Large-cap and upper-mid-market sponsors have quietly turned shared service centers (SSCs) and Centers of Excellence (CoEs) into one of the highest-ROI plays in the portfolio operations toolkit. KKR Capstone, Bain Capital Portfolio Group, Blackstone Portfolio Operations, Carlyle's Global Solutions team, and Advent's Portfolio Support Group now operate cross-portfolio platforms that serve 10-30 portfolio companies in finance, HR, IT, procurement, data engineering, and increasingly AI/ML. The economics are straightforward: a $40M revenue add-on company cannot afford a senior FP&A lead, a cybersecurity architect, or a Snowflake engineer. A platform spanning 18 portcos can — and amortize the cost at $80-150K per portco per year instead of $400-600K standalone.
This article walks through how to design, build, and run these platforms. We focus on the unit economics, the captive-vs-BPO-vs-hybrid decision, the technology stack that determines whether the SSC scales past three portcos, and the governance traps that have killed otherwise sound programs. The discussion connects to the 100-day integration playbook and the fractional talent operating model covered elsewhere in this guide.
The Economics: Why Cross-Portfolio Platforms Beat Standalone Functions
In a typical mid-market buyout (EV $250M-$1.5B), G&A runs 8-14% of revenue. Finance, HR, IT, and procurement together account for 60-70% of that G&A spend. When an operating partner consolidates these functions into a sponsor-level shared service — typically structured as a separate legal entity that bills portcos at cost-plus 5-8% — the blended unit economics improve through three levers: labor arbitrage (35-55% savings by relocating to Hyderabad, Krakow, Manila, or Monterrey), span-of-control expansion (one AR analyst now processes 4-6 portcos' invoices versus one), and automation overlay (UiPath bots, Workday, NetSuite, and Coupa eliminate 25-40% of remaining manual touches).
At a 6-8x EBITDA exit multiple, $2M of recurring G&A savings translates to $12-16M of enterprise value per portco. Across an 18-portco platform, that is $216-288M of created value before any revenue synergies, cross-selling, or commercial CoE work. This is why sponsors like Vista Equity Partners (through Vista Consulting Group) and Thoma Bravo have built 200+ person operating organizations dedicated to running these platforms across their software portfolios.
Functional Scope: What Belongs in Shared Services vs. CoEs
The cleanest design separates transactional shared services (volume-driven, process-standardizable) from Centers of Excellence (expertise-driven, project-based). Sponsors that conflate the two end up with helpdesks staffed by data scientists — expensive and unhappy. The split should be deliberate.
Shared service centers handle invoice processing (P2P), order-to-cash (O2C), record-to-report (R2R), payroll administration, benefits ops, IT service desk Tier 1/2, master data management, and travel & expense. Volumes drive economics: an SSC absorbing 200K invoices per year across 12 portcos hits the productivity curve. Coupa, Bill.com, Tipalti, and SAP Concur typically anchor the AP stack; ADP Workforce Now or Workday handles payroll across the portfolio.
Centers of Excellence sit one layer up. They house scarce, expensive talent that no single portco can justify: cybersecurity engineers (often supporting the work described in cross-portfolio cyber posture management), Snowflake/Databricks data engineers, AI/ML scientists, Salesforce CPQ specialists, SEO/performance marketing pods, pricing analytics teams, and FP&A modeling experts who parachute in for budgeting season. CoE pricing is typically project-based or capacity-based (e.g., 40 hours/month retainer) rather than per-transaction.
| Dimension | Shared Service Center | Center of Excellence |
|---|---|---|
| Primary driver | Transaction volume | Specialized expertise |
| Talent profile | Process operators, accountants, L1/L2 IT | Engineers, data scientists, domain specialists |
| Pricing model | Per-transaction or per-FTE supported | Project fee or capacity retainer |
| Typical location | Hyderabad, Manila, Krakow, Bogotá | Bangalore, Warsaw, Austin, Lisbon, Dublin |
| Tooling spine | NetSuite, Workday, Coupa, ServiceNow | Snowflake, Databricks, GitHub, Salesforce, Looker |
| Onboarding a portco | 6-12 weeks | 2-4 weeks for first engagement |
| Margin target (cost-plus) | 5-8% | 10-15% |
Operating Model: Captive, BPO, or Hybrid
Three structures dominate. The captive model — owned and operated by the sponsor — gives maximum control and captures all the arbitrage but requires real management bandwidth and 18-24 months to reach steady state. KKR Capstone's India captive in Gurgaon and Blackstone's Hyderabad-based BXAccess (built out in 2023-2024 to over 600 FTEs) are large-cap examples. The third-party BPO model, used heavily by mid-market sponsors, contracts with Genpact, WNS, EXL, Infosys BPM, or Accenture Operations to run the shared services on a managed services basis. Setup is faster (90-120 days), but margin leakage to the BPO is 12-20% and process customization is constrained.
The hybrid — a captive shell with BPO-supplied labor ("build-operate-transfer," or BOT) — has become the default for sponsors with $5-25B AUM. The sponsor signs a 3-5 year BOT contract with a partner like WNS or Quatrro, which stands up the center, hires the team, and transitions ownership in years 2-3. The sponsor gets time-to-value of a BPO with end-state economics of a captive. Hahn & Company, Stone Point Capital, and several Nordic sponsors have used BOT structures with success.
