CPG demand forecasting is the discipline of estimating how many units your brand will sell, by item and by week, so that production, inventory and cash line up with real demand. For an emerging brand it is the difference between a retailer relationship that grows and one that ends after two stock-outs. This guide sets out a practical approach that a small team can run without a large planning department.
Brand Refinery is a CPG consulting firm that helps brands build the supply chain and retail operations they need to win and keep placements at Walmart, Target, Kroger, Costco and other retailers. This article is general information, not financial or legal advice; retailer compliance terms change, so confirm current requirements in each retailer's supplier portal.
Why forecasting matters more at retail
Direct-to-consumer sales are forgiving: if you run out, customers wait. Retail is not. Large retailers measure suppliers on whether orders arrive on time and in full, and many assess fines or deductions when they do not. Walmart's program, for example, is commonly cited at about 3 percent of the cost of goods on non-compliant cases, though you should confirm current targets and rates in Retail Link. Under-forecasting causes short shipments and fines; over-forecasting ties up cash, risks expiry and forces discounting. Our guide to EDI setup and chargeback prevention explains how those deductions arise.
Step 1: Gather the right data
A useful forecast starts with clean inputs. Collect the following by item and, where possible, by retailer:
- Shipments from your warehouse or 3PL to each customer.
- Retailer point-of-sale or scan data, where the retailer shares it.
- Syndicated data from providers such as SPINS, NielsenIQ or Circana, for category and competitor context. See our guide to CPG retail data.
- Your promotional calendar, with dates, depth of discount and mechanics.
- Open purchase orders and any new store or distribution gains.
- Supplier lead times and minimum order quantities.
Shipments are not consumption. Retailers and distributors build and draw down inventory, so a forecast based only on shipments will swing more than real shopper demand.
Step 2: Choose a method that fits your stage
Most emerging brands do not need sophisticated software. They need a repeatable method.
- Baseline from history. Use a simple moving average or seasonal average of recent weekly consumption to set the non-promoted run rate.
- Add known events. Layer in promotions, new distribution and seasonal peaks as separate, explicit adjustments.
- Apply judgment, then record it. Sales and category teams know about shelf resets and competitor moves. Write each adjustment down with a reason so you can check later whether it was right.
- Review monthly at minimum. Weekly is better when a launch is under way.
Brands with no history should forecast from comparable items and category rate-of-sale data, then replace the estimate with actuals as soon as sales begin.
Step 3: Measure accuracy and bias
You cannot improve what you do not measure. Track two things:
- Accuracy. Weighted mean absolute percentage error, usually written WMAPE, compares the size of forecast errors with total sales, so large items count for more than small ones.
- Bias. This shows whether you consistently over-forecast or under-forecast. A brand that always runs a few percent short has a bias problem, not a randomness problem, and the fix is different.
Measure at the level where decisions are made: item by retailer by month is a sensible start. Review the results in a short monthly meeting that includes sales, operations and finance.
Step 4: Plan promotions separately
Promotions are the largest source of forecast error for most brands. Treat them as their own forecast.
- Record the baseline sales for the period before the promotion.
- Estimate the lift from past promotions of the same depth and mechanic.
- Account for the dip after a promotion, when shoppers have stocked up.
- Agree the inventory build with your co-packer far enough ahead to meet its lead time.
Keep a log of every promotion with its planned and actual lift. After a few cycles you will have your own lift ranges, which are far more reliable than rules of thumb.
Step 5: Connect the forecast to production and cash
A forecast only matters when it drives decisions. Convert the unit forecast into:
- Production orders, taking co-packer lead times and minimums into account. Our guide to choosing a co-packer covers the trade-offs.
- Packaging and ingredient purchases, which often have longer lead times than the finished product.
- A 13-week cash view that shows when you pay suppliers and when retailers pay you.
- Safety stock, set from your forecast error and your supplier reliability rather than a fixed number of weeks.
If you also use 3PL or distributor warehouses, see our overview of 3PL, DSD and warehouse distribution.
Step 6: Run a monthly planning cycle
A simple monthly cycle, often called sales and operations planning, keeps the forecast honest.
- Week 1: update actuals and calculate accuracy and bias.
- Week 2: sales proposes changes for promotions, distribution gains and risks.
- Week 3: operations checks capacity, lead times and inventory against the plan.
- Week 4: leadership signs off one plan and the cash view that goes with it.
Common mistakes
- Forecasting from shipments only and ignoring retailer inventory.
- Treating promotions as part of the baseline.
- Failing to record why a number was adjusted.
- Setting safety stock without reference to forecast error.
- Giving the co-packer a forecast that sales has not signed off.
Frequently Asked Questions
What is CPG demand forecasting?
CPG demand forecasting is the process of estimating future unit sales by item, customer and time period so that production, inventory and cash can be planned. It combines historical sales, retailer data, promotional plans and known events, and it is checked regularly against actual results.
How do I forecast demand for a brand with no sales history?
Use comparable items and category rate-of-sale data from syndicated sources to set a starting estimate by store and week. Plan the first production run conservatively, track actual sell-through from the first week and replace the estimate with real data as soon as you have a few weeks of sales.
What is a good forecast accuracy for a CPG brand?
There is no single benchmark that fits every brand, because accuracy depends on item volume, promotion intensity and data quality. Track WMAPE and bias at the item and retailer level, and aim for steady improvement over time rather than chasing a headline number.
How do retailer fines relate to forecasting?
Large retailers measure on-time and in-full delivery and may fine or deduct for shortfalls. Walmart's program is commonly cited at about 3 percent of cost of goods on non-compliant cases, but rates and targets change, so confirm them in the retailer's portal. An accurate forecast reduces short shipments and the deductions that follow.
Do small CPG brands need forecasting software?
Not at the start. A well-structured spreadsheet with clean data, a documented method and a monthly review is enough for many brands. Consider software when the number of items, retailers and promotions makes manual updates slow or error-prone.
Next step
If forecasting errors are causing stock-outs, fines or excess inventory, Brand Refinery can review your data, planning process and retailer obligations. Learn more about our consulting services or our team, then book a strategy call.