Audit growth incrementality experiment integrity

What it's for

Gives growth and finance leaders an auditable answer to 'did this channel create incremental contribution?' before a dashboard ROAS number reaches budget planning.

What you give it

Inputs split into evidence read from your connected systems, calibration your team owns, and numerical controls that affect precision but never the result's meaning.

{
  "bootstrap_draws": 200,
  "experiment_units": [
    {
      "assignment_probability": 0.5,
      "baseline_contribution": 100,
      "channel_id": "paid-search",
      "evidence_verified": true,
      "experiment_id": "paid-search-geo-holdout",
      "id": "search-geo-0",
      "incremental_spend": 10,
      "outcome_contribution": 120,
      "spillover_exposed": false,
      "treatment": true,
      "unit_id": "geo-0"
    },
    {
      "assignment_probability": 0.5,
      "baseline_contribution": 101,
      "channel_id": "paid-search",
      "evidence_verified": true,
      "experiment_id": "paid-search-geo-holdout",
      "id": "search-geo-1",
      "incremental_spend": 10,
      "outcome_contribution": 121,
      "spillover_exposed": false,
      "treatment": true,
      "unit_id": "geo-1"
    },
    {
      "assignment_probability": 0.5,
      "baseline_contribution": 102,
      "channel_id": "paid-search",
      "evidence_verified": true,
      "experiment_id": "paid-search-geo-holdout",
      "id": "search-geo-2",
      "incremental_spend": 10,
      "outcome_contribution": 122,
      "spillover_exposed": false,
      "treatment": true,
      "unit_id": "geo-2"
    },
    {
      "assignment_probability": 0.5,

What you get back

This is the actual output of running the example above — computed by the same function the platform calls, not an illustration.

{
  "configuration": {
    "bootstrap_draws": 200,
    "confidence_level": 0.9,
    "maximum_absolute_standardized_difference": 0.25,
    "maximum_spillover_fraction": 0.05,
    "minimum_units_per_arm": 10,
    "minimum_verified_fraction": 0.95,
    "seed": 7
  },
  "decision": "growth_incrementality_evidence_supported",
  "experiment_diagnostics": [
    {
      "average_treated_incremental_spend": 10,
      "baseline_adjusted_incremental_contribution": 20,
      "baseline_standardized_difference": 0,
      "channel_id": "paid-search",
      "confidence_interval": [
        20,
        20
      ],
      "control_units": 20,
      "experiment_id": "paid-search-geo-holdout",
      "failed_gates": [],
      "incremental_return_on_spend": 2,
      "spillover_fraction": 0,
      "treated_units": 20,
      "verified_fraction": 1
    }
  ],
  "guardrails": [
    "Aggregate randomized units only; this is not person-level customer or employee scoring.",
    "Last-click attribution, platform-reported conversions, and observational spend correlation do not establish incrementality.",
    "Baseline imbalance, interference, unverified outcomes, or missing control spend discipline requires repair or a different design."
  ],
  "method": "aggregate_randomized_growth_incrementality_did_ipw_bootstrap_audit",
  "summary": {
    "aggregate_unit_count": 40,
    "experiment_count": 1,
    "failed_experiment_count": 0,
    "supported_experiment_count": 1
  },
  "truncation": {
    "experiments_omitted": 0

How it works

Causal inference & experiment design — Separate what a change caused from what merely moved alongside it.

1. 1 Freeze aggregate experimental unit, channel, assignment, propensity, pre-period contribution, mature post-period contribution, incremental spend, spillover and evidence fields before outcomes are inspected.
2. 2 Reconstruct each experiment, reject duplicate units or mixed channels, require both randomized arms, calculate inverse-propensity weighted pre/post contrasts and subtract the baseline contrast to obtain a difference-in-differences incremental contribution estimate.
3. 3 Bootstrap whole aggregate units within arm, compute uncertainty and iROAS, then gate the claim on sample support, baseline standardized difference, interference, verified evidence and zero incremental control spend.

Before you trust it

Every tool in the catalog ships with the conditions under which its answer is meaningful — and the conditions under which it should abstain instead of guessing.

Calibrating it to your org

Calibration workflow

1. 1 Define the management decision, target outcome, aggregate unit, privacy boundary, cadence, and prediction/intervention horizon for this organization.
2. 2 Build a tenant-scoped historical cohort using only information available before each prediction or decision; preserve zero periods, censoring, assignment probabilities, and unresolved outcomes when the method requires them.
3. 3 Estimate statistical parameters on training history, but obtain costs, utilities, risk tolerance, practical-effect thresholds, capacity, and policy constraints from accountable decision owners.
4. 4 Validate on later time windows or held-out aggregate units at the deployment grain, against a simple baseline and the function-specific validation strategy.
5. 5 Deploy only if the returned decision clears evidence, overlap, calibration, robustness, and guardrail checks; warning, unsupported, schema-gap, and fallback decisions are abstentions.
6. 6 Monitor realized outcomes, data drift, coverage, and decision regret; recalibrate at a governed cadence or after a detected regime/definition change, never merely because a stakeholder dislikes the result.

Call it from your AI

You don't wire up 388 tools in your MCP client. The GitRevio MCP server exposes 18 tools, three of which let an agent search the catalog, read a tool's schema, and run it — so the assistant finds this one on its own.

gitrevio_capabilities_search
  { "q": "audit aggregate randomized growth experiments before" }
  → finds "audit_growth_incrementality_experiment_integrity"

gitrevio_capability_describe
  { "capability_id": "audit_growth_incrementality_experiment_integrity" }
  → returns the input schema and agent guidance shown on this page

gitrevio_capability_run
  { "capability_id": "audit_growth_incrementality_experiment_integrity", "arguments": { ... } }
  → returns the result shown above

Works in Claude Desktop, Claude Code, Cursor, Cline, Continue.dev, Goose and Aider. See the MCP server.

Related tools

See every tool in Causal evidence & experiments →