Beverage distribution: reach August with the beer your bars will actually order.
For distributors supplying hospitality, breweries and wineries with their own delivery fleet. Hundreds of bars that order a little every day, a summer that multiplies sales and local festivals that move from village to village each week. The system learns that calendar from your history and proposes what to order from the plant and what to load on each truck.
Book a 30-minute demoHow it works
Read-only connection · Tested on your own data first · Data hosted in the EU · Demos in English or Spanish
The short answer
Beverage distributor inventory forecasting predicts what the bars on each route will order, with summer and local festivals built in, and proposes what to order from the brewery or plant and what to load on each truck. It treats each format (keg, returnable bottle, can) as a separate product, keeps the balance of kegs and crates held by customers, and does not propose ordering more than will sell before its date. It suits distributors, breweries and wineries with their own delivery.
By Manuel Gómez González, founder of Arume · Reviewed on
What it does
The festival calendar of each area, learned
A mountain village's festival drives keg sales for five days in August and the rest of the year that bar orders two crates a week. The system finds those peaks in your sales history by customer and area, places them on their dates even when they move from one year to the next, and uses them when ordering from the plant weeks ahead.
Factory orders format by format
The same beer in a 30-litre keg, a returnable bottle and a 33 cl can are three products with three audiences and three seasons. Each format is forecast separately and the order comes out per format, in full pallets if the plant ships that way.
Returnable packaging in view
Every keg and crate that goes out has to come back. The system keeps the balance by customer and route, flags bars that accumulate empties, and calculates whether what will come back this week is enough for the next load from the plant, which many breweries only ship against returned packaging.
Each route's load, with pre-sales in front
With the orders the salesperson took in pre-sales and the forecast of what each customer usually asks for on the day, the system proposes what to load on each truck. Fewer trips back to the depot for a bar that ran out of sparkling water, and less product that rides the whole route without coming down.
Short shelf life, batch by batch
Unpasteurised craft beer, juices and some soft drinks have short dates. The system knows which batch you hold and when it expires, and does not propose ordering more than you will sell before then. A batch that is tight goes to the top so deliveries take it first.
Customers that open only in season
The beach bar open from June to September, the pool terrace, the ski-station bar. None has history in March and all order heavily on day one. They are marked as seasonal and forecast from their previous seasons, not from last month.
Why does a year of beverages fit into ten weeks?
A distributor supplying hospitality sells in July and August what it sells between January and April together, and a good part of that summer is concentrated in long weekends and local festivals. The rest of the year the job is the opposite: hundreds of bars ordering little and often, sometimes daily, expecting delivery the next day. Both at once, with the same warehouse and the same trucks.
The factory order cannot wait to see what happens. A large brewery ships weekly and asks for a forecast in advance, a small winery bottles when it bottles, and water and soft drinks arrive in pallets that take the space they take. If you are short of keg in June, there is no recovering in August, and if you overbuy cans, they are still there in October.
Today that order is made by one person with last year's sheet in front of them and what they remember of the summer. It works while that person is there and the year resembles the last one. Inventory forecasting for a beverage distributor does the same sum for every customer and every format, and explains it.
Why is each area's season different from the warehouse average?
The warehouse average misleads. The coast peaks in August and halves in September, the inland area has its peaks at each village festival from June to October, and the city dips in August because half the bars close. Added together it looks like a mild summer. Area by area, it is three different businesses.
The system draws the season of each area and product family with all its customers together, because a single bar has too little data, and adjusts each customer's level with its own sales. Peaks that repeat on the same dates every year (patron-saint festivals, a pilgrimage, a music festival) are learned as what they are. One-off peaks (a large wedding, a match) stay as isolated peaks and do not inflate next year's order.
If a festival falls a week earlier because the calendar says so, it moves with it. With your area's festival calendar loaded the date is fixed; without it, the system estimates it from previous years. Temperature can be added as a further variable and is kept only if it improves the forecast on your history.
How do you forecast by format?
One beer reaches your warehouse in three or four formats and each has its buyer. Keg goes to bars and festival counters. Returnable bottle goes to the traditional bar. Cans and multipacks go to shops, beach bars and open-air events. A sales rise for that beer in July tells you nothing useful unless you know in which format it happened.
So the forecast is per format, and the factory order comes out per format and in the units the plant ships: complete pallets, layers, cases. If a can pallet is 2,400 units and you are short of 600, the system does not order 600: it compares the cost of ordering the full pallet with the cost of being short until the following week, and shows both figures.
- Keg: follows festivals and bars with a tap; very sensitive to local season.
