplace. On the fourth, nine days.
A lighting manufacturer wanted to enter the Leroy Merlin market. We suggested a strategy that meant the next three retail chains cost him a fraction of what the first one did.
The challenge
The client treated each retail chain as a separate project: separate data, separate export, a separate person to manage it. With two channels, this worked. With six planned, the numbers no longer added up; the internal cost estimate came to half a full-time equivalent per channel.
No one on the client side knew that Leroy Merlin, Castorama, Media Expert and Obi.de all run on the same platform, Mirakl.
About the client
A manufacturer of decorative and technical lighting, with its own factory and an extensive catalogue offering a wide range of options.
He sold on Allegro and through his own shop; retail chains’ marketplaces were a new type of sales channel for him.
What we’ve done
The decisions that determined the outcome.
We treated the first roll-out as an investment in the next three
Leroy Merlin, Castorama, Obi.de and Media Expert operate on the Mirakl platform. Their product file structure, variant logic, order processing and quality requirements are very similar. We therefore carried out the initial implementation not for a single retail chain, but to a standard, which extended the timeframe from an estimated 25 days to 42.
That difference – 17 days that seemed to have been wasted – reappeared during the very next roll-out and kept reappearing with every subsequent one.
We have created a single source file instead of four exports
Instead of creating a separate file for each network, we have created a single database containing a full set of attributes and mappings to the categories for each platform. Today, a new network means a new mapping, not a new catalogue.
We linked the statuses and prices via integration so that the four channels wouldn’t drift apart manually. With 1,260 indices, this wasn’t a matter of convenience, but a prerequisite for managing the system at all.
We stopped after the fourth network
The plan had envisaged six. After analysing the results from the first four, we recommended putting the project on hold: the remaining two were in categories where the client’s product range had no competitive advantage, either in terms of price or product. The cost of handling them would have exceeded the projected profit.
Course
Key stages.
- March 2026 A standard approach rather than a one-off integration Development of an attribute database to meet Mirakl’s requirements, not specific to any single network. Category mapping, variant logic, and images in the required formats.
- April, May 2026 Castorama, 42 days The most demanding customer first. Three rounds of quality improvements on the platform’s side.
- June 2026 Leroy Merlin, 14 days The catalogue was approved without any amendments. Time was spent on mapping categories and sorting out the logistics.
- July 2026 Obi.de, 11 days The first overseas market in this series. The translations have been added; everything else remains the same.
- August, September 2026 Media Expert, 9 days A small selection from the catalogue: 310 entries from the home and technical lighting categories. Launching after the summer break in August.
- October 2026 Decision to detain Analysis of margins by channel. Recommendation to suspend the two remaining networks from the plan.
Results
The timeframe for rolling out further networks.
The decline is not down to the team’s expertise, but to the fact that the catalogue had already been designed to meet the requirements of the entire family of platforms.
Labour cost per channel.
| Credits | Customer respect | Actual | Change |
|---|---|---|---|
| The first network | 25 days | 42 days | +17 days |
| Each subsequent | 25 days | 9–14 days | −44 to −64% |
| Four networks in total | 100 days | 76 days | −24% |
| Monthly service, 4 channels | 2.0 full-time equivalents | 0.6 full-time equivalent | −70% |
| Restructuring the catalogue for the next network | put on every time | not even once | , |
How we did it.
The entire value of this project was created in the first three months, before any sales had been made. We built the catalogue to the Mirakl standard and started with the platform that had the most stringent requirements, which meant that the next three were implemented without the need for any reworking. In the end, we decided against two of the networks in the plan, as technical feasibility is not the same as business sense.
Repeatability
Why do we think it wasn’t a coincidence?.
The order in which you enter has its price
A catalogue designed for the most demanding audience goes down a treat with less demanding audiences without needing any changes. The reverse scenario means as many redesigns as there are platforms.
Platform families, not individual websites
A dozen or so European marketplaces run on shared platforms. Anyone who realises this at the time of the initial implementation pays for integration just once, rather than four times.
Technical feasibility is not a reason to proceed
A channel that goes live in nine days still incurs a monthly cost. The decision is based on the margin in that category, not the time taken to set it up.
What this case does not prove: Not every platform in Europe belongs to such a group. Amazon, Allegro and eMAG have their own systems and their own requirements, so these savings will not apply there.
Are you planning to use more than one platform?
We’ll check which of them share the same engine and requirements, and we’ll organise the implementation order so that the first one does the work for the others. You’ll receive a plan setting out the time and cost per channel.
Plan the order of roll-outs
Polish 
English 
German 
Czech 
Hungarian 
Slovak 