A campaign aimed at running enthusiasts can already call for several versions: depending on practice, previous purchases, loyalty status or available products. As targeting becomes more refined, production work increases. You have to choose the elements, compose the messages, check their consistency and prepare for their distribution.
This burden ultimately determines the level of customization actually accessible to teams. “The teams will limit themselves in segmentation since the more segments I have, the more campaigns I will have to do,” explains Vincent Martinet, founder of Reelevant. Knowledge of the customer can be fine, when the resources necessary to exploit it remain limited.
It is on this constraint that the company positions itself. Intended in particular for CRM teams responsible for loyalty, its platform automates the composition of content based on information available within the company. The objective is to allow more variations without imposing equivalent manual work for each.
Bring personalization through content
In the organization described by Vincent Martinet, the teams build a commercial animation plan around launches, promotions and highlights of the year. They then define the populations to be addressed and the associated content.
Reelevant proposes to shift part of this differentiation when producing content. The teams prepare reusable elements and the rules governing their use. The platform assembles them based on the data available for each recipient.
“We are going to assemble assets, like layers,” describes the manager. These elements may be products, banners or content from the site. Their selection can take into account loyalty status, in-store purchases or a navigation path.
Vincent Martinet takes the example of Decathlon: a trail runner could receive content taking into account his interests, his location and the weather. Personalization then concerns the composition of the message and its context of use.
The segmentation criteria remain. Their application moves into the content engine, which should allow teams to reduce the number of versions to be produced separately. The economic promise rests on this ability to increase the volume of variants faster than the workload.
Automation that is not always AI
In a sector where artificial intelligence occupies an increasing place in commercial discourse, Vincent Martinet introduces a useful distinction: “The automation part is not necessarily AI. »
Applying a loyalty rule, selecting an authorized element or assembling several components can be based on a logic defined in advance. The value of these operations lies in their systematic execution and scale.
The manager also mentions the uses of AI in calculations and in interactions with the product. But he locates a significant portion of current value in the ability to automate processes that would be too costly to perform manually.
This distinction allows us to better appreciate the work of the editor. For a marketing department, the benefit must be translated into production time, reliability and commercial performance. The use of AI provides information on the means used; it is not enough to measure the result.
Work shifts toward rules and control
Automation, however, requires preparation. The available elements must be usable, the data sufficiently reliable and the rules consistent with the campaign objectives.
Reelevant has a content construction tool in which teams define colors, structures and brand constraints. Vincent Martinet cites Decathlon, Sephora, Leroy Merlin and FDJ to illustrate these requirements: “There are still business rules that are put in place, which will constrain the capacity of what we can display or not in a given context and for a given client. »
The workload thus shifts towards the definition and supervision of the system. A team that produces fewer variations manually must ensure that the authorized combinations remain relevant. A poorly configured rule or information that has become obsolete can affect many variants.
An additional expense to defend
Reelevant integrates with the customer’s equipment, the platform can use data from Snowflake, BigQuery, product feeds or internal files. The content is then used by the distribution solutions already present.
“We don’t send marketing messages,” explains Vincent Martinet. The manager cites Salesforce and Adobe in particular among the environments in which content can be consumed.
This position allows existing campaign tools to be preserved. It also involves justifying an additional expense. “We are not in a replacement market, we are really in an equipment market,” believes the founder. According to him, the commercial process often involves developing a budget and demonstrating that the problems identified represent a sufficient cost to justify the investment.
The speed of integration comes into play in this equation. Vincent Martinet indicates that the implementation at Sephora took between four and six weeks. In addition to the technical connection, there is the preparation of the contents, validations and the organization of their use. It is all of these expenses that the expected gain must cover.
Measure what each variation yields
The business model described is based on an annual subscription, the price of which varies depending on the number of people exposed to Reelevant content during the year. Billing is therefore linked to the audience covered.
Vincent Martinet, for his part, relates the value of the product to “performance results”, ROI and the income generated. These two dimensions must be distinguished: the interview does not describe remuneration contractually indexed to additional sales.
To assess the economic benefit of the platform, several measures are necessary. The time spent on creation and validation makes it possible to evaluate production savings. Conversions and margin inform business results. Unsubscribes can also signal the limits of perceived relevance.
A comparison with a control group would make it possible to distinguish additional purchases from those which would have taken place without personalization. Producing more content, or associating sales with its distribution, is not enough to establish the incremental gain.
Agents could reduce setup time
The next step mentioned by Vincent Martinet concerns how to order the software. A user could ask an agent to create a rule or banner from their work environment, for example Teams, and then obtain performance feedback without browsing dashboards.
After assembling the content, part of the configuration could thus be automated. However, transcription does not make it possible to precisely distinguish the capacities already deployed from the uses envisaged.
The manager above all identifies an adoption constraint: “I think the biggest obstacle is not the technology, but the ability to adopt it. » In large accounts, the availability of a function must still meet procedures, authorizations and teams capable of integrating it into their work.
For marketing departments, the question will remain that of the use of this additional capacity. If producing a new variant becomes less expensive, it will still be necessary to decide which one is worth releasing. Personalization also finds its limit in the usefulness of each message for the person who receives it.
In April 2025, Reelevant completed a Series A of 6 million euros led by Seventure Partners, with the participation of GO CAPITAL and business angels, including Nathalie Balla, former director of La Redoute.
