Expansion Map Stage 2 Review
Crew Preliminary Expansion Map
This page renders the full non-canonical Stage 2 working packet for Crew, the AI dev-tool cost product path under research/crew/. It updates the same review page after approved scope YAML and stops before canonical Stage 3 writes.
Stage 2 stop: the working packet exists at research/crew/_working/preliminary-expansion-map-research.md. Canonical files remain proposed only until a later complete approval payload approves Stage 3.
The previous Stage 1 page was archived at docs/history/archive/2026-06-24/131616/alignment/expansion-map-crew.html before this page was replaced.
Research Scope Approved
approval recordApproval status: ready-for-agent-review. Required gates: complete. Unanswered required questions: none.
Approved source plan: use Crew repo artifacts, Trace sibling context for cross-sell, and current external validation for time-sensitive vendor/pricing claims.
Approved emphasis: shared-account Crew-to-Trace and Trace-to-Crew cross-sell mechanics first.
Approved format request: use more visual account-flow and expansion-trigger diagrams in Stage 2.
Approved file flow: create the Stage 2 working packet at research/crew/_working/preliminary-expansion-map-research.md; reserve research/crew/expansion-map.md and research/crew/expansion-map-interview.md for Stage 3.
In scope
Expansion triggers, account rollout, role growth, tier pressure, shared-account cross-sell, referral/advocacy paths, health thresholds, and source gaps.
Out of scope
Trace canonical expansion map, pricing page, billing implementation, sales playbook, and canonical Crew output writes.
Non-canonical packet
Stage 2 synthesizes research for review only. Stage 3 must consume a later approval payload before final artifacts are written.
Section feedback
Executive Findings
top findings
1. Shared-account scope expansion comes first
Claim: Crew should expand by turning one retained account into a broader CalcLLM account that can hold both Crew and Trace.
Evidence: Crew and Trace docs repeatedly state that both products share one product-line account and cross-sell bidirectionally.
Confidence: High for account model; medium for exact trigger timing.
2. Crew-only expansion starts when visibility becomes governance
Claim: A retained Crew customer expands when dashboard visibility becomes showback, budgets, alerts, forecasting, team dashboards, chargeback, or budget enforcement.
Evidence: Crew journey-map tasks center on monthly budget review, anomaly investigation, team budgets, and ROI justification.
Confidence: High for workflow; medium for packaging.
3. Role expansion is as important as seat expansion
Claim: Expansion adds stakeholders before it simply adds seats: platform champion, team leads, Finance/FP&A, VP Eng or CTO, Security/Compliance, and Procurement.
Evidence: Crew ICP names these roles and the journey-map describes finance-triggered attribution and executive approval.
Confidence: High for role map; medium for timing.
4. Vendor budget controls are validation and risk
Claim: GitHub, Claude, Cursor, and Devin/Windsurf controls validate budget volatility, but reduce Crew's ability to sell single-vendor budget alerts.
Evidence: Current docs and pricing pages show AI credits, spend limits, usage analytics, admin dashboards, SSO, and enterprise controls.
Confidence: High for GitHub/Claude; medium for Cursor/Devin granularity.
5. Cross-sell should fire from a missing answer
Claim: Crew-to-Trace should trigger when production-cost attribution appears; Trace-to-Crew should trigger when team AI-tool spend becomes visible.
Evidence: Crew and Trace journey maps both name the opposite product as an expansion gap.
Confidence: Medium-high.
6. Advocacy follows proof artifacts
Claim: Referral and public-story asks should follow CFO answers, budget surprise prevention, cost-per-PR proof, or tool consolidation wins.
Evidence: Crew journey-map advocacy paths include platform talks, peer referrals, engineering blogs, and FinOps case studies.
Confidence: Medium.
