In travel retail, a stock-out is not simply a deferred sale. The passenger may leave the country, change terminal or not return for months. Beirut Duty Free wants to reduce that risk by automating forecasting and replenishment with Slimstock in an operation where demand can change quickly by season, route, nationality and geopolitics.
- The platform will automate forecast and replenishment processes that previously required more manual work.
- Integration with Microsoft Dynamics is intended to provide current visibility of sales, inventory and planning.
- The stated objective is to reduce both product shortages and excess inventory while responding faster to traffic changes.
What is happening
Beirut Duty Free and Slimstock announced a partnership on 30 July to improve inventory control across the airport retail operation. Facilitated by Megatech, the project will automate forecasting and replenishment and use current sales data to support planning decisions.
The solution will integrate with Microsoft Dynamics. Beirut Duty Free had already modernised part of its management infrastructure using Dynamics and LS Retail. The new planning layer is intended to turn transactional information into decisions on purchasing, service levels, availability and excess stock.
The announcement identifies three essential variables: season, route and passenger type. Travellers do not buy the same assortment. A return of the Lebanese diaspora, a Gulf route, a European flight or a leisure departure can create different baskets even when total passenger numbers are identical.
Slimstock describes its platform as using statistical models, machine learning and demand-pattern classification. The companies have not yet disclosed the specific model architecture used in Beirut, go-live date, SKU coverage or quantitative targets for accuracy, availability or inventory reduction.
Why airport retail needs another inventory model
The passenger cannot wait for the next delivery
In domestic retail, a customer may return or order online. Duty-free opportunity is constrained by the flight. A bestseller stock-out may transfer demand to a substitute, but it may also destroy the sale. Service levels therefore need to reflect value, margin and substitution.
Demand changes by route and time band
A monthly aggregate forecast can get total volume right and miss the moment or product. Planning needs daily or weekly granularity, with factors for flight, nationality, time, season and event. Replenishment must respond to when demand arrives, not only to past sales.
Beirut operates with extreme volatility
The airport moved from 7.01 million passengers and 24.7% growth in 2025 to 2.02 million passengers in the first half of 2026, about one third below the previous year. Monthly traffic fell to 139,000 in March and recovered above 442,000 in June. A half-year average would conceal both extremes.
Excess stock destroys value too
Overbuying locks working capital, consumes space and increases markdown, ageing and expiry risk. This matters in confectionery, food, cosmetics, seasonal products and items connected with one nationality or route.
Signals
Flights, sales, events and stock create the input.
Model
Patterns, trends and risk shape the forecast.
Decision
Planners set orders, allocation and exceptions.
Execution
Teams replenish, transfer and manage availability.
Learning
Error, bias and outcomes improve the next cycle.
The system improves when every decision feeds the next cycle. Advanced planning connects external signals, modelling, operational judgement, execution and measured results.
What it means for travel retail
Automation can free planners from reviewing thousands of SKU-period combinations. The saved time should be used to manage exceptions: new routes, promotions, supplier delays, regulatory changes and signals the algorithm does not yet understand.
Availability needs commercial prioritisation. Not every product requires the same service level. Exclusives, gifts, bestsellers, local products and items with limited substitution may deserve more protection than a slow-moving SKU with close alternatives.
The supplier relationship changes too. A more stable purchase forecast can support better lead times and fewer emergency orders. In disruption-prone markets, collaboration may include minimum quantities, split deliveries, safety stock and alternative logistics routes.
For the airport, stronger availability can improve conversion and experience without adding floor space. The project becomes strategic only when it connects inventory with traffic, margin, space, promotions and satisfaction, rather than replacing one spreadsheet with another interface.
Where the commercial opportunities appear
Plan by exception
Automate stable demand and focus teams on high-value deviations.
Share operational signals
Connect schedules, cancellations, terminals and flows with commercial planning.
Improve availability
Align production, packs, promotion and stock with routes and seasons.
