Economic-model assumptions sidecar

Sourcing and confidence labels for the economic table in consulting-partner-brief.md. This is the back-pocket document for the technical follow-up after the first partner conversation — when the partner's CFO or practice leader asks "where did you get these numbers?"

The point of this document is transparency, not argument. Every cell is open to revision based on the partner's view of actual engagement economics. They have run more of these engagements than we have. The model exists to seed the conversation, not to win it.

---

Confidence labels (used throughout)

• validated — backed by published data, ARX standard contract terms, or industry-norm ranges that have remained stable for 5+ years
• benchmark-comparable-company — derived from comparable companies' published deal sizes (Anaplan, Workday, ServiceNow, Snowflake, etc.) or from third-party research (ALM Intelligence, Gartner, IDC, Forrester)
• estimate-needs-validation — ARX hypothesis. The partner will know better and we expect them to revise

---

Footnote source table

[¹] "Initial strategy + design: $4–8M today"

Confidence: benchmark-comparable-company. Source: ALM Intelligence Vault Consulting Firms 2024 publishes engagement-size ranges for the top-10 strategy consulting firms. AI / digital transformation strategy engagements at the F500 cluster sit at $3–10M for the initial strategy + design phase, with McKinsey and BCG at the upper end and Deloitte / Accenture at the lower end of the strategy-only band (Deloitte / Accenture monetize on the implementation side). Validation ask: confirm against [Firm]'s last 5 AI-strategy engagements at F500 customers — does $4–8M match the median?

[²] "Initial strategy + design: $5–10M with ARX"

Confidence: estimate-needs-validation. Source: ARX hypothesis. The reasoning: with ARX as the infrastructure, the strategy phase produces a deployable plan with real cost / productivity numbers grounded in the customer's actual systems (via the Engagement Canvas), not a slide-based hypothetical. That justifies a ~25% fee uplift because the deliverable is materially better than the slide-based equivalent. Range $5–10M is the [¹] range × 1.25. Validation ask: would [Firm]'s practice leader price a strategy engagement that produces a board-ready deployable plan (vs. a slide deck) at a 20–30% premium?

[³] "Integration / deployment delivery: $0–3M today (handed off)"

Confidence: estimate-needs-validation. Source: ARX hypothesis based on the typical Big-3 strategy-vs-implementation split. McKinsey and BCG core typically hand off implementation to a designated SI (Accenture, Deloitte, Cognizant, Infosys); the strategy firm retains a Program Management Office (PMO) role at $0–3M / yr. Deloitte / Accenture / Capgemini retain implementation work in-house — for them, today's number is much higher (the hand-off problem is moot for SIs). Validation ask: for [Firm], what fraction of an AI-transformation engagement do you keep in-house vs. hand off? This number changes shape per firm.

[⁴] "Integration / deployment delivery: $15–35M with ARX"

Confidence: benchmark-comparable-company. Source: Workday / ServiceNow / Anaplan F500 implementation typical fee ranges per ALM Intelligence + Gartner 2023–2024 — Workday HCM at-scale F500 implementations cluster at $20–60M; ServiceNow Now Platform enterprise transformations at $15–40M; Anaplan FP&A transformations at $10–25M. ARX is comparable in delivery scope (multi-platform, multi-year, requires ongoing configuration) but lower in upfront customization (less custom code needed because the agents are configured, not built from scratch). $15–35M is the conservative midpoint of the comparable bands. Validation ask: against [Firm]'s ServiceNow / Workday / Salesforce implementation deal sizes — does this range track?

[⁵] "Outcome-tied success fees: $5–15M / yr for 3 yrs"

Confidence: estimate-needs-validation. Source: ARX hypothesis. The reasoning: ARX produces a defensible measurement of productivity gain per cohort (from app/core/usage_meter.py + workforce dashboards in the build plan); that measurement is what success fees can be tied to. Industry precedent for outcome-tied consulting fees: BCG's results-based pricing pilots, Accenture's "outcome economics" engagements, McKinsey's "value-based fees" experiments — typical structures cap success fees at 20–40% of measured savings. At a $1.9B Cisco-shape annual recovery, a 1–3% success fee per consulting firm = $20–60M / yr; we've cut that by 4× to land at the $5–15M / yr range as a conservative anchor. Validation ask: what is [Firm]'s current largest success-fee structure, and what is the typical % of measured savings? This number is the most variable cell in the model.