Section feedback
Visual Account-Flow Diagrams
visual review
Diagram 1 - Shared CalcLLM Account Expansion Flow
| Node | Meaning | Expansion Role |
|---|---|---|
| Shared CalcLLM account | One account holds Crew and Trace. | Prevents forced tier graduation and supports bidirectional product addition. |
| Crew | AI dev-tool cost module. | Expands to Trace when production-cost questions appear. |
| Trace | Production LLM cost module. | Expands to Crew when team AI-tool spend questions appear. |
| Shared surfaces | Identity, teams, cost centers, billing, Slack reports. | Make cross-sell feel like scope expansion inside the same account. |
Diagram 2 - Crew Expansion Trigger Map
| Trigger | Expansion Motion | Outcome |
|---|---|---|
| More engineers or teams | Add teams, cost centers, and team-lead dashboards. | Broader Crew account usage. |
| New AI tool adopted | Add provider integration and cross-tool normalization. | Higher aggregation value. |
| Finance asks who spent this | Add showback, chargeback, and FP&A exports. | Governance expansion. |
| Usage volatility | Add forecasting, guardrails, anomaly workflows. | Budget-control expansion. |
| Production-cost question | Add Trace module. | Shared product-line account expansion. |
Diagram 3 - Role Rollout Path
| Role | Expansion Artifact | Why It Matters |
|---|---|---|
| Platform Eng Lead | Cross-tool attribution dashboard. | Champion and primary user. |
| Admin | Provider, team, and cost-center mapping. | Maintains data quality. |
| Team leads | Team-specific dashboards and alerts. | Creates role expansion beyond the champion. |
| Finance/FP&A | Showback, chargeback, and budget exports. | Turns visibility into governance. |
| VP Eng or CTO | ROI and budget justification artifact. | Economic buyer and renewal owner. |
| Security/Compliance | SSO, RBAC, audit logs, retention controls. | Higher-tier readiness gate. |
| Procurement | Invoice, PO, custom terms. | Enterprise account process. |
Section feedback
Evidence Matrix
claims mapped to evidence| Claim | Evidence | Inference | Confidence | Assumption Status | Decision Impact |
|---|---|---|---|---|---|
| Shared-account cross-sell is primary. | Crew and Trace journey, positioning, and ICP docs all describe one shared account and bidirectional cross-sell. | Expansion should be a domain-scope addition, not maturity graduation. | High | Evidence-backed for strategy; mechanics unbuilt. | Put cross-sell first in Stage 3. |
| Crew-only expansion begins with governance workflows. | Crew journey-map tasks include monthly budget review, anomaly investigation, team budgets, and ROI justification. | A retained dashboard becomes expansion when roles and workflows multiply. | High | Packaging provisional. | Separate retention from expansion. |
| Role expansion matters. | Crew ICP names champion, buyer, finance influencer, security blocker, and secondary users. | Account rollout should map roles and artifacts. | High | Evidence-backed. | Include role table in canonical map. |
| GitHub validates budget-control urgency but is also risk. | GitHub docs show AI credits, pooled allowances, budgets at multiple levels, blocking behavior, and no automatic fallback to cheaper models. | Vendor-native controls can solve single-vendor budget problems. | High | Current as of 2026-06-24. | Crew must lead with cross-tool value. |
| Claude validates per-developer cost volatility. | Claude Code docs describe token charges, $13 active day average, $150-250 monthly average, usage commands, spend limits, and reporting. | Teams need pilots, baselines, and spend tracking. | High | Current as of 2026-06-24. | Use Claude as trigger evidence, not sole wedge. |
| Cursor and Devin/Windsurf support enterprise tier surfaces. | Current pricing pages list team billing, usage analytics, admin dashboards, SSO, enterprise controls, quotas, and extra usage. | Buyers expect admin and governance controls in AI coding tools. | Medium | Public pages lack deep API details. | Keep enterprise controls as likely tier drivers. |
| Cohrint narrows the empty-space claim. | Current Cohrint page claims routing, dashboards, Claude Code/Copilot/Cursor support, budget enforcement, and savings-based pricing. | Crew should avoid claiming no AI coding cost player exists. | Medium-high | Vendor claims not customer-verified. | Say no validated cross-tool read-only governance layer, not no competitors. |
