Strategy Switch Decision Tree¶

A decision-support tool — instantiates Plateau Detection and Switching

A branching decision aid that turns a confirmed plateau into a structured choice of switch — reroute, redesign, expand capacity, change modality, resegment, or revise the objective — by matching the switch type to the plateau's cause.

Strategy Switch Decision Tree is the chooser. Once a plateau is confirmed and escalation stopped, a team faces the hardest question the archetype poses: switch to what? This tool's distinguishing feature is that it doesn't detect, stop, or enforce anything — it selects among kinds of switch by branching on the plateau's cause. Saturation of a channel points to rerouting; a structural bottleneck points to redesign; a capacity ceiling points to expansion; a mechanism mismatch points to a modality change; a wrong-audience plateau points to resegmentation; and a genuinely exhausted goal points to revising the objective. It holds these options as a live menu and matches the diagnosis to the response, so switching is a reasoned selection rather than a reach for whatever comes to mind.

Example¶

A city's road-safety program has spent three years on enforcement — more speed cameras, more patrols — to cut traffic fatalities, and the numbers have flattened despite rising enforcement. The plateau is confirmed; the question is what to do instead. The Strategy Switch Decision Tree branches on the cause. Are the remaining crashes concentrated where enforcement can reach? No — they cluster at a few intersections with dangerous geometry, which enforcement can't fix. That branch rules out "more enforcement" and "resegment the patrol map" and points to redesign the road (protected turns, narrowed lanes) as the matched switch, with public education held as a parallel bet from the option menu. The tool doesn't decide the whole policy, but it turns "enforcement plateaued, now what?" into a defensible selection tied to the actual failure cause — rather than simply doubling down on the lever that stopped working.

How it works¶

Branch on the plateau's cause. The tree's first splits ask why response flattened — saturation, bottleneck, capacity, mechanism mismatch, wrong segment, exhausted goal — because the cause determines the right family of switch.
Map each cause to a switch type. Every leaf names a kind of response (reroute, redesign, expand, change modality, resegment, revise objective), so the diagnosis produces a matched action rather than a generic "try something else."
Draw candidates from a live option menu. Concrete alternatives are kept ready under each branch, so the selection is from prepared options, not improvised on the spot.

Tuning parameters¶

Branch granularity — how many cause categories the tree distinguishes. Fine branching gives precise matches but is heavier to navigate; coarse branching is quick but can route different plateaus to the same generic switch.
Option-menu breadth — how many candidate switches sit under each leaf. A broad menu offers real choice but risks decision overload; a narrow one is decisive but may miss the best move.
Evidence gate per branch — how much diagnostic evidence a branch demands before it's taken. Strict gates prevent misrouting; loose ones let the tree run on assumption.
Reversibility weighting — whether the tree prefers cheap, reversible switches first. Favoring reversible moves buys learning at low cost; favoring decisive ones commits faster when the cause is clear.

When it helps, and when it misleads¶

Its strength is preventing the default failure of switching — reaching for a familiar move regardless of why the plateau formed. By forcing a cause diagnosis before naming a switch, it stops "our answer to everything is a redesign" and matches the response to the constraint.

Its failure mode is a mis-diagnosed root branch: take the wrong fork early and every downstream option is wrong, with the tree lending false rigor to a bad call. A tree can also ossify — its branches encode last year's causes and quietly exclude a switch nobody thought to add. And it is easy to run backwards, steering the branching toward the switch someone already favored. The discipline is to gate each branch on real evidence of the cause (often from a capacity-context read), keep the option menu current, and treat the tree's output as a candidate to validate, not a verdict.

How it implements the components¶

Strategy Switch Decision Tree fills the switch-selection subset — it chooses the kind of switch, matched to cause:

alternative_strategy — its core: it selects the specific alternate path (reroute, redesign, expand, change modality, resegment, revise objective) to pursue.
experiment_backlog — the live menu of candidate switches held ready under each branch, so selection draws on prepared options.
capacity_context_signal — the diagnostic input the branches read: whether the plateau is saturation, bottleneck, capacity, or mechanism mismatch, which determines the fork.

It does not detect the plateau, enforce the stop, or validate the chosen switch (marginal_response_metric, escalation_stop_rule, false_plateau_check → Diminishing Returns Detection, Escalation Stop Workflow, Controlled Experiment After Plateau); it selects the candidate they then act on and test.

Instantiates: Plateau Detection and Switching — it is the selection step that turns a confirmed plateau into a matched choice of switch.
Consumes: Diminishing Returns Detection confirms the plateau; a capacity-context read supplies the cause its branches split on.
Sibling mechanisms: Controlled Experiment After Plateau · Process Redesign After Plateau · Diminishing Returns Detection · Marginal Gain Dashboard · Escalation Stop Workflow · Plateau Review Cadence · Ad Fatigue Switching · Training Plateau Adjustment · Product Growth Plateau Response · Saturation-Aware Resource Allocation · Expert-Governed Modality Change

Notes¶

This tool selects a switch; it does not test it. Its output is a matched candidate that Controlled Experiment After Plateau should validate before full commitment — the tree narrows the field, the experiment confirms the pick.