• The pricing engine is a black box — 187 price rules, 9 calculator plugin scripts and 44 discount schedules interact in an order nobody can fully explain; 52 active rules have not fired in 12 months and pricing behaviour is verified by "quote it and see".
• Amendments and renewals run on contract-centric logic RCA doesn't have — 3,150 amendments a year flow through SBQQ__Subscription__c and contract-based amendment quotes; RCA's asset-centric model (Asset / AssetAction / AssetStatePeriod) is a re-architecture, not a mapping.
• Revenue recognition lives in legacy Revenue Schedules — 82% of revenue is ratable and recognised through constructs RCB deliberately replaces with GL Journal Entry Rules; finance's month-end process is built on objects that will not exist post-migration.
• Two-thirds of the catalog is noise — 858 of 1,246 products are inactive or unquoted in 24 months, 31 of 96 bundles have zero recent usage, and 12% of products have no PricebookEntry at all. Migrating this as-is would pour six years of drift into a clean platform.
• Bundle nesting exceeds RCA's sweet spot — 14 bundles nest three to four levels deep with option-constraint chains; RCA's flatter ProductRelatedComponent model needs these redesigned, not transplanted.
• Quote templates don't come along — 18 CPQ quote templates (5 active) have no native RCA equivalent; document generation must be re-implemented on Document Builder or an equivalent before cutover, or reps lose the ability to send quotes on day one.
• The org's real logic is in code — 61 Apex classes and triggers reference SBQQ objects, including twin-field sync, a quote calculator plugin and renewal uplift logic. Each needs an explicit retire / refactor / retain decision against RCA's Expression Set and CML framework.

• 1,246 Product2 records: 388 active, 858 inactive or with no quote line in 24 months. 149 products (12%) have no PricebookEntry in any pricebook — unquotable by definition.
• Subscription behaviour is driven by SBQQ__SubscriptionType__c (5 values in use, including two legacy variants of "Renewable"), SBQQ__SubscriptionPricing__c and SBQQ__SubscriptionTerm__c defaults of 12/24/36 months.
• Portfolio split by billing behaviour: 61% recurring subscription, 24% one-time (services, onboarding), 15% usage-rated (API calls, consumption tiers).
• 7 pricebooks beyond Standard — regional (ANZ, EMEA, Americas), a partner book, and three "promo" books last updated over two years ago. Multi-currency active across AUD, USD, EUR, GBP with 214 currency-specific PricebookEntry overrides.

• Run a catalog rationalisation sprint before any RCA build: confirm the 388-product active set with product management, archive the rest with a documented restore path.
• Define the three Product Selling Models and their term options as the first RCA build artefact — everything downstream (pricing, bundles, assets, billing) hangs off them.
• Rebuild the pricebook estate as four books (three regional + partner) with currency coverage validated against the 214 existing overrides.
• Move edition/tier variants (Starter / Pro / Enterprise duplicated as separate SKUs) to attribute-driven products where the variant count justifies it.

• 96 bundle products carrying 412 SBQQ__ProductFeature__c records and 1,830 SBQQ__ProductOption__c records; selection methods split between Click (78%) and Add (22%).
• 14 bundles nest 3–4 levels deep (bundle-in-bundle-in-bundle), all in the enterprise platform family; the deepest chains exist to reuse option sets rather than to reflect how the product is actually sold.
• 65 bundles were quoted in the trailing 12 months; 31 have zero usage. Average options selected per bundle instance: 4.2 of an average 11 available.
• 214 SBQQ__ProductRule__c records: 118 Validation, 61 Selection, 35 Alert. 38 rules (18%) have not fired in 12 months; 9 pairs of rules overlap or contradict on the same products. Option visibility logic additionally lives in 47 SBQQ__ConfigurationRule__c records and 3 Apex validation classes — three mechanisms for one job.

Rule firing inferred from quote error logs and rule-evaluation history. Dormant rules migrate as documentation, not configuration.

