Hypothesis Ledger
Operating Record
Working document · yours to extendLive walkthrough included
RetainerCadence ·Continuous · weekly batch
Continuous creative testing on a documented hypothesis.
A weekly-batched testing operation. Hypothesis library, kill criteria at one thousand impressions, statistical reads attached to every winner. Closes each cycle with a Hypothesis Ledger entry.
Inside the Dynamic Optimization layer of the Dynamic Ad operating system. Run as a wedge, or with the full stack.
Buyer trigger
Slot · The Volume-Without-Read Optimization OperatorSignal · Active
The Volume-Without-Read Optimization OperatorThe creative team is producing assets, the platforms are absorbing them, and nobody can name what the account has actually learned. Variants ship without hypotheses. Winners are declared on weekly dashboards before the read clears the statistical bar.
Signals firing the engagement
Hundreds of assets in rotation, no hypothesis log, no documented kill criteria, no statistical read on declared winners
A flagship variant has been called a winner three weeks running and the CPA is back at baseline
The team cannot answer which angle, format, or audience layer is actually compounding into the next batch
Hypothesis Ledger. The compounding record at every weekly batch.
A continuously updated ledger of every test the creative engine is running. Hypothesis in, statistical read out, winner or kill on record.
How the operation runs.
Operating access · continuous cadence · operator-approved
Phase 01 · Intake
Scope and access
Current creative program inventoried: volume in rotation, active angles, existing hypothesis documentation (or absence of it), and platform assignment. Account model confirmed against spend and channel mix. Read access provisioned across paid-media platforms. Angle library and existing creative backlog reviewed to establish the starting state of the hypothesis ledger before calibration.
Phase 02 · Calibration
Hypothesis bar and kill threshold
Statistical kill threshold calibrated against account-average CTR and primary conversion rate. Hypothesis schema agreed: angle, format, audience layer, success metric, and minimum sample per variant documented before any creative ships under the Signal Loop. Kill-criteria policy signed. No variant enters rotation without a filed hypothesis from this point forward.
Phase 03 · Operation
Weekly batch, weekly read
Every variant ships against a filed hypothesis, a named kill threshold, and a minimum sample requirement. Kill criteria apply at one thousand impressions against account-average CTR; variants that miss are paused and logged, not declared inconclusive. Winner reads require full statistical clearance: thirty percent lift on the primary metric at the declared sample size, signed before the result posts to the Hypothesis Ledger.
Phase 04 · Readout
Cycle close and ledger update
Cycle closes with Hypothesis Ledger updated: winners posted with hypothesis, statistical read, and survival check at thirty days; killed variants logged with residual and cause of death. Angle library compounded: next batch sequenced against what has already been learned, not from a blank brief. The operator inherits a library that grows each cycle, not a dashboard that resets.
Questions buyers ask before scoping Creative Testing Loop.
Does this include creative production?
No. The Signal Loop runs the testing layer: hypotheses, kill criteria, statistical reads. Production is owned by your in-house team or paired with Dynamic Ad Engineering.
What is the variant ceiling?
Documented per active angle in the SOW so volume scales with budget without breaking unit economics.
What statistical bar declares a winner?
Calibrated to your account at intake. The default is a thirty percent lift on the primary metric over account-average baseline at a minimum sample of one thousand impressions per variant.
How does the angle library compound?
Every winner enters the library with its hypothesis, its statistical read, and its survival rate. The next batch is sequenced against the library, not from a swipe file.
Scope the Creative Testing Loop. Hypothesis Ledger running.
Continuous operationCompounding loopHypothesis Ledger live
Alongside the retainerDynamic Lead OS.Stack the testing loop into the full operation. Production, reallocation, and analytics on one ledger.Other engagements under Dynamic Optimization
Each is sold standalone or stacks with the others into a single foundational build.