Creator: Alexander M. Dinola. Date of first use / creation: August 03, 2026. Description: The Decision Design Workbench is the licensed software instrument that implements the Decision Design frameworks as an inspectable, deterministic deal-scoring application, published at defensibility.design as the companion to The Decision Design of Selling. The same inputs always produce the same verdict, every threshold is visible and editable in the interface, and any number displayed traces back to a row the user entered. The engine reads five components: Status Quo Burden, Governed Change Exposure, Certainty, Portability and Decision Weight. Each carries its own gate. Gates resolve before the composite Defensibility Index (zero to one), so a component below its gate caps the recommendation regardless of the overall score, and an eighteen-check integrity audit sits above both, where a critical failure halts the verdict. The recommendation ladder has six rungs: Not Yet Inspected, Stop / Rebuild Evidence, Hold / Close Critical Gaps, Reset / Narrow the Move, Conditional, and Advance. Every input takes conservative, base and upside values, and one control re-reads the whole instrument under each. Five tables feed the score: Tax Ledger, Switch Risks, Certainty, Stakeholders and Buyer Actions. Each row records an evidence class, source, owner, confidence level and review date. A provenance badge is computed from those fields directly, so relabeling a row cannot improve it. Eight surfaces present the case: Dashboard, Case, Field Cards, Analysis, Timeline, Audit, Control and Portfolio. Analysis runs a correlated simulation that reads entered ranges as coherent scenarios, identifies which single open item would change the recommendation, projects an evidence horizon showing when the verdict goes stale, and withholds outcome calibration until twenty of the user's own deals have closed. Timeline snapshots each material change in verdict and flags the pattern where the score climbs while buyer-owned Decision Weight stays flat, naming the date it began. Portfolio reads across saved deals, treats the same gate failing on five or more as a defect in the selling motion rather than in any account, and declines to compare deals measured under different settings. Nine deeper models sit alongside the core: Cost of Delay, Network, Proof NPV, Weight Calibrator, Pilot, Reframe, Portability, Guardrail and Expansion Gate. Each is complete, carries its own integrity checks, and answers one question a difficult deal turns on. None alters the Defensibility Index. The Dashboard instead compares each model's conclusion against the core tables and reports disagreement. Facts entered in more than one place are cross-checked, all dates run on a single as-of date, and a model not run is reported as not run rather than as agreeing. Supporting content: The product keeps six rules. An empty form reads "not yet inspected" rather than failing. Missing data is labeled "not yet named" and never filled with decoration. Gates precede the index so a strong average cannot cover a failed requirement. Provenance is computed from evidence fields, never from names. Costs are never reduced by a certainty factor while benefits are. A risk that cannot honestly carry a number is carried in words instead. A saved deal generates a read-only link containing the verdict and an optional buyer-inspectable consequence case with seller-side notes removed, printable to PDF. The instrument is licensed at $290 for one year as a single non-recurring purchase, with a thirty-day code to the full product printed in the back of the book.