Post-Acquisition

What Should Move to Shared Services and What Should Stay Local?

For an operating partner or shared-services lead: route one process at a time instead of centralizing a department by label.

Uli PrantzBuilds and operates all-agents
Published
Key takeaways

Key takeaways

  • Centralize a process only when its definition, evidence and service boundary are stable across participating companies.
  • Local variance is not automatically waste. Customer promises, licenses, safety conditions and accountable judgment may require local control.
  • Use more than two destinations: shared execution, shared platform with local execution, shared standard, coordinated exception service, or fully local work.
  • Pilot with a service contract, exit criteria and a fallback path. A new shared queue without authority and exception ownership only moves the bottleneck.
Move recurring work to shared services when participating companies agree on the definition, source evidence, permitted actions, exception service and finish condition. Keep execution local when the result depends on local relationships, physical conditions, jurisdiction-specific requirements or accountable judgment. Many processes belong between those poles: share the standard, platform or evidence preparation while a local owner controls the consequential decision.

Decide at process level, not department level

“Centralize finance” or “keep operations local” bundles unlike work. A weekly KPI pack, a disputed customer promise and a physical safety decision may sit in one department but need three different operating models.

The GSA shared-services model emphasizes common business and data standards, accountable governance and an escalation path for unresolved inconsistencies. The useful lesson for a small portfolio is structural: a provider needs a defined service and a customer needs a retained owner.

Score the process against five questions before choosing a destination.
QuestionShared signalLocal signal
DefinitionSame finish and field meaningMaterially different promise or obligation
EvidenceAccessible, structured sourcesContext lives in people or physical inspection
VarianceKnown, bounded exception classesNovel cases dominate
ConsequenceReversible preparation or routine effectSafety, legal, financial or relationship commitment
AuthorityShared role has explicit permissionAccountability remains with named local role

Use five operating routes instead of a binary answer

Each route changes a different layer of the process.
RouteWhat is sharedWhat stays local
Shared executionComplete routine path and exception serviceDemand signal and accountable service owner
Shared platformData structure, queue and evidenceExecution and decision authority
Shared standardDefinitions, minimum controls and output contractTool and operating path
Shared preparationCollection, validation and proposalApproval and consequential effect
Local executionPortfolio visibility and escalation onlyProcess, authority, relationships and recovery

Write the service contract before moving the queue

  1. 01Name the service eventTrigger
    Define what enters, who may submit it and when the service clock starts.
  2. 02Fix the evidence contractCode
    List required sources, freshness, identifiers, validation and how missing evidence is returned.
  3. 03Bound shared authorityHuman approval
    State what the provider may decide, propose, reject or never perform.
  4. 04Design exception ownershipHuman approval
    Every refusal, conflict and timeout needs a named local or shared owner plus an escalation clock.
  5. 05Test exit and fallbackCode
    Define acceptance, stop conditions, local fallback and how in-flight cases are reconciled if the pilot ends.

One weekly report moves; one customer commitment does not

Simulated portfolio routing decisionSimulated example data

Three Northline branches prepare the same weekly service pack. Definitions differ for callbacks and completed work, but every metric can be tied to a source record. The portfolio first agrees the definitions, preserves branch-level source fields and pilots shared preparation with local controller sign-off.

A disputed customer credit remains local. The shared team can assemble the job record and policy, but the branch service lead owns the relationship context and approval. The portfolio gets a visible exception and response clock rather than a centralized decision.

The example makes no staffing, savings, system or product claim.

Review the pilot as a service, not an organizational victory

  • Was the right case accepted, rejected or returned with a reason?
  • Did local and shared owners agree on the final state?
  • Which exceptions lacked authority or evidence?
  • Could a branch use the fallback without losing in-flight work?
  • Did standardization erase a local requirement that should remain explicit?

Use the shared-services readiness calculator to keep portability separate from local-dependence pressure. Neither score approves a move.

Limitations and when not to use this

  • This is a process-routing framework, not organizational, staffing, legal, financial, employment, tax, cybersecurity or regulatory advice.
  • Federal shared-services sources illustrate governance and standards; they are not a ready-made model for a small-business portfolio.
  • A common platform does not create common authority. Retain named accountable owners and specialist review for consequential work.
  • Northline is simulated. No savings, service level, connector or customer outcome is claimed.

Sources

  1. Federal Shared Services: Adoption Challenges Underscore the Need for Consistent LeadershipU.S. Government Accountability Office Accessed 9 August 2026
  2. Enterprise Shared Services Governance EcosystemU.S. General Services Administration Accessed 9 August 2026
  3. Contingency Planning Guide for Federal Information SystemsNIST Accessed 9 August 2026

Assess shared-services readiness

Compare portability with the need for local knowledge and accountable control.

Assess shared-services readiness
About the author

Uli Prantz

Builds and operates all-agents

Uli Prantz builds all-agents, the process-automation platform this site documents. He writes about the operational side of automating recurring business work: where deterministic code beats model judgment, where it does not, and where a human still has to approve.

Bring one process. We will scope it in 30 minutes.

You leave the call knowing whether it is a fit, what can become code and what still needs a person.

Book a discovery call