Demand Planning · FMCG

    Demand planning for FMCG, where the history lies to you

    Promotions, distributor loading, festive peaks and launches all move your numbers for reasons that have nothing to do with demand. Likwid forecasts every SKU and region with those causes modelled rather than averaged over — then nets it into what to make and buy.

    Planning pharma instead? See demand planning for pharma.

    app.likwid.co.in/inventory/demand-forecasting
    Demand Forecasting
    3,242
    SKUs
    1,583
    Smooth
    785
    Intermittent
    408
    Erratic
    434
    Lumpy
    ItemPer Day
    Detergent Powder 1kgHome Caresmooth2,747
    Hair Oil 200mlPersonal Caresmooth1,563
    Biscuits · Glucose 300gFoodssmooth3,733
    Toilet Soap 100gPersonal Careintermittent2,143
    Face Cream 50g · WinterSeasonallumpy297
    The short answer

    What makes FMCG demand planning different?

    The maths isn't the hard part. The record you forecast from is distorted — and by things you did on purpose.

    Every scheme, every bit of distributor loading, every festive week and every launch leaves a mark on sales history that looks exactly like demand. FMCG demand planning is the work of separating those causes from the underlying baseline, across thousands of SKUs and every region — then turning the result into a plan for what to make, hold and buy.

    Why it breaks

    Four ways an FMCG forecast goes wrong

    None of them are arithmetic errors. They're all the same mistake — treating a caused number as an ordinary one.

    01

    Every promotion rewrites your history

    A scheme month sells three times the baseline. Next year the model reads that spike as ordinary demand and plans for it — so you build to a number that only existed because you discounted it. Unless the promotion is recorded as a cause, the uplift quietly becomes the forecast.

    02

    You forecast dispatches, not offtake

    Primary sales tell you what left your warehouse. Secondary sales tell you what the consumer actually bought. When the distributor is loaded, primary looks healthy while offtake is flat — and the correction arrives as returns and a dead month.

    03

    The festive peak moves every year

    Diwali, Eid and the wedding season don't sit on the same dates twice, and they don't scale evenly by region. A month-on-month model built on last year's calendar plans the peak a fortnight late — which in FMCG is the whole peak.

    04

    A launch quietly eats its neighbours

    The new variant hits its number, and nobody notices that most of it came out of the SKU next to it on the shelf. Both forecasts stay wrong until someone reconciles the category by hand.

    Inside the platform

    From a forecast to a purchase order

    The same three screens every cycle — what demand looks like per SKU with the pattern named, what it nets to once stock and open orders are counted, and the order that goes out.

    ARIA · Forecast
    Signals → SKU-level demand
    4,180 SKUs
    ItemForecastConfPattern
    Detergent Powder 1kgFG-DET-1K82,400 No88%Promo
    Hair Oil 200mlFG-HRO-20046,900 No84%Seasonal
    Biscuits · Glucose 300gFG-BIS-3001,12,000 No86%Trending
    Toilet Soap 100g · SandalFG-SOP-10064,300 No79%Steady
    Face Cream 50g · WinterFG-FCR-0508,900 No61%Lumpy
    How Likwid plans it

    Model the cause, then forecast the baseline

    01

    Promotions modelled as a cause, not noise

    Scheme periods are marked and their uplift held separately from baseline demand, so the promotion explains the spike instead of becoming next year's plan.

    02

    Forecast on secondary, plan on primary

    Where secondary sales are available, demand is sensed from offtake and the distributor's stock position is netted off — so you plan against consumption rather than loading.

    03

    The festive calendar, not the month number

    Seasonality is tied to the actual event dates and their regional spread, which is why the peak lands where the peak lands rather than where last year's cell sat.

    04

    Every SKU, including the long tail

    Thousands of SKUs across branches each get their own method and confidence — including the slow and lumpy ones nobody had time to plan in a spreadsheet.

    Built on Likwid's demand forecasting engine, feeding S&OP and procurement.

    0+
    SKUs forecast per cycle
    0
    branches, each its own signal
    Promo · festive
    modelled as causes
    Primary · secondary
    reconciled where available

    Illustrative of how the platform is designed to behave — actual figures depend on your catalogue and data.

    Questions

    FMCG demand planning, answered

    Food & beverage specifically? See ERP for food & beverage manufacturing · inventory · pricing.

    See it forecast your own catalogue

    A 30-minute walkthrough on your SKUs — promotions separated from baseline, the festive peak where it actually falls, and the buy that follows.

    Talk to us

    See Likwid on your own data

    Leave your details and we'll set up a 30-minute walkthrough on your numbers — no slides. We reply within a day.

    Prefer email? admin@likwid.co.in