Trupeer Blog

Shared Service Model Explained: Structures, Savings, Setup

Shared Service Model Explained: Structures, Savings, Setup

Synthétiser

Table des matières

Créez de superbes vidéos produit et des documents avec l’IA

Commencez gratuitement

A shared service model consolidates recurring functions such as finance, HR, IT and procurement into one specialized center that serves several business units or countries. Run that center under a named global business services leader and the payback gets serious: about 55% of organizations with a global GBS leader cleared more than 20% average savings in 2025.

It's a Tuesday, and you're staring at three onboarding checklists from three countries. Same steps, different wording, different approval boxes, and a new hire in each location waiting on a form nobody agrees on. That's the moment shared services stops being a slide in a board deck and becomes a decision about who owns the work.

The pattern is consistent. Teams that get this right start with a narrow set of repeatable processes, name one owner, and measure quality as hard as they measure cost. Teams that get it wrong move headcount into a building, keep every local rule, and call it transformation.

The trade-off in one line: centralization buys consistency and costs you proximity. You get one standard process, and you lose the ability to walk down the hall and ask a controller why an invoice is stuck.

Table of Contents

What Does a Shared Service Model Cover

A shared service model pulls recurring corporate work out of individual business units and runs it from one specialized center that serves several units or countries. Think of a single payroll office covering every branch, instead of each branch keeping its own payroll clerk.

The analogy holds because the underlying work doesn't change. Someone still validates the invoice, still checks the bank details, still runs the month-end close. What changes is who does it, under which standard, and against which service level. I treat that as an operating decision rather than a cost program, because once you centralize a process you also centralize the rules, and the rules are the hard part.

Here's where the work usually lands.

Function

Recurring work moved

Who gets served

Finance and accounts payable

Invoice processing, reconciliation, month-end close

Business units and country entities

Human resources

Payroll, employee onboarding, benefits administration

Every business unit and location

Procurement

Supplier sourcing, purchase orders, supplier onboarding

Operating units buying similar goods

IT service desk

Access requests, helpdesk tickets, infrastructure support

Employees across all sites

Customer and sales support

Order handling, customer queries, sales operations

Regional sales teams and customers

This hasn't been an experiment for a long time. Deloitte has tracked the model as a formal management practice since 1999 through its biennial global surveys, and its 2015 survey alone drew 311 respondents headquartered in 35 countries representing more than 1,000 shared service centers, per the Deloitte global shared services survey.

The scope has widened since then. Deloitte's later research found adoption of more complex, knowledge-based processes had doubled, and in some cases tripled, since 2013, as reported in the Deloitte shared services survey report. Documentation, product training, and customer education now sit alongside accounts payable as fair candidates.

Shared services are not outsourcing. The work stays inside your control, your standards, and your systems. A provider can sit beside the center, but the process owner and the service level remain yours. That distinction matters most in year two, when the novelty is gone and someone has to decide what happens to an exception nobody planned for. Getting the sequencing right is easier if you start with business process harmonization before anyone moves a role.

Say a distributor runs four branches. Each manager keeps a local invoice approval sheet in a personal spreadsheet, and the finance director spends two days a month reconciling them. Same task, four copies, one standard. That's your first candidate, not the messiest process you can find.

Test you can run this afternoon: list every recurring activity that three or more units perform in roughly the same way. The longest list is your shortlist.

The takeaway: a shared service model is defined by shared standards and one accountable owner, not by a building where people happen to sit together.

Which Operating Structure Fits Your Company

Three structures cover most setups: a single centralized center, a hub and spoke with a shared core plus regional satellites, and a hybrid where an internal team works alongside an outside provider. Structure decides how fast a plant manager gets an answer, so settle it before you settle on a city.

The diagram below shows how four common operating designs distribute authority across functions, divisions, a matrix, or a network, which is the same question a shared services decision raises in practice.

A comparison chart outlining four common business operating structures: functional, divisional, matrix, and network organizations.

Structure

How it works

Main trade-off

Best when

Centralized center

One team serves all units from a single location

Strong standards, slower local response

Volume is high and processes are uniform

Hub and spoke

Shared core plus regional satellites

Balances consistency with local speed

Regions need local language or regulatory handling

Hybrid with outside provider

Internal team plus an external partner for part of the work

Lower flexibility, needs clear service-level agreements

Some work is repetitive and the provider has proven capacity

Blended delivery is ordinary now, not exotic. In Deloitte's 2021 global survey, roughly 65% of respondents included outsourcers alongside their internal shared services, according to the Deloitte global shared services report. That makes written service levels between the internal and external parties a design requirement, not a legal afterthought.

