What the planner sees
Demand forecasting
For every SKU, customer and region, 30–90 days ahead, factoring in history, seasonality and sales channel.
Home · Case studies · FMCG
Case study · FMCG, commercial planning
From Excel files to a unified commercial planning platform
An international personal care products manufacturer unified forecasting, pricing, discounts and promotions in a single tool — instead of dozens of scattered Excel files.
10×faster planning cycle
Sell-in ↔ Sell-outin one loop
AIanswers questions about the data
10×faster commercial planning cycle
001
Why dozens of Excel files stopped working
An international personal care products manufacturer ran commercial planning in scattered spreadsheets — forecasting, pricing, discounts and promotions each lived separately.
01
Dozens of scattered files
Forecasting, pricing, discounts and promotions were kept in separate Excel files with no single source of truth.
02
Two weeks per planning cycle
Collecting and reconciling data from different files and systems took up most of the cycle.
03
Sell-in and sell-out — tracked separately
There was no unified picture of what was shipped to the customer versus what was actually sold to the end buyer.
Forecasting
→
Pricing
→
Discounts and promotions
→
Sell-in ↔ Sell-out
Specifics
Four tools instead of data exports
Users work with interfaces, not data exports: forecasting, pricing, discounts and data automation — in one tool, with country-specific price lists and forecast versioning.
What the team gets
Pricing, discounts and promotions
A discount and terms builder: rules, recommendations and a change history by customer, region, volume and promotion.
What’s visible every day
Commercial dashboard and data
Sales, prices, ASP, stock levels, turnover and plan-vs-actual in a single window. Excel, 1C and BI — updated daily.
Results
What the business gained
A unified commercial planning platform instead of scattered spreadsheets.
Margin
+10–20%
Growth from accurate forecasting and unified prices and discounts for every SKU and customer.
Warehouse and logistics costs
−15–20%
A reduction from more accurate demand forecasting and stock planning.
Manager time spent on reporting
−50–70%
Data updates automatically — less manual reconciliation and fewer exports.
Project payback
3–6 months
The platform grew out of the data warehouse: first the data, then the models, then the interfaces.
In the interface
One loop for sell-in and sell-out
We plan what to ship and at what price, and see what is actually being sold to the end buyer — in a single window, broken down by channel and period.
udev.forecast · Dashboard · Sales and profit over time
Sell-in ↔ Sell-out
Shipment plan and actual sales side by side
Country price lists
Russia, Belarus, Kazakhstan
Forecast versions
A history of plan changes
AI assistant
From a plain-language question to verifiable SQL
A team of agents works on top of the unified data. The question is asked in plain language, and the system breaks it down and assembles queries according to a verified contract.
01
A question in natural language
→
02
Agents work in parallel: points of sale, assortment, price, demand
→
03
The query is assembled by code following a verified contract
→
04
An answer traceable to specific queries and rows
udev.forecast · Sales analyst · Agent council
udev.forecast · Research log
41 of 68 points of sale are declining year over year−14.3% year over year (2,296,400 → 1,968,100 units), but 12 points of sale account for almost all of the decline.
23 of 68 points of sale — for the latest month−6% month over month — and the set of problem points of sale has already changed.
The principle of trusting the numbers
The model does not write SQL itself — it describes what needs to be calculated, and the query is assembled by code following a verified contract. Every figure is traceable to a specific query and data rows. The connection is read-only. Example run: 3 of 5 waves, 14 queries, 21 facts, 7 hypotheses, coverage 7 of 7.
One platform instead of dozens of files
The company got a connected commercial planning system: forecasting, pricing, discounts, promotions and AI analytics — on one platform, with transparent sell-in and sell-out.