GL Planning + simulated-ML price/volume forecasting in OneStream
Quick Facts
- Industry: Retail / FMCG
- Role: OneStream Analyst (Accenture)
- Impact: A GL Planning module with custom business rules on the input and review forms, a simulated-ML price/volume forecast built from historical data and run entirely inside OneStream, and a workforce → planning cube integration that removed manual re-entry. All deliverables shipped independently, no escalations.
Overview
Early-career OneStream work at Accenture across two clients. For a global retail client I built a GL Planning module — input and review forms with custom business-rule computations behind them — alongside metadata, dashboards for the key business metrics, and the complex business rules the out-of-the-box needs required. For an FMCG client I built a simulated-ML model that forecasts price and volume from historical patterns without an external ML stack, and an integration that fed the workforce cube into the planning cube.
GL Planning Module (Retail)
The core build was a GL Planning module with custom business-rule computations on both the input and review forms:
- Planners enter base figures on the input forms; custom business rules compute the derived values from those entries.
- The review forms show the aggregated, reviewed numbers using the same computations, so the planner view and reviewer view stay consistent.
Around the module, I also delivered the supporting pieces the client needed: metadata, dashboards for the essential business metrics, complex business rules for the out-of-the-box requirements, enhancements to a prebuilt model per the client's requirements, and rapid administrative rules such as customised copy and clear data rules.
Simulated-ML Price & Volume Forecasting (FMCG)
For the FMCG client I built a model that simulated machine-learning capabilities to forecast price and volume from historical data — but kept it entirely inside OneStream. Rather than standing up an external ML/Python stack, it used historical patterns in the data through rules/formulas to compute forecasted price and volume metrics. The effect was ML-style forecasting that stayed in-platform and auditable, with no black box to govern.
Workforce → Planning Cube Integration (FMCG)
I built an integration that fed the Workforce cube's data into the planning cube, so workforce numbers flowed into the plan automatically instead of being keyed in by hand each cycle.
Enablement
Alongside the build work, I conducted OneStream technical trainings — Finance, Spreadsheet, Extender, Dashboard — for the support team and people newer to the platform.
Results
| Metric | Before | After | Change |
|---|---|---|---|
| GL planning math | pushed out to spreadsheets | custom rules on input + review forms | in-platform |
| Price/volume forecasting | offline / manual | simulated-ML from historical patterns, in OneStream | auditable, in-platform |
| Workforce → plan numbers | re-entered by hand | fed from the workforce cube via integration | automated |
| Input vs review math | drift risk | same computations on both forms | consistent |
Learnings
What worked. Keeping the "ML" forecasting as a simulated, in-platform model rather than a real external ML stack — it forecast price and volume from historical patterns while staying auditable and easy to govern inside OneStream.
Skill developed. Early, broad OneStream exposure — planning forms and business rules, metadata and dashboards, a cube-to-cube integration, a forecasting model, and delivering trainings — all shipped independently. A strong foundation for everything that came after.