Workshop: Opportunity audit

First workshop, first artifact. The opportunity inventory is the raw material for every later decision: a single table of every credible candidate, scored consistently, owned specifically. Run it for your real organization if you can (strongly preferred) or for Alder using the case details — the worked example below shows the standard.

1. 1Run the three passes (chain walk, money leaks, floor harvest) for your scope. Target 15–25 candidates: fewer means the passes were shallow; many more means you're listing tasks, not opportunities. Use the capability-matching prompt to pressure-test each.
2. 2Fill the seven columns per candidate (see Alder's excerpt): the one-line opportunity, the capability verb, the value lever (cost / revenue / risk / speed / quality) with the line item it touches, the volume (runs per month), data readiness (green: exists and accessible / yellow: exists, messy / red: doesn't exist), error tolerance (draft-reviewed / triaged / must-be-right), and a named owner — a person, not a department.
3. 3Do the owner check for real. Fifteen minutes with each named owner: 'would this help, would your team use it, what did I get wrong?' Candidates whose owners shrug get marked — a shrug now is a stalled pilot later. (Expect two surprises: an owner who kills a favorite, and an owner who upgrades one you almost cut.)
4. 4Red-flag but keep the ineligible. Candidates that fail data readiness or error tolerance stay in the inventory, flagged, with what would unblock them ('needs 12 months of appointment data'; 'needs the quality gate from a data-foundations effort'). Today's red flags are next year's pipeline — and the flags themselves are input to your data-investment case.
5. 5Skim for themes before Module 2. Alder's inventory clustered around documents-in-motion (extract/classify everywhere) and promises-to-customers (predict ETA, predict profitability). Clusters hint at platform choices — one document-AI capability serving six candidates beats six point solutions, a thought Module 3 will finish.

OPPORTUNITY                CAP.     LEVER (line item)      VOL/MO  DATA  ERR-TOL     OWNER
exception-email triage     classify cost+speed (svc credits) 3,000  grn   triaged     R.Diaz, CS
quote-tender extraction    extract  speed+rev (win rate)       400  grn   draft-rev   M.Chen, Sales
predictive ETA, top-20     predict  rev-retention (churn)    9,000  yel   triaged     J.Okafor, Ops
contract-risk summarizer   generate risk (legal spend)          25  grn   draft-rev   L.Park, Legal
...
flagged red: dock-detention predictor (appointment data doesn't
exist yet) - unblock: capture appointment times in the TMS
(transportation management system) starting Q3