[⁶] "Post-deployment optimization retainer: $3–8M / yr"

Confidence: benchmark-comparable-company. Source: Accenture Operations and Deloitte Managed Services published retainer rates for enterprise platform optimization. Typical F500 ServiceNow / Workday post-deployment managed-services contracts run $2–10M / yr depending on platform complexity and customer in-house capability. ARX's optimization retainer is shaped similarly — ongoing performance monitoring, drift triage, cohort tuning, evidence-package preparation, regulatory-update response. Range $3–8M / yr is the midpoint of comparable. Validation ask: does [Firm] currently run managed-services retainers in this range for ServiceNow / Workday / Salesforce post-deployment? If so, ARX's retainer slots in alongside, not replacing.

[⁷] "ARX revenue-share: 20% of ARX ARR"

Confidence: validated. Source: ARX channel-program standard rate. ISV channel industry norm is 15–25% rev share for partner-sourced deals (Salesforce AppExchange partner economics, Microsoft Cloud Solution Provider rates, AWS Partner Network APN rebate tiers). 20% is the midpoint. Validation ask: none — ARX commits to 20% as the baseline; can be tuned higher (up to 25%) for first-mover or named-account-exclusivity terms.

[⁸] "Total per F500 engagement, 3-year basis: $48–117M"

Confidence: derived (carries forward the confidence of the highest-uncertainty inputs, which are [⁵] success fees and [⁴] integration delivery). Source: arithmetic sum of [²] + [⁴] + [⁵] × 3 + [⁶] × 3 + [⁷] × 3 with the ARX ARR assumed at $200K–$1M / yr. Range covers conservative-low-end ($48M) and aggressive-high-end ($117M) scenarios. Validation ask: even at the conservative low end, this is an 8–10× expansion vs. today's strategy-only engagement. Does the order of magnitude track with [Firm]'s view of where AI transformation engagements are heading?

[⁹] "Three engagements per year: $130–350M incremental"

Confidence: estimate-needs-validation. Source: ARX assumption that a Big-3 firm with active F500 AI-transformation pipeline could close 3 ARX-anchored deals per year through this channel after the channel motion is established (post-pilot). Range = [⁸] × 3 with conservative-3-deals at the low end and full-pipeline-3-deals at the high end. Validation ask: how many AI transformation engagements does [Firm]'s practice currently close per year at F500 customers? If it's 12, three of them being ARX-anchored is 25% — plausible. If it's 4, three is 75% — too aggressive.

---

Methodology note

This model uses a 3-year accounting horizon because that's the typical contract length for both ARX (3-year SaaS contracts standard) and the equivalent SI managed-services retainer. The model does not include:

• Year 4+ revenue. Existing engagements continue to bill the success fees + retainer + ARX rev share. Realistic LTV per F500 account is 5–7 years; the 3-year number is conservative.
• Pipeline development effects. Once [Firm] has 2–3 ARX-anchored engagements as references, the pitch motion becomes materially easier and the close rate improves. The model assumes flat conversion, which is conservative.
• Industry-pack add-ons. When ARX ships the financial-services / healthcare / public-sector industry packs (per the build plan in /root/.claude/plans/what-tools-in-arx-glimmering-tiger.md), each pack creates additional engagement scope per F500 account in that industry. Not modeled here.
• Cross-engagement learnings. Once [Firm] has portfolio-level data across ARX engagements (per the Firm Portfolio analytics in the build plan), the firm has differentiated benchmark IP. Not modeled here.

The model also does not include any deal-cost line items (legal review, ARX co-sell SE time, etc.) on the assumption that they're sub-rounding-error against the totals shown.

Where this is wrong

The single largest unknown in this model is the success-fee structure ([⁵]). If [Firm]'s typical success-fee ratio is dramatically lower (e.g., 5–10% of measured savings vs. the 20–40% industry-precedent we cited), the success-fee row collapses by 3–4× and the total comes down to roughly $35–80M per engagement — still a ~7× expansion vs. today, but a different headline number. We expect this to be the cell most aggressively renegotiated.

The second-largest unknown is whether [Firm] retains implementation in-house or hands off ([³] / [⁴]). For McKinsey OPE / BCG core, the hand-off pattern likely persists even with ARX — McKinsey doesn't typically run multi-year implementation. In that case, the row [⁴] does not accrue to McKinsey directly; it accrues to whichever SI partners with McKinsey. ARX's economic argument to McKinsey then collapses to [²] + [⁵] + [⁶] + [⁷] = ~$25–55M per engagement — still a 3–5× expansion, still meaningful, but a different conversation.

The variants in partners/variants/{firm}.md adjust this table per firm to reflect these structural differences.