SAC Predictive Planning — Time-Series Forecasting Inside Planning Models
As of 2026-09-25
What is SAC Predictive Planning?
SAC Predictive Planning is time-series forecasting triggered from inside a Seamless Planning model version rather than a generic dataset — SAP documents a named workflow for generating and integrating APL forecasts into such a model. The underlying engine is the same APL Auto Time Series capability behind Smart Predict's time-series scenario (C341); the difference is delivery surface, not algorithm: the forecast writes back as a reviewable, lockable planning version.
Predictive Planning is a delivery surface, not a separate engine
SAC Predictive Planning is the name for time-series forecasting triggered from inside a planning model version rather than from a generic dataset — a planner working in a Seamless Planning model version can generate a forecast for a plan-relevant measure and have it written straight into that version as reviewable, adjustable numbers, without leaving the planning workflow. SAP documents a concrete, named workflow for this: generating and integrating Automated Predictive Library (APL) forecasts inside a Seamless Planning model. That naming is the fact worth anchoring on — Predictive Planning is not a separate machine-learning engine sitting beside Smart Predict (concept C341); it is the same underlying automated forecasting capability, APL's Auto Time Series function (concept C344), surfaced specifically inside the planning-model experience rather than the general Smart Predict scenario-building screen.
Why it matters
- Predictive Planning is not a separate engine from Smart Predict's time-series scenario — SAP documents both as running on APL's Auto Time Series function (C344); pricing or scoping them as unrelated capabilities double-counts the same underlying modeling work.
- The forecast a planner triggers lands as a reviewable, adjustable, lockable planning version through the same approval workflow as any other version — a concrete, demonstrable answer to 'how does AI touch our planning process' rather than an abstract claim.
- A planning model without clean, continuous history for the target measure limits what Predictive Planning can produce regardless of APL's performance — remodeling the planning model is often the larger part of a Predictive Planning engagement, not the forecast call itself.
Key points
- Predictive Planning = time-series forecasting triggered from inside a Seamless Planning model version, not a separate ML engine.
- SAP documents a named workflow: generating and integrating APL forecasts inside a Seamless Planning model.
- Same underlying engine as Smart Predict's time-series scenario (C341) — APL's Auto Time Series function (C344), not a planning-specific algorithm.
- Difference from Smart Predict time series is the delivery surface: Predictive Planning writes into a planning version with audit trail, lock and approval workflow; Smart Predict writes to a dataset.
- The forecast call is, on the HANA-backed path, a server-side APL Auto Time Series execution against the planning measure's history.
- Result: a reviewable planning version (or version-category combination) a planner can adjust and then lock/submit through normal approval.
- Same seasonality data bar as Smart Predict time series (commonly ~2 full historical cycles) applies — a forecast on too little history is technically generated but not trustworthy.
- Planning-model design prerequisite: the target measure needs clean, continuous history modeled consistently inside a Seamless Planning model structure before Predictive Planning can produce useful output.
Terms used on this page
- Seamless Planning model
- The SAC planning-model type into which Predictive Planning writes back APL-generated forecasts as a reviewable version.
- Predictive Planning
- SAP's name for triggering an APL Auto Time Series forecast from inside a planning model, writing the result back as a plan version.
- Planning version
- A distinct, lockable, auditable set of plan numbers; a Predictive Planning forecast is delivered as one, following the same approval workflow as any manually entered version.
- Seasonality bar
- The minimum historical span (commonly ~2 full cycles) needed for a reliable time-series forecast; applies equally to Predictive Planning and Smart Predict time series.
Sources
Full card available to members. What the full card adds: the full decision framework · the common pitfalls and their fix · the cheat sheet · the facts worth quoting.