Outsourcing vs Insourcing Decision Framework

Executive summary

The question "should we outsource this?" usually arrives as a quote: a vendor will do the work for a fixed fee, and the fee is lower than the cost of the people who do it now. That comparison is incomplete in both directions. The internal side is often understated because it omits management time, tools, recruiting, and turnover. The vendor side is often understated because it omits transition, oversight, rework, escalations, price increases, and the cost of leaving.

This framework does four things in order:

1. A strategic screen: is this activity a source of advantage or a utility? Utilities are candidates for outsourcing. Sources of advantage usually are not.
2. An economic model: total cost for both options over the same horizon, with one-time and recurring costs separated and a break-even test.
3. A risk and control screen: what can go wrong, who feels it first, and how reversible the decision is.
4. A governance plan: the contract terms, service levels, and review cadence that make the arrangement safe to run.

The framework applies in both directions. Insourcing an outsourced function follows the same logic, with the vendor's fee on one side and the cost of building the capability on the other.

Anything that touches employees (headcount, severance, transfers of staff to a vendor, consultation duties) or the legal terms of the contract belongs with HR and counsel. This guide covers the finance and operating analysis only.

When to use this

Use the framework when:

• A vendor has proposed to take over a function (payroll, IT support, logistics, customer service, engineering services, manufacturing of a component).
• An outsourced service has become expensive or unreliable and someone proposes bringing it back.
• You are reviewing overhead and a function looks like a candidate for an alternative delivery model.
• You are integrating an acquisition and two companies do the same work in different ways.
• Capacity is the constraint and the choice is to hire, to build, or to buy.

Do not use it to avoid a management problem. If a function underperforms because of unclear ownership or bad process, a vendor inherits the same problem and charges for it.

Diagnostic questions

1. What exactly is the activity, in terms of outputs delivered, volume, and frequency? Can you describe it without naming the team that does it today?
2. Does the activity differentiate you from competitors, or would a customer not notice who does it, provided it is done well?
3. What does it cost today, fully loaded: people, management time, tools, facilities, training, recruiting, turnover?
4. What are the quality and delivery measures today, and who tracks them?
5. How variable is the volume? Would a vendor's pricing handle peaks and troughs better than your fixed capacity?
6. What proprietary data, customer contact, or intellectual property does the activity touch?
7. What would happen if the service stopped for a week? A month?
8. How many qualified vendors exist, and how easy is it to switch from one to another?
9. What is the horizon of the decision: one year, three, five, and what is the exit route?
10. What regulatory, contractual, or lender requirements apply to who may perform the work or where the data sits?
11. What institutional knowledge sits in the current team, and what happens to it?
12. Who in the company will manage the vendor, and do they have the capacity and the skill?

If question 6 or 10 raises a flag, involve counsel early. If question 12 has no name, you do not have a governance plan yet, and an outsourced service without an owner decays quickly.

The framework

Test 1: strategic role

Sort the activity into one of three groups:

Be honest about which group an activity is in. Leaders tend to call everything core. A useful check is to ask whether you would be comfortable if a competitor used the same vendor for the same work.

Test 2: total cost over a stated horizon

Build both options on the same cost lines for the same number of years.

Keep or insource (recurring): people at fully loaded cost, management share, tools and licenses, facilities, training, recruiting and turnover allowance.

Outsource (recurring): the vendor fee including expected escalators, internal vendor-management time, quality and rework cost you absorb, tools you still need to keep, and any retained staff.

Outsource (one-time): transition and knowledge transfer, parallel running, contract and legal costs, systems integration, and the costs of any change to your workforce, which HR and counsel must size.

Insource (one-time): recruiting, training, tools, setup, and the exit cost from the current vendor.

Total cost (n years) = One-time costs + Sum of recurring costs for each year, with escalators
Break-even one-time cost = Total cost of the more expensive option − Total cost excluding one-time items of the cheaper option

The second line shows how large the one-time costs can be before the apparent saving disappears. It is the most useful number in the model, because one-time costs are the hardest to estimate.

Test 3: risk, control, and reversibility

Score each option on these questions, using a simple high/medium/low and a sentence of evidence:

Reversibility is the one executives forget. A function that has been outsourced for five years, with the internal team gone and the systems migrated, is expensive to reverse. Price that into the decision: a contract with a short initial term and a documented exit plan is worth more than a lower fee that locks you in.

Test 4: governance

Decide before signing who owns the relationship, what the service levels are, how they are measured, what the remedies are, and how often performance is reviewed. A good decision with weak governance performs worse than a mediocre decision with good governance.

