Lyreco and RELEX: Reliable Forecasting Requires More Than an AI Planning Tool
Guillaume Trécan, journalist and editorial leader at Républik Supply Le Média, interviewed Lyreco Demand & Supply Director David Licour in an article published on December 14, 2023. The discussion covered Lyreco's selection of RELEX for forecasting and replenishment planning, its efforts to strengthen sales forecast reliability and its ambition to develop a broader Sales and Operations Planning (S&OP) operating model. This insight examines Guillaume Trécan's reporting as it stood on December 14, 2023. It is historical credited coverage, not a new announcement or a newly verified account of Lyreco's current implementation status, operational performance or subsequent results.
What the original reporting established
David Licour described a need to improve the reliability of sales forecasts because forecast weaknesses had affected replenishment and inventory management. According to his account, unreliable forecasts had contributed to excess inventory, while some product ranges had experienced two to three times their usual level of stockouts. Supplier service performance was also a concern. Licour indicated that conditions had improved since late 2022, but remained below Lyreco's ambitions.
Lyreco had invested in the RELEX Advanced Planning System (APS), selecting it for its forecasting capabilities. At the time of the interview, teams were designing and configuring the solution, with a first-country pilot planned for the end of 2024. The article does not establish whether that pilot or any subsequent country deployments were completed.
The initiative extended beyond software implementation. Lyreco was strengthening its demand forecasting function and developing S&OP processes involving Demand & Supply, Merchandising and Sales. The planned training program also involved Purchasing and senior management. Central and local key users were intended to support the deployment across countries.
Licour described a trust problem with the previous planning tool: replenishment teams frequently corrected its outputs because they did not fully trust the recommendations. During RELEX configuration, Lyreco was restricting many opportunities for manual forecast correction to encourage greater reliance on the new system. These statements describe the intended change in planning behavior, not an independently demonstrated improvement in forecast accuracy or planner adoption.
Master-data governance was another priority. Lyreco wanted product, supplier and customer information to remain current across its countries. Merchandising was expected to determine which product ranges required particularly high availability. For selected major B2B customers, customer-level forecasting was also envisaged through dialogue with Sales teams.
Supplier service was part of the operating challenge. Licour explained that weaker service from some suppliers could lead Lyreco to hold additional safety stock without fully eliminating stockouts. This connected demand forecast reliability with supplier performance, inventory buffers and customer service rather than treating forecasting as a standalone statistical problem.
Dataleo analysis: forecasting trust is a decision-design problem
The important connection in this case is between forecast automation, the quality of business inputs and the authority to change the plan. A forecasting engine can produce recommendations, but a replenishment organization must decide which outputs to accept, which exceptions require intervention and how conflicting information from suppliers, product teams and customers is reconciled.
Reducing manual overrides can help create a consistent planning baseline, but limiting planner intervention also introduces a trade-off. If the new system receives outdated product information, incomplete customer signals or unrealistic supplier lead times, restricting corrections may increase rather than reduce decision risk. The practical requirement is to distinguish uncontrolled manual changes from justified, documented exceptions.
Lyreco's reported approach to Merchandising illustrates another dependency. Product availability is not solely a forecasting metric. It reflects commercial priorities, assortment decisions, inventory policies and supplier performance. Identifying priority ranges can help translate commercial strategy into replenishment parameters, but the organization must decide who owns the trade-offs between service, working capital and stockout exposure.
The move toward S&OP is therefore more than a training exercise. It requires a shared planning cadence, agreed assumptions and escalation rules across Demand & Supply, Merchandising, Sales and management. Customer-level forecasts for major B2B accounts also require a clear process for incorporating commercial intelligence without turning sales expectations into unchallenged demand commitments.
Supplier service adds a further constraint. Better demand forecasts cannot by themselves eliminate shortages caused by unreliable supplier delivery. Nor can additional safety stock indefinitely compensate for poor supply reliability without increasing inventory exposure. A credible assessment should evaluate forecast accuracy alongside supplier service, inventory coverage, stockouts, planner overrides and the quality of replenishment execution.
Dataleo angle: confidence in automation must be earned
Lyreco's reported transformation illustrates why confidence in an AI-enabled APS cannot be imposed simply by reducing manual adjustments. Its decision value depends on current master data, commercially agreed availability priorities, realistic supplier constraints and clear responsibility for forecast exceptions. A controlled forecast baseline can improve consistency, but decision owners must retain an explicit process for correcting material errors and challenging unrealistic recommendations. The main failure mode is replacing low-trust manual planning with high-confidence automation built on incomplete business assumptions. This is Dataleo's interpretation of the December 2023 interview, not evidence that Lyreco achieved or failed to achieve its stated objectives.
Historical evidence and limitations
The source is a trade-media interview published on December 14, 2023. It records Lyreco's reported challenges, management intentions, APS configuration work, training plans and planned deployment timetable. It does not provide an independently verified before-and-after assessment of forecast accuracy, service levels, inventory reductions, planner adoption, financial return or completed implementation. The reported first-country pilot planned for late 2024 must not be interpreted as proof that deployment subsequently occurred or delivered operational improvements.
Source and credit
Original interview and reporting by Guillaume Trécan: Lyreco : Nous ressentons un besoin crucial de fiabiliser la prévision de vente, Républik Supply Le Média, December 14, 2023. The factual account and statements attributed to David Licour are credited to Guillaume Trécan and the original publication. The analysis of planning authority, forecasting trust, decision ownership and failure modes is Dataleo's independent editorial interpretation.
Illustration credit
Editorial stock illustration: warehouse stock-handling photograph by cottonbro studio, sourced from Pexels. The image is thematic only. It does not depict Lyreco facilities, RELEX software, the reported project, David Licour or Guillaume Trécan. Its use does not imply endorsement by the photographer, Pexels, Lyreco, RELEX or Républik Supply.