- Returnable: steady hospitality customer, demand tied to the balance of packaging.
- Can and multipack: shops, beach bars and events; rises with heat and long weekends.
- Bag-in-box and post-mix: few customers, large and spaced-out orders.
Why does packaging count as stock?
A distributor with returnables has two warehouses in one: full product and empty packaging. Kegs and crates leave with the delivery and come back, or do not, days or weeks later. If they do not come back, the plant will not ship the next order in full, and in summer that is a stock-out under another name.
The system keeps the packaging balance by customer and route using delivery and collection notes. It flags bars holding more empty kegs than usual so the driver collects them on the next visit, and calculates whether the empties coming back this week are enough for the plant load. It also works the other way: a customer that has not returned packaging for weeks and orders less is usually buying elsewhere, and your salesperson can call before you lose them.
Reuse is also a regulatory direction in Europe. According to the European Commission, Regulation (EU) 2025/40 on packaging and packaging waste includes binding reuse targets and applies from 12 August 2026. Which obligations reach your company depends on your products and country, and it is for you and your adviser to confirm. The system helps by keeping the packaging balance in order.
From the factory order to the truck load
There are two chained decisions. The weekly one: how much to order from the plant for each product and format, counting each supplier's lead time and the coming season. And the daily one: what to load on each truck from pre-sales orders and what each customer tends to ask for on the road.
The first moves the most money, and the system solves it with its risk calculation: for each product, the probability of reaching zero before the next truck arrives from the plant, and the quantity that pays in money terms. The second is simpler but saves trips back to the depot and product that rides the whole round without coming down. The order of stops stays with your routing software.
Short dates, vintages and what the law says about dates
Not all drinks last the same. Water and soft drinks have long dates, while unpasteurised craft beer and some juices do not. If your system records batch and date, the order is calculated to sell before expiry and tight batches leave first. If it does not, we work with the date of receipt and the average life of the product.
Wineries have their own problem: the vintage. Every autumn the same wine appears under a new code and, for an ERP, with no history. In the first review we join the vintages of each wine so the forecast continues, and the system warns when the remaining vintage will run out against when the next one is released. Dates help little here: under Regulation (EU) No 1169/2011, Annex X, drinks containing 10 percent or more alcohol by volume do not need a best-before date, so the thing to watch is the vintage.
What do you need to provide, and how do you start?
Sales by customer, product, format and date (delivery notes will do), current stock, receipts from the plant with order and arrival dates and, if you keep them, packaging movements. With two summers of history the system already sees each area's season. No personal data is needed: customer, area, product and quantity are enough. It reads from Odoo, Microsoft Dynamics 365 Business Central, Sage, Holded, SAP Business One or an Excel export, with a read-only user.
First, the review: we clean the data, rebuild stock day by day and replay your last summer without knowing the future to compare with what you did. That is the simulation with your data. Then one or two months in shadow mode, proposing each week's factory order while you decide as usual. Questions about stock can later be put to the AI assistant, and urgent alerts can arrive by WhatsApp.
Frequently asked questions
Does it work for a distributor that uses pre-sales?
Yes, and that is where it shows most. Orders the salesperson takes the day before are certain demand; what the system adds is the forecast of what has not been ordered yet: the bar that calls mid-morning, the customer the driver persuades on the spot. With both, the truck load is calculated and, a layer above, the plant order.
How does it know when each village festival takes place?
It sees them in your history: a group of customers in the same area that multiplies orders for a few days every year. With two summers of data they appear. If you also give us the festival calendar of your delivery area, it is loaded and each peak is placed on its exact date, even if one year it falls a week earlier.
What happens with kegs and returnable packaging?
They are tracked as another stock, with a balance per customer: what went out full and what came back empty. The system flags customers accumulating packaging and calculates whether the empties due back this week are enough for the next plant order. Deposits and packaging invoicing stay in your ERP.
We sell craft beer with a short date. Is that handled?
Yes. If your system stores batch and best-before date, the order never proposes more than will sell before expiry, and tight batches appear at the top to go out first. If it does not store batches, we work with the receipt date and the average life of each product.
We are a winery. What happens when the vintage changes?
The new vintage is a different code in your system but sells to the same customers in the same season. In the first review the vintages of each wine are joined so history does not break every year, and the system warns in good time when the current vintage will run out.
Can it also decide the load of each truck?
It proposes what to load on each route from pre-sales and each customer's forecast, and the warehouse manager adjusts it. The route itself (order of stops) stays with your routing software or your team: we do not touch it.