Step-by-step process

1. Define the scope as outputs, volumes, and service levels, not as a team. Owner: function head with FP&A. Output: a one-page scope.
2. Run the strategic screen with the leadership team. Owner: CEO or COO. Output: group classification and reasoning, recorded.
3. Build the internal baseline. Owner: FP&A. Output: fully loaded cost, tied to the ledger, with quality and delivery measures.
4. Collect comparable vendor proposals on the same scope and service levels. Where only one vendor is quoting, the price is only a starting point. Owner: Procurement. Output: normalized proposals. See vendor consolidation playbook and procurement playbook for running a sourcing event.
5. Build the total cost model for both options over the chosen horizon. Owner: FP&A. Output: model with one-time and recurring costs separated.
6. Run sensitivities: volume up and down, fee escalator, transition cost 50% higher, a service failure. Owner: FP&A. Output: break-even one-time cost and a range.
7. Complete the risk screen. Owner: function head, with IT and legal as needed. Output: risk table with mitigations.
8. Send HR and legal items to HR and counsel: workforce impact, transfer obligations, contract terms, data protection, and anything in your lender documents that restricts the change. Owner: CFO. Output: written advice before any announcement.
9. Decide, record, and set review dates. Owner: CEO. Output: decision memo with assumptions and the conditions that would trigger a re-review.
10. Run the transition with a named lead, parallel running where feasible, and a go/no-go gate. Owner: function head. Output: transition plan with dates.
11. Review at six and twelve months against the original model. Owner: CFO. Output: variance between modeled and actual cost and service.

Templates

Scope sheet

Total cost comparison

Contract checklist for counsel review

1. Term, renewal, and termination rights, including termination for convenience and the fees attached.
2. Service levels, how they are measured, and remedies for misses.
3. Price escalators and what triggers a price change.
4. Data ownership, security, audit rights, and return of data on exit.
5. Transition-out assistance and its cost.
6. Subcontracting and location restrictions.
7. Liability caps and insurance.
8. Key personnel and change-of-control clauses.

This is a list of topics to raise with counsel, not legal advice. Your agreements and your counsel decide the actual terms.

Worked example

Illustrative. The numbers below are made up to show the method. They are not benchmarks.

A company is considering outsourcing tier-1 customer support. The horizon is three years, recurring costs for both options escalate at 3% a year, and one-time costs do not escalate.

Internal option, Year 1 recurring:

Outsourced option, Year 1 recurring:

One-time for the outsourced option: transition, knowledge transfer, and parallel running estimated at $90,000. Workforce-related costs are excluded here because they must be sized by HR and counsel.

Three-year totals with 3% escalation:

The headline comparison of fee to payroll ($430,000 versus $468,000) suggests a saving of $38,000 a year, or $114,000 over three years. The full model shows a three-year difference of $1,817,449 minus $1,619,996, or about $197,000, which is larger because the internal side carries tools, management, and turnover cost. The model is more favorable to outsourcing than the headline.

But the break-even test matters more. The apparent advantage of about $197,000 is the amount that all unmodeled one-time costs must stay under. If HR and counsel advise that workforce-related costs, contract exit fees on existing tools, or stranded costs add up to more than $197,000, the saving is gone. A 50% overrun on transition ($135,000 instead of $90,000) alone cuts the advantage to about $152,000.

Other things the numbers do not show, and the decision memo should address:

• Quality: the model assumes equal service levels. If customer satisfaction falls, the cost shows up in churn, not in the support budget.
• Reversibility: bringing the function back after three years requires hiring and training from scratch.
• Dependency: the vendor's second-year price escalator is modeled at 3%, but contract terms may differ.

A saving of this size, with these uncertainties, is a reasonable candidate for a pilot or a shorter initial contract rather than a full transfer. That is a judgment, not a result of the calculation.

Metrics and KPIs

Targets by method. Do not take an outsourcing saving percentage from a vendor deck. Set the target from your own baseline: your current cost per unit of output, your best trailing service levels, and contractual floors for service. A decision rule such as "outsource only if the three-year advantage exceeds the break-even one-time cost by a stated margin, and service levels are no worse than our own trailing best" is a method, and you can write it down before any quote arrives.