Picture a manufacturer with plants in four countries. It centralizes finance into one location, and within two quarters the local controllers are complaining that answers take four days and nobody understands a specific tax rule. Moving to a hub with two regional satellites keeps a single chart of accounts and one close calendar, while restoring same-day answers for the two regions that need them. The work didn't change. The structure did.

How I'd choose

Volume first. Thin, sporadic work doesn't need a center. High transaction counts with predictable inputs do.

Languages and regulatory regions second. Two languages and one regulatory regime is a very different build from nine languages and four tax authorities, which is where a global capability center design starts earning its keep.

Process standardness third. If three units run the same process, centralize. If each one is a special case, harmonize before you relocate anything.

Appetite for central decision rights last. If leadership won't let one person override a country controller, a centralized model will fail while looking fine on paper.

My verdict: start centralized for uniform, high-volume work, and add satellites only where local response changes the outcome.

Governance in three decisions

Name one leader who owns policy, process standards, and performance reporting for every unit the center serves. Then choose the funding model before launch, either a straight cost pass-through to business units or a retained corporate budget, because a center that argues about chargebacks every quarter never gets to process improvement. Finally, publish service levels and an escalation path, and put the same terms in writing when a hybrid uses an outside provider.

The Deloitte shared services survey makes this point bluntly: centralization alone doesn't create savings, coordinated governance does. Three country teams documenting the same onboarding process three different ways is not a training problem. It's an ownership problem, and a global leader with real decision rights fixes it in one meeting.

How Large Are Shared Service Savings

About 55% of organizations with a global business services leader achieved more than 20% average savings in 2025, according to Deloitte's 2025 Global Business Services Survey. Note the qualifier in that sentence. The savings cluster around organizations that gave someone enterprise-wide authority.

The mechanism is operational efficiency, and it's worth stating plainly because it explains both the wins and the disappointments. One standardized process, one control framework, one technology stack, and one trained team serve many business units instead of every unit duplicating the same capability. Duplicate it five times and you pay for five control frameworks, five training programs, and five sets of auditors asking similar questions.

Deloitte's wider shared services work puts average annual savings at roughly 8%, with new, significant center implementations delivering about 15% in initial savings, per the Deloitte GBS reader. Those are averages across a mixed population, which is exactly why the cost benchmark below is more useful than the average for setting a target.

ScottMadden and APQC benchmarked 103 finance shared service organizations in 2024 and put median operating cost at $5.33 per $1,000 of organizational revenue, with a top quartile at $2.97 and a bottom quartile at $8.90, according to the 2024 financial shared services benchmarking study.

Performance group

Operating cost per $1,000 revenue

What usually differs

Top quartile

$2.97

Fewer staff, broader services, higher automation

Median

$5.33

Typical baseline for mid-maturity centers

Bottom quartile

$8.90

Weaker process maturity and governance

Read that spread carefully. The bottom quartile spends three times what the top quartile spends for the same revenue base, and both would describe themselves as shared services. A label creates no efficiency. Fewer staff carrying a broader service catalog, supported by automation, is what separates the two groups.

Say your finance center costs $14 million a year and the company books $2 billion in revenue. That works out to $7 per $1,000, which puts you above the median and well short of the top quartile. Before anyone sets a 20% savings target, that arithmetic tells you whether the target is ambitious or just arithmetically implausible.

The takeaway: compute your own cost per $1,000 of revenue, compare it to the median, and size the target from where you stand.

How Do You Measure Value Beyond Labor Savings

Savings tell you whether the center got cheaper, not whether it worked. Only 18% of respondents had achieved their GBS goals, while 21% struggled to align GBS KPIs with corporate objectives, 21% cited insufficient data availability or quality, and 20% lacked KPI standardization, all from the PwC findings summarized in Deloitte's 2025 Global Business Services Survey coverage.

That gap is the whole argument for a balanced scorecard. A program can post a real cost win and still miss the business goal it was created for, and the two facts coexist comfortably in the same annual report.

Measure group

Example KPIs

Enterprise outcome it supports

Service quality

Cycle time, first-contact resolution, accuracy

Customer retention

Adoption

Reuse rate across teams, active use of procedures

Employee productivity

Financial

Cost per completed deliverable, savings against plan

Budget discipline and reinvestment

Control

Exception rate, rework rate, freshness compliance

Risk reduction

Customer experience

Satisfaction scores, complaint volume

Revenue enablement

The control row is the one people drop first, and it's the one that catches the expensive problems. Centralization can cut unit cost while business owners lose sight of the whole process, because the number they see is a rate rather than a result.

The failure mode is predictable. A support center reports cost per ticket every month and it falls for three quarters. What nobody reports is the rework rate climbing underneath, and the sales team stops routing complex questions to the center because answers come back wrong the first time. The cost metric looked excellent right up to the moment the business stopped using the service.

Lock this before launch: six to eight measures, agreed with the business owners who will be judged on them, reviewed monthly with the same people in the room.

