Forecasting & Demand Planning
Baselines, scenarios and forecast improvement built for real S&OP cycles in life sciences and consumer goods, with exception views and accuracy tracking your planners will actually use.
Life Sciences
Life science demand runs on patients, dosing and channels, not just shipment history. We build forecasts that brand, finance and supply teams can all work from, from the 0–24 month plan to the long-range view.
What we factor in
- Patients on therapy
- Dosing and pack mix
- Wholesaler and specialty channels
- Label and competitive events
- Seasonality
Patient-based demand
Patients → units by dose and pack
We forecast patients on therapy from new starts and persistency, then convert them to units by dose, strength and pack size, so the plan reflects how the product is actually used.
What it gives you: Run patient-based S&OP that brand, finance and supply teams can each trace.
Specialty and oncology planning
Specialty pharmacy, buy-and-bill, weight-based dosing
Specialty and oncology products move through specialty pharmacies, distributors and buy-and-bill accounts, with weight-based or cycle-based dosing. We model each channel and regimen separately, so specialty demand is not forced into a retail pattern.
What it gives you: Plan specialty demand by channel and regimen, not as one national number.
Sell-in vs. demand
852/867 data and wholesaler inventory
Using 852 inventory and 867 sales-out data, we separate what wholesalers order from what patients and pharmacies consume, so stocking swings and buy-ins aren't mistaken for changes in underlying demand.
What it gives you: Forecast true demand and see channel inventory building before it unwinds.
Event-based forecasting
Label expansions, competitor entry, guideline changes
We layer known and likely events onto the baseline, such as a new indication, a competitor launch or a treatment guideline update, each with its own timing, size and probability.
What it gives you: Show leadership exactly which events move the plan, and by how much.
Near-term to long-range plans
0–24 month demand linked to the 3–10 year plan
We link the 0–24 month operational forecast to the 3–10 year long-range plan, so near-term demand, supply plans and LRP assumptions come from one model.
What it gives you: Keep the S&OP number and the long-range plan from drifting apart.
Accuracy and governance
MAPE, bias and forecast value added
We score every brand, strength and channel forecast for error and bias each cycle, and set governance rules for who can change the number and when, so accuracy is reported, not assumed.
What it gives you: Hold the forecast process to a measured standard in every S&OP review.
Consumer Goods & Retail
Thousands of SKUs, locations and channels, each with its own pattern. We give planners a statistical baseline for every series, scenarios for S&OP, and a clear view of where their time matters most.
What we factor in
- SKU × location × channel history
- Point-of-sale data
- Promotions
- Seasonality
- Distribution changes
Statistical baselines
Best-fit per series (Geneva)
The Geneva engine tests ten statistical methods on every SKU, location and channel series, tunes them, detects seasonality and outliers, and keeps the best fit, with MAPE, MAD and RMSE reported.
What it gives you: Start every S&OP cycle from a tested baseline, not last year plus a percentage.
S&OP scenarios
Plan, upside and downside
Plan, upside and downside scenarios share one baseline and one set of assumptions, so sales, finance and supply discuss the same numbers and the gap between them, not whose spreadsheet is right.
What it gives you: Agree on one demand plan faster, with the risks visible.
Short-horizon demand sensing
POS-driven adjustments
For the next few weeks, we adjust the forecast with the latest point-of-sale and order data, so replenishment reacts to what shoppers are buying now rather than last month's plan.
What it gives you: Catch demand shifts in days, before they become stockouts or overstock.
Override tracking
Forecast value added (FVA)
We measure whether each planner, sales or management override made the forecast better or worse than the statistical baseline, by product, customer and person, cycle after cycle.
What it gives you: Keep the overrides that add information and drop the ones that add bias.
Category growth forecasting
Category trends and 3–5 year volume
We forecast category and segment growth from market data alongside your own sales, so share targets and the longer-range volume plan rest on a realistic view of where the market is heading.
What it gives you: Set share and volume targets against a category forecast, not a guess.
Quota and expense planning
Baselines for sales targets and budgets
Statistical baselines by territory, account and product give sales quotas and expense budgets a defensible starting point, so targets reflect expected demand rather than last year's number plus a stretch.
What it gives you: Set quotas and budgets your sales and finance teams accept as fair.
How do you improve forecast accuracy?
You improve forecast accuracy by matching the method to the shape of each series, measuring error the same way every cycle, and limiting overrides to the ones that add information. Most planning teams use one method for everything and judge accuracy by feel. That’s where the error comes from.
The Geneva Forecasting™ Engine tests ten statistical methods against every series, tunes each method’s parameters, detects seasonality and outliers, and keeps the best fit. A steady seller gets exponential smoothing, a seasonal line gets Holt-Winters, a launch gets an S-curve, and a spare part with many zero weeks gets Croston’s method. You get the error metrics for each one, so the accuracy conversation starts from numbers.
Forecasting that fits your S&OP cycle
A forecast only helps if planners use it. We build the baseline, the scenarios and the exception views around the meetings you already hold, so the statistical forecast becomes the starting point of the S&OP discussion instead of one more spreadsheet. Planner overrides are tracked against actuals, which shows which adjustments improve the plan and which don’t.
Related case studies
Genentech: oncology and specialty therapy planning
Genentech partnered with RoadMap Technologies to unify short-term operational demand forecasts with long-range strategic supply plans across oncology, neurology, and transplant anti-infective therapy portfolios, eliminating the planning disconnects that drove inventory imbalances and capital misallocation.
0–24 mo. + 3–10 yr. · 3 therapy portfoliosSamsung and Best Buy: collaborative retail replenishment
Samsung implemented a Collaborative Planning, Forecasting and Replenishment (CPFR) framework with Best Buy in partnership with RoadMap Technologies, shifting from shipment-based forecasting to consumption-based planning and reducing supply chain volatility across high-revenue retail accounts.
CPFR framework · Best Buy (Fortune 100)Dell, GE, and Samsung: forecasting in consumer electronics
RoadMap partnered with leaders in consumer electronics, including Dell, GE, and Samsung, to address the sector's rapid product cycles, price compression curves, and multi-tier channel and service complexity.
Lifecycle, price compression & installed-base service
Frequently asked questions
What is forecasting & demand planning?
Demand planning consulting is expert help building the statistical baselines, scenarios and accuracy tracking a planning team runs every S&OP cycle. RoadMap Technologies helps life science and consumer goods teams by putting a governed demand model on their data in about two weeks, using the Oracle-licensed Geneva Forecasting™ Engine, then proving the ROI before they scale.
How is pharma demand forecasting different from consumer goods forecasting?
Pharma demand is driven by patients rather than shoppers. Forecasts start from new starts, persistency and dosing, run through wholesaler, specialty pharmacy and buy-and-bill channels, and have to account for payer access, label events and gross-to-net. Consumer goods forecasts lean on SKU history, promotions and point-of-sale data.
How long does it take to get a forecast live?
We typically have a governed demand model running on your data in about two weeks. Most clients see measurable results within 30 days.
What data do you need to start?
Usually two or more years of sales or shipment history at the level you plan, plus any known drivers such as promotions or price changes. We can work from ERP extracts or spreadsheets.
How do you measure forecast accuracy?
Every series is scored with MAPE, MAD and RMSE against holdout periods before go-live, and against actuals every cycle after. Planner overrides are tracked separately so you can see which ones help.
Will this replace our planners?
No. The statistical baseline handles the routine series so planners can spend their time on exceptions, new products and the judgment calls that need people.
Start with a pilot
Measurable outcomes agreed upfront, a live model on your data in about two weeks, and the ROI proven before you scale.