| Advocacy follows proof. | Crew journey-map lists platform talks, peer referrals, engineering blogs, and FinOps case studies. | Ask for advocacy after a proof artifact exists. | Medium | No customer feedback yet. | Add advocacy thresholds, not generic prompts. |
Section feedback
Working Packet Review
complete rendered packetExpansion Triggers
| Trigger | Signal | Expansion Motion | Evidence | Confidence |
|---|---|---|---|---|
| Engineering headcount growth | 50 engineers becomes 100-150; more teams need cost-center mapping. | Add teams, team dashboards, cost centers, user-level views, Slack alerts. | Crew journey map lists headcount growth and team-level rollup. | High |
| New AI dev-tool adoption | Account adds Cursor, Claude Code, Copilot, Devin/Windsurf, or another coding agent. | Add provider integration and cross-tool normalization. | Crew positioning says aggregation across vendors is the wedge. | High |
| Budget volatility under usage-based plans | Shared pool burns early; power users consume more than planned; budget caps become necessary. | Add forecasting, anomaly workflows, budget guardrails, executive reporting. | GitHub and Claude current docs validate usage volatility and spend limits. | High |
| Showback becomes chargeback | Finance asks for allocation by team/cost center; teams dispute numbers. | Add FP&A export, audit trail, allocation rules, role-specific dashboards. | Crew journey and ICP name finance's "who spent this?" question. | Medium-high |
| Production-cost attribution question | Platform lead or CTO asks why production LLM bill is 3x expected. | Add Trace to same account. | Crew and Trace journey maps both name cross-sell as a gap. | Medium-high |
| Trace account scales engineering team | Startup CTO using Trace asks what engineers spend across AI coding tools. | Add Crew to same account. | Trace journey and positioning name Trace-to-Crew trigger. | Medium |
| Enterprise readiness | SSO, RBAC, audit logs, SCIM, invoice/PO, support, security review. | Upgrade tier or sales-assisted motion. | Crew ICP and current Cursor/Claude/Devin pricing pages expose enterprise controls. | Medium |
| ROI proof | Cost-per-PR, budget surprise prevented, tool consolidation win. | Referral, case study, platform-community talk, FinOps story. | Crew journey-map advocacy paths. | Medium |
Upgrade And Seat Growth Paths
| Path | Baseline State | Expansion State | Roles Added | Product Surface | Caveat |
|---|---|---|---|---|---|
| More engineers | One platform lead reviews account spend. | Multiple teams and cost centers review their own spend. | Team leads, engineering managers. | Team dashboards, cost-center filters, Slack digests. | No monetization artifact defines seat-based versus usage-based packaging. |
| More tools | One or two providers connected. | Three or more AI coding tools normalized. | Tool owners, platform admins. | Provider connectors, normalized usage and cost views. | Per-developer granularity depends on vendor API detail. |
| Governance workflow | Monthly visibility. | Showback, chargeback, budget caps, allocation reports. | Finance/FP&A, VP Eng, team leads. | FP&A export, allocation rules, budget alerts, audit trail. | Chargeback may be too heavy pre-PMF. |
| Higher assurance | Self-serve admin. | Enterprise controls. | Security, compliance, procurement. | SSO/SAML, SCIM, RBAC, audit logs, retention, invoice/PO. | Defer full enterprise sales until post-PMF unless pulled by customer. |
| Product-line account | Crew only. | Crew plus Trace. | CTO, production platform owner, product/feature owners. | Shared account navigation, billing, identity/team taxonomy. | Cross-sell requires a specific production-cost question. |
| Advocacy | Private ROI proof. | Public or peer proof. | Champion, comms/exec reviewer. | Case-study export, referral prompt, community-talk support. | Needs permission and evidence quality. |
Team Or Account Rollout
| Step | Account State | Primary User Job | Expansion Test | Health Signal | Risk Signal |
|---|---|---|---|---|---|
| 1. Single champion | Platform lead connects first provider. | Answer who spent what. | Can Crew reveal an unexpected per-tool or per-user fact? | First dashboard viewed in first session. | No cross-tool reveal within 5 minutes. |