If your team can't name the enterprise outcome behind each measure, cut the measure. A dashboard with 40 KPIs is a way of avoiding the conversation about which six matter, and it's how shared services teams end up defending a cost number while service quality degrades. There's a practical method for finding where the automation effort should go first in this guide to automation opportunity identification.

One structural note that saves arguments later: your measurement views should mirror your operating model. Trupeer provides workspaces, permissions and brand rules that mirror how the customer is structured, which keeps a regional satellite from reporting on a different definition than the core.

The takeaway: pair every cost metric with a quality and adoption metric, or the scorecard rewards shrinking the service.

How Do You Launch a Shared Service Center

Choose two or three functions that already run similarly in several units, standardize one version, name an owner, and pilot for a full quarter before you automate anything. Half of respondents in PwC's 2025 analysis said GBS was a driver of digital transformation in their company, per the PwC Global Business Services 2025 analysis, which is why the transformation needs a sequence rather than a purchase order.

Here's the sequence laid out end to end.

A step-by-step infographic showing the seven-stage process for launching and scaling a corporate shared service center.
  1. Pick the first two or three functions. Recurring, high-volume work with stable rules: invoice processing, payroll inputs, access requests, order handling. The usual slip is choosing the process with the loudest internal sponsor instead of the most repeatable one, and then spending a year on exceptions.

  2. Map each process end to end. Inputs, exception paths, approval points, and the systems of record. Skip this and you automate exceptions nobody documented, which is how a center gets slower after it gets automated.

  3. Standardize one global version. One set of case categories, one approval logic, one control framework. Keep genuine statutory differences as explicit named variants rather than letting each country keep a private copy of the procedure.

  4. Name the executive owner, the process owners, and the escalation path. Write it down with names. An escalation route that ends at "the center" ends nowhere, and the transition sequencing in this GBS transition methodology walkthrough is a useful reference for who signs off at each gate.

  5. Set access and roles once. Manual account setup becomes a permanent tax on a center serving thousands of users across dozens of sites. Trupeer's SSO and SCIM plug into your identity stack, and access runs with zero manual provisioning.

  6. Pilot in one or two units for a full quarter, measured against the common scorecard rather than a local one. Three months is long enough to see a month-end close twice, which is where actual exceptions hide.

  7. Automate and scale only after the pilot hits its targets. Scaling before the pilot means copying unfinished rules into four more countries, and then rebuilding all five.

Automation underperforms in shared services for one reason: the tool inherits every local variation you left in place. It reproduces the exception logic of whichever unit built the original, then gets blamed when a second country's invoices route the wrong way.

At 90 days, a working launch looks boring. One intake form, one procedure library with named owners, one dashboard with six measures, and a pilot group that can tell you its exception rate without leaving the room.

The takeaway: standardize first, pilot second, automate third, and treat any proposal that reverses that order as a warning sign.

Why Do Shared Service Models Fail

A center that hasn't recovered its investment inside three years is telling you something. Deloitte's 2019 global survey found that 80% of respondents recovered the investment in a new center within three years and 50% within two, that 49% passed lower costs back to business units, 20% invested savings in technology, 14% invested in process improvement, and 10% invested in talent development, as reported in the 2019 global shared services survey infographic.

The reinvestment split is the more interesting half. Organizations that push every saving straight back into business unit budgets have no money for the automation and process work that keeps the center competitive in year three. That's how a good first-year result turns into a plateau.

The infographic below sets out the strategic and leadership failures that sit behind most stalled programs.

An infographic detailing five common reasons why business shared service models fail, including strategy and leadership issues.

Four failures account for most of the trouble:

  • Local workarounds that persist. Symptom: shadow spreadsheets reappear within a quarter. Cause: the center never got authority over the exception path, so units solved it themselves. Fix: one intake route and a published decision on what counts as a valid local variant.

  • Automation that makes errors faster. Symptom: cycle time falls while rework climbs. Cause: the process was automated before it was standardized. Fix: freeze automation until the exception rate is measured and understood.

  • Savings that stall after two years. Symptom: the year-three business case shows nothing new. Cause: every saving went back to the units, leaving no budget for automation or process redesign. Fix: agree the reinvestment share in writing before launch.

  • Teams that route around the center. Symptom: volume drops while headcount holds. Cause: service levels were never published, so nobody knows what to expect. Fix: publish the service level, log the misses, and review them with the business owner monthly.

Take the accounts payable scenario. A center automates invoice coding before standardizing vendor records, so duplicates and mismatched tax IDs now flow through the workflow faster than three clerks ever managed manually. Nothing in the automation was broken. The input rules were.

Monthly checklist, four questions: is payback still on the original schedule, is the exception rate falling, are the named owners answering escalations, and were the reinvestment rules actually applied this quarter?

