Purpose
Turn the "Building an AI-native revenue organization" guide into a working plan. The guide argues that returns grow with the access and trust the organization extends to Claude, and that whole-team adoption compounds beyond individual time savings. Details live in companion files:
- references/maturity-and-rollout.md — maturity ladder, setup decisions, phases, ROI, pitfalls, FAQ answers, cited customer results.
- references/sales-plugin-catalog.md — Sales plugin commands and skills, connectors, other Claude surfaces.
- templates/rollout-plan.md — fill-in plan and checklists.
When to use
- A revenue leader owns the department's AI program and needs a rollout plan.
- A pilot is running and someone must decide whether to scale.
- Leadership wants an ROI readout or a budget case for next year.
- Spend is rising and the team must decide between coaching and capping.
Instructions
Place the org on the maturity ladder. Ask what Claude can read and write today, how much of the work it is trusted to finish, and whether top performers' methods are written down as shared skills. Name the next level and what it needs (connections, trust, governance: who can publish skills, what data each workflow can reach, who owns results).
Make the setup decisions before any pilot date.
- Owner: RevOps is a strong default (owns CRM and pipeline/forecast reporting).
- Connect systems (CRM, call recording, email, Slack) so reps stop uploading files by hand.
- Name an IT owner (provisioning, SSO, admin consents) and agree provisioning dates first.
- Start the security review early: data boundaries, connector permission model, auditability, telemetry export.
- Pick one activity metric and one revenue metric, set a baseline, commit to readout dates.
- Set spend limits by org, group, and user; restrict costlier capabilities by role; track usage from day one; give champions more headroom.
- Pick two or three teams with motivated leads; provision plugins at the admin level.
Set up the team's toolkit. Provision the starter pack (see the catalog). Preload the Sales plugin settings file (name, quota, positioning, competitors) before first login so every rep has a working command on day one. Connectors inherit each rep's existing permissions.
Run the three phases.
- Setup: IT handles workspace and SSO while RevOps configures connectors and plugins. Choose champions with sales managers: two or three per department, or one per 25-50 users, with dedicated hours.
- Pilot: each pilot team runs two or three use cases. Champions demo on live accounts, staff office hours, and write the first skills. Track how many champion-built skills pilot teams use regularly — the best early signal for scaling.
- Scale: move pilot skills into the shared bundle; provision in waves (e.g., 25 → 150 →
everyone); bring Claude output into standing meetings (e.g.,
/forecastin the weekly call).
Measure ROI the way the guide recommends. Compare pilot vs non-pilot teams in the same quarter on pipeline per rep, cycle length, and win rate, rather than before/after. Sort returns into efficiency, expansion, and new capabilities. Scale when output persists past the novelty, quality checks hold, and pilot teams outperform on the chosen outcome metric. After a full quarter, set daily spend per rep next to results, and lead budget talks with expansion and new capabilities — not hours saved.
Read spend in context. High spend plus high output (often a champion) is the design working. High spend with little output → coach first; lower the cap only if coaching does not help.
Pre-empt the three pitfalls (details in the reference): pilot with no end date, seats scaled without champions, spend ignored until the invoice.
Use the template in templates/rollout-plan.md to produce the written plan. Do not invent numbers or customer claims beyond the reference.
Examples
- User: "Our 25-seat pilot has run for three quarters and everyone likes it." → Pilot without an end date. Put a scale-decision date on the sponsoring executive's calendar, name the decider, and agree what a "yes" requires.
- User: "We're going from 25 to 150 seats next month with our three champions." → Hold one champion per 25-50 users; ask incoming managers to name champions now so current champions can train them.
- User: "The CFO asked what we got for the spend." → Report pilot-vs-non-pilot pipeline per rep, cycle length, win rate; frame as efficiency, expansion, new capabilities; avoid a case built only on hours saved.
- User: "One rep is top of the spend table." → Check what they produced (briefs, opportunity updates, proposals). If low output, have a champion coach them on commands, skills, and connectors before cutting the cap.