Scale without overbuying
Forecast diaspora, gifting and cultural-occasion demand.
Receive a steadier signal
Use forecasts and scenarios to improve delivery and capacity.
Reduce emergencies
Optimise batches, consolidation, alternatives and lead times.
Unify decisions
Integrate ERP, POS, inventory, flights, promotions and master data.
Release working capital
Balance service, margin, risk and cash tied up by category.
Can your forecast distinguish a real demand decline from a temporary disruption?
Risks and practical barriers
- Automating poor data. Inventory, SKU and promotion errors become poor recommendations.
- Confusing sales with demand. A stock-out lowers observed sales and can mislead the model.
- Unrecorded promotions. The system may treat a promotional peak as baseline demand.
- Structural breaks. War, airspace closure or a new route can invalidate history.
- One-dimensional targets. Minimising stock without protecting availability destroys sales.
- Too many overrides. Constant manual intervention prevents learning and evaluation.
- Weak adoption. Planners need to understand logic, limits and exception handling.
- No baseline. A project can appear successful when it has only changed the workflow.
Slimstock describes its platform as using AI and machine learning. The Beirut announcement confirms automation, integration and operational objectives, but does not disclose the models used, expected improvement or achieved results.
How Marksyte can help
Marksyte can connect planning technology with airport-specific data and rigorous commercial measurement.
Data diagnosis
Review sales, stock, stock-outs, promotions, SKUs, flights and calendars.
Demand segmentation
Classify products by stability, intermittency, seasonality and substitution.
Passenger variables
Add route, nationality, time, connection, season and purpose.
Probabilistic forecasts
Generate ranges and scenarios instead of one apparently exact number.
Inventory policies
Define service levels, safety stock, order quantities and category rules.
Exception management
Prioritise alerts by economic impact and ability to act.
Value measurement
Separate availability, inventory, margin, capital and planner productivity.
AI governance
Assign ownership, limits, validation and override monitoring.
A practical 90-day agenda
- Build the baseline. Measure forecast error, bias, availability, excess, emergencies and capital.
- Clean the signal. Separate normal sales, promotion, stock-out, return and exceptional events.
- Classify the assortment. Define families with different patterns and service levels.
- Add flight data. Test routes, times, nationalities and operational changes as variables.
- Design governance. Agree alerts, overrides, accountability and performance review.
AI-driven inventory planning cannot remove Beirut’s uncertainty. It can turn it into faster, more consistent and measurable decisions. The outcome will depend less on algorithm sophistication than on data quality, process discipline and the ability to act when reality moves away from the forecast.
Frequently asked questions
What will Beirut Duty Free automate?
The announcement covers demand forecasting and replenishment, integration with Microsoft Dynamics, and improved visibility of inventory and planning. A detailed timetable, SKU scope and implementation results have not yet been published.
Will artificial intelligence replace inventory planners?
It should not. The platform can select models, identify patterns and recommend orders, but decisions still need context on flights, security, suppliers, promotions, launches and commercial priorities. Value comes from combining automation with operational judgement.
Which indicators should evaluate the project?
Availability, stock-outs, excess and ageing, inventory turn, emergency orders, service level, forecast accuracy and bias, working capital, lost margin and planner time. They should be measured by SKU, category, route, time band and forecast horizon.
Sources
- Slimstock, Beirut Duty Free and Slimstock team up, 30 July 2026
- Global Travel Retail Magazine, partnership overview, 28 July 2026
- Slimstock, forecasting, machine-learning and alert capabilities
- BLOMINVEST, Beirut airport traffic in 2025, 14 January 2026
- Al Modon, first-half 2026 airport traffic, 20 July 2026
- LS Retail, Beirut Duty Free Microsoft Dynamics infrastructure
- Beirut Duty Free, corporate profile
Project facts and traffic figures come from the cited sources. Forecasting, inventory and measurement recommendations are a Marksyte analytical framework.