If any answer is no, fix that before you approve another automation project, and read up on why digital transformations fail for the organizational patterns that repeat here.

Are You Ready to Commit to Shared Services

Four yes-or-no answers decide this, and three yeses means run a pilot rather than open a center. I'd ask them in this order, because each one invalidates the next if the answer is no.

  1. Do at least three business units already perform the same recurring work in similar ways? If the work is different in each unit, you have a harmonization project, not a shared services project.

  2. Will one executive take central decision rights and hold the center's budget? Without that, you're building a coordination forum with a cost target attached.

  3. Can the sponsor name the scorecard measures and the payback window today, before any hiring or system build? Vague measures agreed after launch always resolve in favor of cost.

  4. Are the chosen processes mapped well enough to standardize inside one quarter? If mapping alone would take six months, the scope is too wide for a first wave.

Three or more yeses: run a one-quarter pilot on one or two functions and stop there until the measures hold. Fewer than three: don't open a center. Fix process ownership and source data first, because centralizing an unowned process just moves the argument to a different building.

One action for this month. Block 30 minutes, list the recurring workloads three or more units share, and name the pilot function before the next leadership meeting. That single list will tell you more about readiness than any business case drafted without it.

What Do People Ask About Shared Service Models

Is a shared service model the same as outsourcing?

In shared services, the work remains inside your organization, governed by its standards and systems. An outside provider may handle part of the volume, but outsourcing transfers control of the process as well as its execution. For a hybrid arrangement, document the division of responsibility in a service-level agreement and assign one owner to the end-to-end process.

How quickly should a shared service center pay back?

Set the payback window before launch and review progress at the 12-month mark. If the center is behind schedule, examine process ownership and exception volume before adding automation. Automating inconsistent work can extend payback while obscuring the operational cause.

Which processes should move into a shared service center first?

Begin with high-volume, rule-stable work that is already similar across units, such as invoice processing, payroll inputs, access requests, or order handling. Documentation and customer education can follow once templates and review rules are in place. Judgment-heavy work is harder to standardize because cases can require different decisions.

What's the difference between shared services and global business services?

Shared services typically means a center serving several units, often within one function. Global business services integrates multiple functions across regions and may include outsourcing partners, with one leader accountable for the end-to-end chain. That wider scope calls for cross-functional process owners, rather than a single function leader.

How do you keep procedures current inside a center?

Use one searchable source of truth, assign an owner to each procedure, retain version history, and make documentation updates part of every process change. Screen recordings can be turned into structured SOPs with screenshots and chapters, reducing the effort required to publish usable instructions. Trupeer provides this workflow for creating documentation and searchable knowledge bases.

Shared-service results depend on more than reducing labor costs. Track whether teams follow the standard process, whether exceptions are handled consistently, and whether procedures stay current across sites. Governance must assign owners for both process performance and documentation, or initial savings can fade as workarounds and outdated instructions spread.

Trupeer turns screen recordings into studio-quality product videos, structured SOPs, and searchable knowledge bases, with translation into 65+ languages. Workspaces, permissions, brand rules, SSO and SCIM, and exportable audit logs support enterprise governance. The Enterprise plan includes a dedicated customer success manager and priority support with an SLA. To start, record a standard procedure with Trupeer at trupeer.ai.

How Genpact transformed training for 140,000 employees across 40 countries

When Genpact rolled out Workday across HCM, finance and data, Davetta Harper, VP Organizational Change Leader, had to make sure 140,000 employees across 40 countries could learn and adopt the new system. That meant 500+ training collaterals across 20 workstreams, in five languages. Done manually, with 20-page SOPs built screenshot by screenshot and human translation, the project would have taken at least 12 months.

Her team recorded existing process design sessions and SME walkthroughs, uploaded them to Trupeer AI, and got back polished SOPs and demo videos automatically translated into all five languages. Templates kept formatting consistent, and AI avatars held up in every language with minimal edits.

Genpact delivered 500+ training collaterals in three months instead of twelve, 75% faster than the standard timeline, and end users consumed the content without realizing it was AI-generated.

“We needed to train 140,000 people across 40 countries on a completely new system. Trupeer gave us a way to actually get there.”

— Davetta Harper, VP Organizational Change Leader, Genpact

Read the full Genpact story →

Questions fréquentes

Besoin d’un monteur vidéo, d’un traducteur et d’un scénariste ?

Essayez Trupeer gratuitement

Réserver une démonstration

Besoin d’un monteur vidéo, d’un traducteur et d’un scénariste ?

Essayez Trupeer gratuitement

Réserver une démonstration

Besoin d’un monteur vidéo, d’un traducteur et d’un scénariste ?

Essayez Trupeer gratuitement

Réserver une démonstration