• Retire the 31 zero-usage bundles and 38 dormant rules before mapping begins — they are 30% of the apparent configuration surface and 0% of revenue.
• Run a bundle redesign workshop per deep-nested product with product management: the target is RCA's flatter component model reflecting how the product is sold today.
• Consolidate all option-visibility logic into the constraint-rule layer during rebuild; do not recreate the three-mechanism split in RCA.
• Resolve the 9 contradictory rule pairs with sales ops before migration — RCA's constraint engine will surface conflicts CPQ silently tolerated.

• 187 SBQQ__PriceRule__c records (evaluation events: 41 on-load, 88 on-change, 58 on-save) with 512 Price Conditions and 391 Price Actions. Only 74 rules fired on quotes in the trailing 12 months; 52 active rules never fired; 61 are inactive but unretired.
• One SBQQ__CalculatorPlugin__c (~1,900 lines of JavaScript) implements currency-rounding, margin-floor enforcement and ramp pricing. It is undocumented and its author has left the business.
• 44 Discount Schedules: 29 volume-tiered, 11 term-based, 4 block. 12 schedules apply to products with zero recent quote activity.
• Discount distribution across 168,300 quote lines: 46% schedule-driven, 31% manual (SBQQ__AdditionalDiscount__c or SBQQ__SpecialPrice__c), 23% list price. Manual discounts cluster at 10/15/20% — anchor-point discounting rather than governed exceptions.

Only the green bar migrates. The pricing procedure rebuild targets 74 rules plus the calculator plugin's three behaviours — not 187 rules.

• Specify pricing behaviour from evidence, not code: derive the calculator plugin's actual effects from 12 months of quote outcomes, then implement that specification in CML.
• Build the golden-quote regression pack (top 200 quote shapes by revenue) and run it against CPQ and RCA in parallel — pricing parity is the cutover gate.
• Retire the 113 non-firing/inactive rules and 12 dead schedules now, in CPQ, so parallel-run comparisons are clean.
• Use the migration to shift the 10/15/20% manual anchor discounts into governed adjustment tiers with approval thresholds preserved.

• 14,800 quotes/year averaging 11.4 lines; 41 custom fields on SBQQ__Quote__c and 58 on SBQQ__QuoteLine__c; Quote Line Groups used on 34% of quotes (multi-year ramps and multi-product phases).
• 63 Twin Field mappings sync quote data to Opportunity/OpportunityLineItem; sync failures logged on 2.1% of quotes, mostly on high-line-count ramp deals.
• Quote status distribution (12 mo): 48% Won-linked, 24% Lost-linked, 22% abandoned drafts, 6% in flight. 18 quote templates exist; 5 are actively used, with conditional sections for region, currency and T&C variants; DocuSign is the e-signature layer.
• 4 approval processes route on discount %, deal size, payment terms and non-standard T&Cs: 6,900 submissions/year, 2.6-day average approval, 9% rejection. Orders: ~640/month generated from won quotes, auto-contracted on activation.

A 22% draft-abandonment rate is a UX signal: reps use quote creation as a pricing calculator. The RCA quote experience should give them a cheaper way to explore price.

• Rebuild only the 5 active templates on Document Builder and get DocuSign touchpoints working before UAT — a rep who can't send a quote on day one defines the migration's reputation.
• Validate the 63 twin-field mappings as a field-parity checklist against RCA's native sync; retire the automation, keep the reporting fields.
• Re-implement approval entry criteria against RCA quote fields, folding in the tightened discount governance from Domain 03.
• Design the RCA quote screen around the "pricing calculator" behaviour — a price-preview path that doesn't create an abandoned quote record.

• 8,412 active SBQQ__Subscription__c records across 2,140 contracts. Term distribution: 58% 12-month, 14% 24-month, 9% 36-month, 19% evergreen (auto-renewing monthly).
• 3,150 amendment quotes/year: 54% upsell (seats/modules), 21% term extension or co-term, 12% downsell, 13% administrative (entity, billing details). Co-termination is used on 41% of multi-product accounts.
• Renewal quote auto-generation succeeds on 87% of eligible contracts; the 13% failure cohort is dominated by evergreen subscriptions and contracts with manually edited end dates. Renewal uplift (default 5%, overridable) is applied by a custom Apex routine, not standard fields.
• Mid-term amendment proration is standard CPQ day-based proration plus one custom Apex override for a legacy customer cohort on calendar-month proration.

