Methodology
Two rubrics. One inspectable-event layer. Claude builds the what; the human does the why.
Mid-market is two motions in a trench coat. Tech accounts buy on agent opportunity and white space. Industries accounts buy on knowledge-worker density and AI commitment. Both rubrics run in parallel against live public evidence.
Rubric · Mid-Market Tech
- 01Agent Opportunity
Surface area for agentic workflows — support, ops, research, code.
- 02Internal Transformation
Evidence the org is actively rewiring around AI internally.
- 03AI Commitment
Leadership posture, AI hiring velocity, product bets.
- 04Whitespace vs. Existing Spend
Room to grow beyond current Anthropic or LLM usage.
- 05Industry Fit & Velocity
Sector alignment with Anthropic ICP and rate of change.
Rubric · Industries
- 01Knowledge-Worker Density
Share of workforce doing reading, writing, analysis.
- 02Public AI Commitment
Measured via open roles, exec statements, press releases.
- 03Automatable Workflow Volume
Volume of repetitive knowledge work Claude could compress.
- 04Transformation Readiness
Data, tooling, change-management maturity to deploy.
- 05Sector AI Velocity
How fast peers in the sector are moving — competitive pressure.
Inspectable-events layer
Propensity says who to call. Inspectable events say whether the buyer is actually moving. Every event is measurable, binary, and described as something the buyer did — not something the seller did. Tri-state per account: Not Done Yet · Yes · No.
- — Co-edited the ROI model
- — Named the budget owner
- — Confirmed quarter for go-live
- — Texting with primary contact
- — Met counterpart on the buyer side
- — Shared internal Slack channel
- — Sandbox API key issued
- — Solutions review scheduled
- — MCP server stood up against staging
- — Named the internal champion
- — Identified the rollout team
- — First success metric agreed
Three cadences
Daily
A skill runs before every external meeting: account snapshot, last three signals, hypothesized use case, comparable.
Weekly
Movers, pipeline health, and the Monday rep queue — assembled from the diff engine, one button, copy-ready.
Quarterly
Overnight scoring of the full book against both rubrics, with reasoning trace, citations, use case, and comparable per account.
How to read the Book
- · Tier mapping: 80+ Priority, 62–79 Build, 46–61 Watch, <46 Park.
- · Confidence is "high" only when at least three dimensions cite live public evidence.
- · Why-Now is the single freshest public signal — careers, press, leadership, or product — that re-ranks the account this week.
- · Every dimension score is linkable back to the source URL it relied on. No black-box numbers.
Under the hood
The scoring engine, prompt library, and per-run inspection are exposed for audit. Nothing is a black box.
- · Skills — the composable operations that run against an account.
- · Prompts — every system prompt and skill template, version-controlled.
- · Runs — recent scoring runs with inputs, tokens, and cost.
- · Proof — throughput and cost math for scaling to a 4,000-account book.
- · Monday agenda — a leader's first day, executed against real skills.