| 2. Team mapping | Teams and cost centers assigned. | Make spend legible by team. | Does the champion invite team leads or add cost centers? | 80%+ spend attributed. | 20%+ spend remains shared or ambiguous. |
| 3. Governance ritual | Monthly review becomes recurring. | Explain budget movement to finance. | Does finance ask for recurring exports? | Monthly report generated twice. | Dashboard viewed only at renewal or crisis. |
| 4. Budget controls | Forecasts and alerts guide behavior. | Prevent surprise spend or justify productive spend. | Does account configure thresholds? | Alert acted on within one business day. | Hard caps block productive work or get disabled. |
| 5. Product-line scope | Production-cost or dev-tool-cost question crosses domains. | Add Trace or Crew inside the same account. | Does an unanswered question belong to sibling product? | Cross-sell preview clicked or requested. | Cross-sell shown before current module proves value. |
| 6. Executive proof | VP Eng/CTO uses ROI narrative. | Defend AI tooling investment. | Does buyer use board/CFO-ready artifact? | Export shared with leadership. | Leadership sees only raw spend. |
| 7. Advocacy | Champion has a credible win. | Refer, review, or speak publicly. | Is there a public-safe proof point? | Referral or case-study permission. | Ask arrives before proof. |
Referral And Advocacy Paths
| Advocacy Path | Trigger | Proof Artifact | Best Ask | Confidence |
|---|---|---|---|---|
| Peer referral | Platform lead can answer the finance question quickly. | Before/after attribution screenshot or sanitized table. | Intro to another platform lead. | Medium |
| Engineering blog post | Tool consolidation or budget surprise prevention. | Cost-per-PR or vendor-normalized spending story. | Co-authored technical post. | Medium |
| PlatformCon/KubeCon talk | Governance workflow becomes repeatable. | Showback-to-chargeback operating model. | Conference talk support. | Medium-low |
| FinOps case study | Finance uses Crew in monthly budget review. | Allocation accuracy, forecast accuracy, saved or avoided spend. | Case study or webinar. | Medium |
| Product review | Startup CTO gets a fast surprise insight. | Self-serve first-value story. | Product Hunt/G2/dev-tool community review. | Medium-low |
Risks, Constraints, And Product Gaps
| Risk Or Gap | Impact | Mitigation Or Follow-Up |
|---|---|---|
| Native vendor controls catch up. | Weakens budget controls as standalone wedge. | Anchor Crew on cross-tool aggregation, team attribution, outcome linkage, and shared-account cross-sell. |
| Vendor branding shifts from Windsurf to Devin Desktop. | Crew docs may sound stale if they keep "Windsurf" without current naming. | Use "Devin/Windsurf" until canonical docs are refreshed. |
| Per-developer attribution gaps. | Account health and chargeback may be unreliable. | Add confidence/coverage indicators and progressive integration options. |
| Cross-sell feels forced. | Premature upsell can damage trust. | Trigger cross-sell only from a concrete missing answer. |
| No scoped monetization artifact. | Tier and packaging recommendations are provisional. | Run monetization or packaging research before final pricing decisions. |
| Lifecycle metrics missing. | Health thresholds need instrumentation definitions. | Run lifecycle metrics after expansion map confirmation. |
| Customer proof missing. | Advocacy and referral paths remain inferred. | Gather customer interviews or simulated discovery notes. |
| Cohrint competes more directly than older repo framing suggests. | Crew's near-empty claim should be narrowed. | Differentiate on read-only governance, attribution, and Trace cross-sell. |
Preliminary Interview Log
| Question | Expected Answer To Validate | Evidence Needed |
|---|---|---|
| When does a monthly Crew budget review become expansion? | When it adds team leads, Finance/FP&A, budget thresholds, cost-center allocation, or chargeback. | Lifecycle metrics or customer discovery. |
| What should trigger Crew-to-Trace cross-sell? | A production-cost attribution question Crew cannot answer. | Onboarding flow, dashboard event, or discovery interview. |
| What should trigger Trace-to-Crew cross-sell? | A Trace account's CTO scales engineering headcount and asks about AI coding-tool spend. | Trace onboarding and expansion map. |