• Treat subscription-to-asset conversion as a formal data migration workstream: extraction, transformation rules, three reconciliation passes (counts, MRR by account, term dates) and a signed-off cutover snapshot.
• Resolve the evergreen design first — it shapes selling models (Domain 01), renewal processing, and billing schedules (Domain 07).
• Clean the 13% renewal-failure cohort in CPQ before migration; migrating broken end-dates makes them RCA's problem.
• Sunset the calendar-month proration exception at migration if commercially possible — one proration model, one set of test cases.

• 61 Apex classes/triggers referencing SBQQ__* objects: twin-field sync helpers, the calculator plugin, renewal uplift, quote-line rollups to Opportunity, contract activation hooks, and integration serialisers. Error logging shows 3 classes account for 84% of CPQ-related Apex exceptions.
• Outbound integrations: ERP middleware consuming Order/Invoice payloads keyed to legacy Billing objects; a data warehouse extract reading SBQQ__Subscription__c nightly; a CS platform webhook on contract activation. One dormant marketing integration has failed silently for 14 months.
• Security: 6 permission sets grant CPQ object access; 240 quote-creating users in the trailing 90 days; sharing on Quote follows Opportunity. Page layouts and the quote line editor carry 9 UX customisations reps rely on daily (column sets, custom actions).

• Freeze the disposition register as a governed artefact — every one of the 61 components has an owner, a target, and a test before cutover.
• Run the ERP remapping as a joint workshop with the middleware owner early in Phase 1; integration lead times, not Salesforce build, set this domain's critical path.
• Parallel-run the data warehouse extract on both schemas for one full cycle so finance reporting never goes dark.
• Rebuild the 9 quote-screen UX behaviours in the RCA quote experience before UAT — they are what "feels the same" means to a rep.

• ~4,100 invoices/month from scheduled batch runs; 3.2% require manual correction (wrong billing period, mid-cycle amendment timing, currency rounding). Invoice line grouping is customised to group by product family for enterprise accounts.
• Payments: 71% gateway-processed (tokenised cards / direct debit), 29% invoice-and-remit. Gateway success rate 96.8%; failures concentrate in expired tokens on evergreen subscriptions.
• Revenue recognition: 82% ratable over term via legacy Revenue Schedules, 18% point-in-time (services, one-time). 940 credit notes/year, top reasons: mid-term downgrades, billing-period errors, goodwill. Avalara handles tax across four jurisdictions with 312 exemption certificates on file.
• Usage-based billing: usage summaries rated monthly for the consumption portfolio (15% of products); rating exceptions on 1.9% of cycles, mostly late-arriving usage files. GL: middleware posts summarised journals to the ERP nightly; ~$18K/month average unreconciled variance absorbed at month-end, plus a recurring manual journal for deferred revenue true-ups.

• Run a full parallel billing month: RCB generates shadow invoices and journals alongside production Billing, reconciled line-by-line for the top 50 accounts and in aggregate for the rest — this is the program's hardest gate and its best insurance.
• Design the GL Journal Entry Rule set with finance in the room, mapping every current recognition treatment and the deferred-revenue true-up it should eliminate.
• Pre-migrate payment token re-authentication for the evergreen cohort — the current failure hotspot — before cutover, not during.
• Keep invoice posting manual for the first cycle post-cutover, then automate once reconciliation is clean two cycles running.

This is a healthier migration than most CPQ estates present. The business's revenue motion is genuinely simpler than its configuration — three selling models, one real pricing engine a third the size of the one on disk, and amendment patterns that map naturally onto RCA's asset model. The hard parts are concentrated and known: an undocumented pricing plugin, a contract-to-asset re-architecture, and a revenue recognition change that finance must own rather than receive.

The discipline that decides the outcome is subtraction. Every dormant product, rule and bundle retired in week one is configuration that never needs building, testing, or explaining in RCA.