Common mistakes

1. Comparing the fee to payroll only. Add tools, management, turnover, and oversight to both sides.
2. Ignoring one-time costs. Transition, parallel running, contract exits, and workforce costs are real and often large.
3. Using a short horizon for a long commitment. A one-year view favors outsourcing; a five-year view with escalators may not.
4. Treating every activity as a utility. If the activity touches how customers experience you, test it harder.
5. No internal owner. Someone has to manage the vendor, review quality, and challenge invoices.
6. Signing before defining service levels. Vague scope leads to change orders and disputes.
7. Underpricing reversibility. If you cannot exit, you cannot negotiate.
8. Announcing before HR and counsel have reviewed. Employment obligations vary and are not a finance call.
9. Never reviewing. The model is a prediction. Compare it to actuals at six and twelve months.

AI prompt assistance

AI can structure the analysis and stress-test the model, but it has no knowledge of your contracts, people, or operations. Supply your own data and check every figure.

1. Build the comparison skeleton.

I am comparing keeping [activity] in-house with outsourcing it. Here are the internal costs and the vendor proposal. List every cost line that should appear on each side, mark each as recurring or one-time, and flag the lines I have not provided data for.

Check: every line must trace to the ledger or a quote. Fill the gaps yourself.

2. Stress-test the assumptions.

Using this three-year model, show the total for each option if: volume is 20% higher, the vendor escalator is 6% instead of 3%, transition costs 50% more, and one-time unmodeled costs are $150,000. Present the result as a table with the break-even one-time cost for each case.

Check: recompute one scenario by hand.

3. Draft questions for the vendor.

Draft twenty questions I should ask a vendor before signing: service level definitions and measurement, escalators, subcontracting, data handling, transition and exit, key personnel. Group them by topic.

Check: have procurement and counsel add and edit. Do not treat the list as complete.

4. Summarize the decision memo.

Write a one-page decision memo from this model: the recommendation, the three assumptions that matter most, what would change the decision, and the review dates. Mark any statement that is not supported by the data I provided.

Check: the CFO should confirm each number against the model.

Implementation guide

Governance: the CEO owns the decision, the CFO owns the model, the function head owns delivery and the vendor relationship, HR and counsel own people and contract matters. Report service levels and cost per unit monthly to the operating review.

Downloadable assets

The scope sheet, the total cost comparison, and the contract checklist above are laid out to be copied into a spreadsheet. The worked example is a useful test case for your own model because its totals can be reproduced from the stated inputs. There is no dedicated calculator for this guide.

Related resources

• SG&A cost guides hub: the wider set of overhead and structure guides.
• Vendor consolidation playbook: running a sourcing decision across several suppliers.
• Procurement playbook: the sourcing process behind a vendor selection.
• Span of control cost analysis: the structure that the function sits in.
• Zero-based budgeting for mid-market companies: where outsourcing proposals often surface.

Frequently asked questions

How do I decide between outsourcing and insourcing?

Start with whether the activity is a source of advantage or a utility, then compare total cost for both options over the same horizon, including one-time and exit costs. Finally screen for risk, control, and reversibility. If the economic advantage is smaller than the plausible one-time costs, the decision is not yet made.

What costs are usually missed in an outsourcing comparison?

The commonly missed costs are transition and parallel running, internal vendor management, rework and escalations absorbed by your team, contract exit fees on existing tools, price escalators, and anything arising from changes to your workforce. The internal side is also often understated: management time, tools, recruiting, and turnover.

How long should the comparison horizon be?

Use a horizon that matches the contract and the reversibility of the decision, commonly three to five years. A one-year view favors outsourcing because one-time costs look small relative to savings, and ignores escalators. State the horizon in the decision memo and test a longer and a shorter one.

Can I outsource without involving HR or legal?

No. Any decision that affects employees or contract terms involves employment obligations, transfer rules, data protection, and liability that vary by jurisdiction and by your own agreements. Finance should build the cost model, and HR and counsel should advise on those items before any announcement or signature.

When does insourcing make sense?

Insourcing makes sense when the activity has become a source of advantage, when service quality or control has been poor, when volume has grown enough that fixed internal capacity is cheaper per unit, or when dependence on the vendor is a risk. Use the same total cost model, with the cost of building the capability and exiting the vendor as the one-time items.

How this works in a System of Performance

An outsourcing decision is a prediction about cost and service, and most companies never check it. The model sits in a spreadsheet, the contract in a folder, and the service data in the vendor's reports. A Performis routine can take on a recurring job like this, with the scope and fit agreed with you: it works from the records your team supplies, prepares the work for your named reviewer, and keeps each approval on record. If you want to see where your own cost base might be a make-or-buy question, bring it to us as a launch partner.