| Which roles must be mapped before chargeback is credible? | Platform admin, team lead, Finance/FP&A, economic buyer, and dispute owner. | Chargeback workflow spec. |
| What makes an advocacy ask appropriate? | Budget surprise prevented, ROI proof, cost-per-PR narrative, or tool-consolidation win. | Customer feedback and permission. |
| Do GitHub AI Credits reduce Crew demand or create urgency? | Both: they validate volatility but weaken single-vendor budget-alert differentiation. | Current docs plus customer discovery. |
| Does Cohrint directly compete? | Yes, as optimization/routing and dashboards for AI coding spend. | Competitive review and customer perception. |
Section feedback
Alternatives And Lower-Confidence Findings
rejected or provisional| Alternative | Verdict | Reason |
|---|---|---|
| Old Solo-to-Platform graduation ladder | Rejected | Superseded by product-domain split and shared-account model. |
| Crew-only expansion with Trace excluded | Rejected for Stage 2 | User explicitly approved cross-sell-first emphasis. |
| Budget controls as primary expansion wedge | De-emphasized | GitHub and other vendors now have native controls; Crew needs cross-tool normalization and outcome linkage. |
| Enterprise-first expansion | Lower confidence | Enterprise controls matter, but repo positions near-term motion as PLG plus sales-assisted growth-stage. |
| Cohrint as non-factor | Rejected | Current Cohrint page makes it a closer competitor than older near-empty language implies. |
| Pricing thresholds and plan names | Deferred | No scoped monetization artifact exists, so Stage 2 should not invent final packaging. |
Section feedback
Source Coverage And Gaps
repo and current marketRepository Sources
| Source | Coverage | Notes |
|---|---|---|
research/crew/journey-map.md | High | Expansion triggers, task journeys, cross-product gap, advocacy paths. |
research/crew/icp.md | High | Buyer/user roles, pain map, willingness-to-pay signals, account model. |
research/crew/positioning.md | High | Shared account, cross-sell, anti-positions, product/sales/pricing implications. |
research/crew/competitive-analysis.md | Medium-high | Competitor framing, GTM lessons, pricing expectations, native-convergence risk. |
research/trace/* | Medium | Sibling context only for cross-sell mechanics. |
| Crew retention, metrics, monetization, GTM, feedback, specs | Missing | Affected claims remain provisional. |
Current External Sources Verified 2026-06-24
| Source | Relevant Observation | Use In Packet |
|---|---|---|
| GitHub Copilot licenses | Copilot licensing combines seats and AI credits; budgets can manage AI-credit consumption. | Validates AI-credit budget governance. |
| GitHub org/enterprise usage-based billing | Business and Enterprise include pooled AI credits; promotional amounts run June 1 to September 1, 2026; budgets can halt user access and additional spend. | Makes GitHub controls Crew's main competitive risk. |
| GitHub budget controls | GitHub recommends guardrails because heavy users or automated agent sessions can consume a disproportionate shared pool early. | Validates pooled-budget trigger. |
| GitHub models and pricing | Token consumption converts into AI credits at 1 credit = $0.01; model and tokens determine cost. | Supports model/usage normalization need. |
| Claude Code costs | Claude Code charges by token consumption; enterprise usage averages around $13 per developer active day and $150-250 per month; admins can set limits and view reporting. | Validates per-developer cost volatility. |
| Claude pricing | Pro includes Claude Code; Team and Enterprise expose central billing, SSO, usage analytics, spend controls, audit logs, and usage-at-API-rates pricing. | Supports higher-tier control expectations. |
| Cursor pricing | Individual is $20/month, Teams is $40/user/month, usage-based continuation, admin metrics, SSO, enterprise pooled usage, SCIM, audit logs, and controls. | Supports tier and readiness triggers. |
| Devin/Windsurf pricing | Windsurf pricing redirects to Devin; Devin Desktop lists Pro, Max, Teams, quotas, admin analytics, and enterprise controls. | Flags naming drift and quota/admin-surface expansion. |
| Cohrint | Cohrint claims savings routing for Claude Code, Copilot, and Cursor; supports MCP, CLI, SDK, OTel, dashboards, budget enforcement, and savings-based pricing. | Updates closest-competitor risk. |