The Technology Stack That Makes Multi-Tenant SSCs Work
A SSC fails the moment it cannot cleanly separate one portco's data, workflows, and reporting from another's. Multi-tenancy is a technical problem with a vendor answer. The reference stack we deploy at mid-market sponsors looks like this:
ERP layer: NetSuite OneWorld (for portcos under $500M revenue) or Microsoft Dynamics 365 with multi-entity configuration. Larger portcos stay on their own SAP or Oracle instance, with the SSC operating those systems remotely via privileged access workstations. Workday Financial Management is increasingly chosen for software portcos with $250M+ revenue. HCM: Workday or ADP Workforce Now multi-EIN configurations. Procure-to-pay: Coupa or SAP Ariba for $100M+ revenue portcos; Bill.com or Tipalti below that line. IT service management: ServiceNow with multi-tenant domain separation, or Freshservice for smaller deployments.
Automation overlay: UiPath or Automation Anywhere bots for repetitive tasks (invoice OCR, bank reconciliation, payroll variance checks, AR cash application). A mature SSC runs 80-200 production bots eliminating 18,000-45,000 manual hours per year. AI document extraction from Hyperscience, Rossum, or AWS Textract has cut invoice processing time from 4.2 minutes per invoice (manual) to 22 seconds (straight-through) on AP volumes we benchmarked in 2024-2025.
Data and analytics: A central Snowflake or Databricks lakehouse ingests financial and operational data from every portco nightly. Fivetran or Airbyte handles the pipelines; dbt manages transformations; Looker, Power BI, or ThoughtSpot serves dashboards. This same backbone powers cross-portfolio benchmarking — letting the operating team show portco CFOs that their DSO of 58 days is 14 days worse than the portfolio median, with named comparables. See the data lakehouse architecture reference for design patterns.
Centers of Excellence Worth Building
Not every CoE pays back. The five that consistently deliver across our engagements:
The procurement CoE deserves a closer look because it is the fastest payback. A sponsor with 22 portcos collectively spending $180M on cloud, software, and indirect categories can typically negotiate 15-20% off list through consolidated commitments. We routinely see AWS EDP discounts move from 8-12% (standalone portco) to 22-32% (sponsor-aggregated commit), Microsoft 365 E5 from $57/user to $42/user, and Salesforce from $165/user/month to $125 at portfolio-pooled volume. The CoE itself costs $1.2-2.4M/year to run. Net savings of $25-45M against $2M cost makes this the single highest-IRR initiative in the operating partner playbook.
Implementation Sequence: 24-Month Build
Diagnostic across 5-8 anchor portcos to size addressable spend. Decide captive vs. BPO vs. BOT. Select location (typically India + one nearshore). Lock in ERP and automation tooling. Build the legal entity, transfer pricing model, and inter-company services agreement (ISA) templates.
Migrate AP, payroll admin, and IT helpdesk for the first 2-3 portcos. Hire core leadership (SSC head, finance ops lead, IT ops lead, transformation lead). Stand up Snowflake/Databricks and the central data layer. Deploy first 20-30 RPA bots.
Industrialize onboarding to a 6-week playbook. Add O2C, R2R, procurement ops. Launch the cybersecurity CoE and procurement CoE in parallel. Hit 60-65% of target run-rate savings.
Onboard remaining portcos and new acquisitions on Day 30 post-close. Launch AI/ML CoE and RevOps CoE. Begin charging back to portcos at cost-plus. Migration of BOT ownership if applicable.
Automation rate above 55% of transaction volume. New portcos onboard in 4-6 weeks. SSC becomes a diligence asset — buyers see lower run-rate cost structures and pay multiple expansion at exit.
Governance: The Things That Kill These Programs
Most SSC failures are not technology failures. They are governance failures. Three patterns recur:
Portco CFOs resist mandatory adoption. If the sponsor's investment committee does not write SSC participation into the management equity plan and the portco CEO's annual scorecard, two of every five portcos will quietly opt out, claiming "unique requirements." By month 12 the platform is sub-scale and unit economics never converge. The fix: the sponsor's operating partner sets adoption as a condition of access to portfolio resources, and the SSC publishes transparent benchmark data showing the portco's cost-per-transaction versus median.
Transfer pricing gets the IRS or HMRC's attention. An SSC billing $4M/year to 15 portcos in 6 jurisdictions is a transfer pricing audit waiting to happen. We have seen sponsors face $8-22M of retroactive assessments where the ISA, cost-plus methodology, and benchmarking studies were not documented to OECD Transfer Pricing Guidelines standards. Engage a Big 4 TP team during design — not after the first audit notice.
Service quality dips during onboarding. Portco controllers measure the SSC on whether last week's payables were processed cleanly, not on the 36-month vision. If onboarding waves cause spikes in error rates or DSO, internal political capital evaporates. Successful programs publish SLAs (e.g., AP cycle time ≤ 4 business days; payroll accuracy ≥ 99.7%; IT ticket resolution P1 ≤ 2 hours) and report on them weekly to portco CFOs and the operating partner.
Measuring Value Creation: What Actually Hits the LP Report
The SSC's value shows up in three places in the LP-facing value creation bridge. First, run-rate EBITDA expansion at each portco — typically 80-180 basis points of margin lift from G&A reduction alone. Second, working capital release — a well-run AR shared service compresses DSO by 6-12 days, freeing $3-9M of cash on a $200M revenue portco. Third, multiple expansion at exit — buyers pay higher multiples for portcos with clean financials, fast monthly close (target: 5 business days), real-time KPI dashboards, and SOC 2 Type II posture, all of which the SSC and CoE backbone delivers.
Sponsors that get this right turn a back-office cost-cutting initiative into a structural competitive advantage. The 2024-2026 cohort of PE platforms with mature SSC+CoE infrastructure are now onboarding new acquisitions in 30-45 days, running monthly close in five business days portfolio-wide, and presenting buyers with diligence packs that include three years of clean, comparable, AI-ready data. That alone has moved exit multiples by 0.5-1.0 turns of EBITDA in transactions we have advised on in the last 18 months — value that compounds with the deal sourcing and add-on identification capabilities covered later in this guide.