Your acquisition cost rises. The first proposal is to change the bids.
Before approving it, ask what happened after the click.
In MIZ OKI Decision Studio, a fictional business called Aster Supply faces that question. Acquisition cost has increased while the auction has barely moved. Checkout completion has deteriorated, and mobile checkout has become slower.
The leading explanation is a checkout problem. The screen calls that explanation supported and marks causal proof pending.
That distinction shapes the next move: hold the bid change and propose a controlled checkout repair.
The Studio is an interactive, synthetic demonstration. It has no connected customer data and does not dispatch production actions. Its purpose is to let you inspect the path from a commercial problem to a decision.
Start in Signal. The evidence includes observations that support the checkout explanation and observations that challenge other hypotheses. Creative fatigue and auction inflation remain visible as alternatives. You can remove a comparison source and see what evidence the decision would then lack.
The displayed support scores describe the fixture. They are not probabilities that a cause is true.
Next, open Proof. A separate synthetic experiment illustrates a treatment-versus-control comparison and an uncertainty interval. The page also states what the example cannot supply: real assignment checks, power, consent, contamination controls, and independent reconciliation.
Opening an experiment example does not qualify the current proposal. The evidence gate remains a separate requirement.
In Media, compare the predefined allocation choices with leaving the current mix unchanged. Change the assumed contribution rate or inspect the downside case. The comparison shows how the choice depends on the economics.
These are scenario calculations, not a trained optimizer's recommendations or measured incremental returns. The separate SKU waterfall has its own assumptions. Remove a required fulfillment cost and the result is excluded rather than completed with zero.
The Intent room examines another kind of evidence: an expiring forecast from a fictional consented session. It is labeled Preview · in development. Its deterministic replay is not a trained model, and its short expiry timer illustrates a behavior rather than specifying production retention.
Finally, open Growth Control. The current proposal must satisfy identity and scope, policy alignment, sufficient context, and hard constraints. Missing causal qualification produces a named reason to withhold. The controls let you explore stale economics, absent configuration, liquidity limits, and missing rollback.
Even simulated approval remains local to the demonstration. It cannot grant real authority.
The Outcomes view brings the record together. It separates the modeled difference from measured business impact and preserves the current disposition. Session state resets on reload, so the interface offers an export of the decision passport.
Follow the journey with a colleague who owns a different part of the business. Ask them to identify the leading explanation, the unproven assumption, the economic tradeoff, and the missing authorization.
If you can explain all four, the demonstration has done its job. The next question is whether the same discipline improves a recurring decision in your operation.
Open Decision Studio, or use All Studio parts to choose a room.