| DX pricing | DX sells modular engineering intelligence and AI measurement with developer-license pricing and annual contracts. | Confirms broad engineering intelligence, not direct cost-input wedge. |
| Worklytics pricing | Worklytics Business starts at $2,500/month and includes integrations and AI adoption insights. | Confirms adoption/work-pattern analytics, not spend attribution. |
Section feedback
Assumptions And Confidence Register
provisional claims| Assumption | Status | Confidence | What Would Change It |
|---|---|---|---|
| Crew and Trace will share one account, identity layer, and billing relationship. | Repo-supported | High | Product architecture decision changes. |
| Shared-account cross-sell should be first in the canonical map. | User-approved and repo-supported | High | User later narrows scope to Crew-only. |
| Chargeback is a genuine expansion motion. | Provisional | Medium | Monetization/lifecycle work shows chargeback must be baseline activation. |
| Cross-tool aggregation remains differentiated despite vendor controls. | Evidence-backed but competitive | Medium-high | Vendors ship robust cross-vendor import/normalization or teams consolidate to one vendor. |
| "Windsurf" should be refreshed to "Devin/Windsurf" in future Crew docs. | Current-source observation | Medium | Product docs clarify separate branding or market usage keeps Windsurf dominant. |
| Per-developer attribution is technically feasible from read-only sources. | Known gap | Medium-low | Vendor APIs expose insufficient user-level data. |
| Advocacy can become a repeatable expansion loop. | Inferred | Medium | Customer interviews fail to show willingness to refer or publish. |
Section feedback
Glossary Additions
optional Stage 3 appendThese terms should be appended only during Stage 3, and only if approved here. Default target is a scoped Crew glossary if created later.
| Term | Definition | Source | Category | Decision |
|---|---|---|---|---|
| Shared-account cross-sell | Expansion in which one CalcLLM account adds the sibling product module, Crew or Trace, because the user's cost-governance scope crosses domains. | expansion-map | business | |
| Showback | A finance or platform reporting workflow that shows teams their attributed spend without directly charging their budget. | expansion-map | business | |
| Chargeback | A finance workflow that allocates attributed AI-tool cost back to the consuming team, product, or cost center. | expansion-map | business | |
| GitHub AI Credits | GitHub Copilot usage unit that converts model token consumption into credits, with 1 AI credit equal to $0.01 USD in GitHub's current billing docs. | expansion-map | technical | |
| Account-health threshold | A usage, coverage, stakeholder, or workflow signal that triggers expansion, risk intervention, referral, or advocacy action. | expansion-map | workflow |
Section feedback
Proposed Canonical Artifacts And File Changes
Stage 3 proposal| Path | Stage | Proposed Action |
|---|---|---|
research/crew/_working/preliminary-expansion-map-research.md | Stage 2 | Active non-canonical working packet. Archive and remove during Stage 3. |
research/crew/expansion-map.md | Stage 3 | Write approved canonical expansion map only after final approval. |
research/crew/expansion-map-interview.md | Stage 3 | Write approved interview/evidence artifact only after final approval. |
alignment/expansion-map-crew.html | Stage 3 | Convert to confirmed status with approval record after canonical artifacts are written. |
research/crew/glossary.md or parent glossary | Stage 3 optional | Append only glossary terms approved above. |
docs/history/archive/YYYY-MM-DD/HHMMSS/... | Stage 3 | Archive the working packet before removal. |
Section feedback
User Format Preferences
required gateConfirm whether this Stage 2 review format matches the requested visual density, grouping, labels, and evidence treatment.
Section feedback
Final Artifact Approval
required gateThis gate controls whether Stage 3 may archive the working packet, write canonical artifacts, and convert this page to confirmed status.
Section feedback
Compile Responses
Answer one or more gates, approve glossary terms, or select section feedback, then compile YAML. Stage 3 can proceed only when every required gate is answered and no blocking option or negative feedback remains unresolved.