Sensefuel - The Blog

Making your marketplace a success by delivering on its promises (episode 6 of 8)

Written by Stéphane Vendramini | April 01, 2025

In an increasingly competitive e-commerce landscape, marketplaces must navigate multiple challenges to ensure their profitability and sustainability. From optimizing margins to gaining deep insights into customer behavior and strategically leveraging data, every decision impacts the platform's attractiveness and economic balance.

In this episode, we revisit these fundamentals and explore how certain marketplaces, like Cdiscount, rely on data and artificial intelligence to strengthen their position and maximize performance.

 

Go back to basics to ensure the success of your marketplace

 

Secure your margins

The margins achieved by a marketplace are on average three times less than those from a traditional e-commerce operation. This margin is dependent on the rate of commission taken on sales, which is, on average, 10%*, but which can range between one and three according to the type of product and marketplace. Not everyone has Amazon’s power to be able to fearlessly increase fees, year after year, to cover the losses caused by services such as Amazon Prime**.

Maintaining margins is a complex exercise because you have to make sure that you remain attractive to all stakeholders. To address this, you need to take into account:

 

  • The need to ensure the success of sellers and to build loyalty with effective investment in marketing.

  • The significant investment in technology needed to ensure the performance of the site.

  • The complexity of the calculations when you choose to offer a logistics service and need to distribute the costs between different businesses.

  • The attractiveness of prices and promotions, as buyers are often looking for bargains on the Internet.

  • The desire to conquer the market quickly, which can lead the marketplace to reduce its fees in order to attract vendors and allow them to offer attractive prices.

  • The importance of preserving your own sales, when you are also an online retailer, in order to avoid lowering your overall margin.

Some experts advocate adopting a holistic approach to calculating and assuring this margin, and not focusing on a simple “pricing vs advertising” equation. Pricing does not generate loyalty – a customer will also remember how easy the platform was to use, the shopping experience, how well presented the range was, and how rich a choice they were offered.

All these factors require investment and innovation. But they will benefit both the sellers and the marketplace’s own business.

Understand your customers

As we have seen, one of the advantages of launching a marketplace for an online retailer is to deepen the understanding of the customer and to draw new strategic direction from this, using the wealth of newly collected data. This profusion of data about customer behavior and product sales can be a goldmine, producing new sources of value and helping to direct the development strategy for the marketplace.

But beware: data collection is a sensitive subject, and not just because of GDPR. The Wall Street Journal pointed the finger of accusation at Amazon in April 2020 for having launched, under its own brand, products that competed with those of its third-party vendors, having first analyzed the sales figures of these products to confirm demand for them.

Nevertheless, to deliver on its promise of capturing traffic, the marketplace needs to collect and analyse this data in order to adapt its marketing tactics. Moreover, as a data producer (since it invests in capturing new clients and generating sales among existing clients), the marketplace can claim ownership of the databases that result from the use of the platform. At the same time, the marketplace must inform its merchants of the existence of a database in the interests of transparency. (Regulation P2B of 12th July 2020). Also be aware of the draft Digital Market Act and Digital Services Act regulations, which aim to limit the powers of marketplaces that are retailers in their own right, to prevent them from competing with their sellers (as in the case of Amazon mentioned above).

Once this data has been captured, the next challenge is to interpret it and to make it available to business teams so that they can use it effectively to optimise the shopping journey. Particularly powerful systems need to be put in place to take into account the growth in the volume of data that can be expected from a marketplace, and the need to use this in near real time to meet customer expectations as closely as possible by individualising their journey.

Focus on a marketplace strategy: cdiscount uses data to optimise personalization

Cdiscount is one of the foremost marketplaces in France. The growth of this business, launched in 2011, accelerated in 2020 when it came to represent 44% of the total sales volume of the group, compared with 38% in 2019, with Cdiscount reaching a total turnover of 2.5 billion euros. In terms of the number of customers, the marketplace took second place, but this was far behind Amazon which grew even more strongly in 2020 (source: Fevad).

That same year, Cdiscount saw its seller numbers increase by 26% and its catalogue by 33%, reaching 100 million SKUs offered by 13,000 sellers.

Making good use of data to improve site performance

Although Cdiscount is one of the leading French e-commerce sites, it is still looking to achieve profitability, and is making significant investments in technology in order to stay in the race. To improve its performance, it is relying on data and began by using this to optimise its search engine, which gathers 2 billion queries each year.

AI was then used to improve the product recommendation systems, using the enormous volumes of data gathered on the marketplace. To be more relevant in terms of recommendations, products have to be ranked as quickly as possible during the customer’s browsing on the site.

The algorithms created by Cdiscount combine offline data (CRM etc) and behavioural data in order to assess the likelihood of purchase or the probability of converting each visitor. The proof of the legitimacy of this approach is that out of ten products put online, half are sold by algorithms***.

Cdiscount also uses this data to offer promotions to undecided customers, thereby avoiding lowering the margin on sales to customers who would have bought without the need for a promotion. This algorithm also takes into account a great deal of data, including the device used, time spent and mouse movements made.

Among the ingredients from Cdiscount’s recipe for success

  • Use artificial intelligence to get the most out of big data.

  • Put in place recommendation algorithms which are even more effective than the seller’s advice.

  • Estimate the customer’s value in order to apply algorithms in the best possible way.

  • Depend on an internal team of 50 data scientists working in close collaboration with the business and IT teams. 

  • Constantly monitor the performance of the site in order to be able to run all personalisation systems with an optimised response time.

* LSA survey 2019 with Sellermania 
** La Tribune - 06/12/2021 - How Amazon is torpedoing global e-commerce
*** Big data exhibition, March 2019

Photo credit: ©istock ©Shutterstock