Incentive Floor Testing¶
Assessment — instantiates Minimum Effective Intervention
Tests lower incentive sizes or frequencies to identify the smallest reliable incentive before larger rewards create cost, dependency, crowd-out, or gaming.
Incentive Floor Testing runs a controlled, stepped experiment across reward levels to find the smallest reward — in size or frequency — that still reliably shifts behavior, before a larger reward starts manufacturing its own pathologies. Its one defining idea is that the dose is a reward magnitude whose distinctive danger is behavioral distortion: past the point of sufficiency, more money or more points does not buy more behavior change but buys dependency, crowd-out of intrinsic motivation, and gaming. The mechanism is built to read those distortion signals directly against the behavioral response, and to bank the resulting dose-response cells as evidence rather than fold them into a single verdict.
Example¶
A coffee chain's app currently gives a free drink every eight purchases, and finance suspects the reward is bigger than it needs to be. Incentive Floor Testing sets up descending test cells: a control at every-eight, and treatment groups at every-ten, every-twelve, and a points variant that lets customers feel progress between rewards. Over a season it measures repeat-visit rate in each cell — the behavioral response — and watches two distortion signals: gaming, where customers split one order into several small transactions to farm stamps, and crowd-out, where regulars who used to come for the coffee now report coming only when a reward is near. The smallest reward whose repeat-visit rate stays within tolerance of the control is the floor; the every-twelve cell holds, every-fourteen does not. The team records all four cells with their responses and distortion readings so next season's test starts from evidence, not from the old every-eight habit. The reward shrinks, margin recovers, and the behavior it was buying is still there.
How it works¶
- Lay out descending cells. Define reward levels stepping below the status quo and assign participants across them, ideally with a holdout control.
- Read behavior and distortion together. At each level, measure the behavioral response and the reward-specific side effects — gaming, dependency, crowd-out — because a reward can hit its behavior target while quietly corroding it.
- Pick the lowest cell that clears the bar. The floor is the smallest reward whose response holds within tolerance of the control and whose distortion signals stay acceptable.
- Archive the cells. Keep every level's result as a dose-response record so the floor is auditable and re-usable, not a one-off guess.
Tuning parameters¶
- Cell spacing — how far apart the tested reward levels sit. Tight spacing locates the floor precisely but needs more sample.
- Sample and power — how many participants per cell. Too few and a noisy cell masquerades as proof the low reward works.
- Distortion-signal set — which pathologies are watched (gaming, crowd-out, dependency). Omit one and the "cheaper" reward may be silently worse.
- Response tolerance — how close to the control's behavior a lower reward must stay to qualify.
- Test duration — how long each cell runs, because dependency and crowd-out often surface only after weeks, not days.
When it helps, and when it misleads¶
Its strength is that it prices behavior change empirically and exposes the perverse economics of over-rewarding: a big incentive that "works" may be buying gaming and eroding the intrinsic motivation that would have carried the behavior for free — the motivation-crowding effect.[n1]
Its failure mode is time. A reward level that looks sufficient during the test can collapse the moment it is trimmed or withdrawn if it has bred dependency the short window never revealed, so a floor found in four weeks may be a mirage. A related misuse is treating one favorable cell as proof and skipping the control. The discipline that keeps it honest is to pre-register the response tolerance and a minimum test duration long enough for dependency to show, and to read the distortion signals as first-class outcomes rather than footnotes to the headline behavior number.
How it implements the components¶
Incentive Floor Testing fills the empirical floor-finding slots — the smallest reliable reward, the distortions that bound it, and the evidence trail:
minimum_effective_input— its product: the smallest reward size/frequency that reliably holds the behavior.side_effect_signal— the reward-specific pathologies (gaming, dependency, crowd-out) tracked at each level.calibration_evidence_record— the archived dose-response cells that make the chosen floor auditable and reusable.response_metric— the behavioral-change measure read in each cell against a control.
It does not define a target social outcome, set a persistence-weighted sufficiency bar, or hold a stronger policy in reserve (target_effect_definition, sufficiency_threshold, escalation_reserve) — that is Minimal Viable Policy Intensity Pilot; and it does not guard against normalizing under-resourcing, check burden equity, or route a step-down (underpowering_guardrail, equity_and_burden_check, de_escalation_path) — that is Staffing Floor Experiment. This mechanism is the reward-dose experiment.
Related¶
- Instantiates: Minimum Effective Intervention — the minimum-intensity discipline applied to reward magnitude.
- Sibling mechanisms: Minimal Viable Policy Intensity Pilot · Staffing Floor Experiment · Alert Sensitivity Floor Tuning · Minimal Effective Training Load · Proportional Enforcement Ladder · Least Intrusive Intervention Policy · Minimum Effective Dose Review
Editorial Notes¶
Form Classification¶
Form family: Experiment, Test & Rehearsal
Rationale: Incentive Floor Testing operates as a bounded trial, probe, simulation, or rehearsal that generates evidence from performance because it tests lower incentive sizes or frequencies to identify the smallest reliable incentive before larger rewards create cost, dependency, crowd-out, or gaming
Independent corroboration: The frozen evidence defines Incentive Floor Testing as 'Tests lower incentive sizes or frequencies to identify the smallest reliable incentive before larger rewards create cost, dependency, crowd-out, or gaming', so its operative form is Experiment, Test & Rehearsal.
Review outcome: Independent reviewer agreement; high confidence.
Origin Attribution¶
Primary origin: Behavioral Economics
Origin pattern: Cross-disciplinary synthesis
Present-day reach: Multi-domain
Rationale: Testing the smallest effective reward while guarding against motivation crowd-out is characteristic of behavioral economics.
Related originating lineages:
- Economics & Finance — Marginal incentive response and cost-benefit analysis provide the optimization frame.
- Psychology — Self-determination and intrinsic-motivation research supplies the crowding-out mechanism.
- Statistics & Experimental Design — Controlled stepped reward levels and dose-response estimation supply the test design.
Review resolution: Both reviewers independently assign behavioral_economics as the primary originating domain, so that shared primary is retained. Alternate domains are the union of reviewer-identified formative or independently originating lineages; later application settings alone are excluded. The final form materially composes methods or concepts from more than one formative domain. It has established independent use across several domains, but that does not make it domain-free. The encyclopedia entry makes that composition explicit.
Encyclopedia synthesis: The exact catalogued form synthesizes established practice rather than reproducing a single standard historical label.
Review outcome: Reconciled after independent review; high confidence.
Notes¶
[n1] Motivation crowding-out is the effect in which an external reward displaces the intrinsic motivation for a behavior, so that removing the reward can leave behavior lower than before it was introduced — the specific side effect that makes an oversized incentive dangerous rather than merely wasteful. ↩