Planner Guide

The Planner stage is the bridge between a retained MMM run and scenario planning inside the IDE.

What Planner Depends On

A retained run is planner-capable only when the IDE can prove that the run has the retained evidence needed for the Python planner services to load it.

In practice, this means the run must have:

a retained run_manifest.json
fit-stage model data needed by the planner
a planner-loadable config path

Native Planner Surface

The native planner client is intentionally narrower than the historical Dash planner, but it is still a real scenario-planning workspace.

The planner surface is organized as a linked tab family:

Planner Overview
Scenario Builder
Curves & Allocation
Comparison
History

What The IDE Owns

The IDE owns:

planner navigation
workspace selection
native table and chart presentation
shell continuity with the retained MMM run

What Python Abacus Owns

Python abacus.scenario_planner owns:

planner truth
scenario evaluation
workspace persistence
cached results
planner exports

v1 Scenario Scope

The native planner aligns to the public planner scenario types currently exposed by Abacus:

current
manual allocation
fixed-budget optimized

More advanced goal-seeking modes should only be documented after the Python planner surface exposes them explicitly.