A make or buy decision analysis compares what it costs you to produce a part, package, or service in-house against what it costs to buy it from a supplier — and then weighs the factors a spreadsheet cannot price. Most teams get the arithmetic right and still make the wrong call, because the cheapest unit price is rarely the cheapest total decision.
This guide gives you the whole framework: the cost heads to include, the costs to leave out, the crossover volume where the two options swap places, the non-cost criteria that usually break the tie, and a seven-step process you can run on a real part in an afternoon. If you make molded parts, there is also a section on tooling amortization and supplier minimum order quantities that change the math more than most people expect.
Table of Contents
- Make or Buy Decision Analysis at a Glance
- What Is a Make or Buy Decision Analysis?
- When Should a Manufacturer Make a Component?
- When Should a Manufacturer Buy a Component?
- How to Compare Total Cost of Ownership
- What Non-Cost Factors Should the Analysis Include?
- Make or Buy Decision Analysis for Plastic Products and Packaging
- How Sensitivity Analysis Changes the Decision
- Common Mistakes in Make or Buy Decisions
- Frequently Asked Questions
- The First Thing to Do
Make or Buy Decision Analysis at a Glance
The table below is the shortest useful version of the framework. Read the middle column and you get the case for making; read the right column and you get the case for buying. A clean analysis ends with a row that is genuinely close, because the factors that are not close usually settle the decision before the cost model does.
| Factor | Leans make | Leans buy |
|---|---|---|
| Total cost at expected volume | Fixed costs spread over enough units; direct labour already on the payroll | Supplier spreads its own fixed costs, economies of scale, no capital outlay |
| Capacity | You have spare machine hours and the control that comes with them | Your machine is loaded; adding the part displaces work that earns more |
| Quality | Design intent lives in your process, tolerances are hard-won, defects stop at the machine | The supplier runs a mature, audited process for this exact part |
| Lead time | Short, predictable, and you can change quantity this week | Longer queue, but dependable if the supplier’s schedule is reliable |
| Risk | Geopolitical, tariff, and port exposure sits inside your own building | Your exposure is supplier financial health, capacity promises, single-source dependency |
| Flexibility | Engineering changes land in days; prototypes stay in-house | New revisions need a new purchase order and a new quote cycle |
| Intellectual property | The design never leaves the building; you own the process | Drawings and tooling are shared with a firm you do not control |
| Strategic importance | The part is core to your product and your differentiation | The part is standard, interchangeable, and not what you sell |
What Is a Make or Buy Decision Analysis?
A make or buy decision analysis is a structured comparison of the total relevant cost of producing a component in-house against the total relevant cost of purchasing it from an external supplier, so a company can choose the sourcing option that best supports its cost, capacity, quality, risk, and strategic goals.
It exists because a sourcing choice is rarely just a purchase order. Insourcing commits capital, people, and floor space to a process. Outsourcing commits the company to a supplier’s schedule, its quality record, and its pricing for as long as the part lasts. The analysis makes both commitments visible before either one is made.
What the analysis actually decides
In manufacturing it decides whether to produce a component or subassembly in-house or buy it from a contract manufacturer, distributor, or job shop. The same logic runs for packaging components, finished assemblies, and services, and it also shows up outside the plant as software build versus buy, equipment ownership versus rental, and HR or IT outsourcing.
The output is normally one of three things: a decision to make, a decision to buy, or a decision to stay open and gather better data. That third outcome is legitimate. A pilot run at low volume often tells you more than a month of spreadsheet work.
Why a unit price comparison is not enough
Because unit price omits most of what you actually spend. A quote of 0.62 per part does not include the tooling sitting on your balance sheet, the freight you pay, the inspection you do on receipt, the scrap rate, the inventory you finance while it sits on a shelf, or the two weeks of production planning that part consumed. The same is true in reverse: an in-house cost that counts rent and management time can hide the fact that your press is idle anyway.
The discipline that fixes this is relevant cost analysis. You count only costs that change if you switch options, and you leave out sunk costs entirely. Money already spent on a machine, a lease, a certification, or a failed supplier audit does not change because you make the part instead of buying it.
Relevant cost and sunk cost in one line each
- Relevant cost is money that goes away or arrives if you choose one option over the other. It is the only kind that belongs in the comparison.
- Sunk cost is money already committed and unrecoverable, such as a paid-down press. Including it in either column is the single most common way these analyses go wrong.
When Should a Manufacturer Make a Component?
In-house production usually wins when the part is proprietary, the volume is steady and high enough to fill a machine centre, or the process knowledge is itself the product. The common signals:
- You own the design and intend to keep it proprietary. If a supplier will hold the drawings, hold the tooling, or quote the part for a competitor, you have given away more than a part.
- Volume is predictable and above the crossover point. Steady demand is what makes amortizing tooling and equipment sensible; lumpy demand is not.
- The part is integrated into your own process. When the component is only one step from assembly, a return trip to the supplier means extra handling, extra queues, and more ways for a schedule to slip.
- You have equipment and skill that already exist. Idle capacity turns a fixed cost into a marginal one, and the economics change completely.
- Supply security matters. Long import lanes, tariff exposure, or a single overseas source can outweigh a modest cost difference.
- Engineering changes are frequent. A part under revision every quarter is expensive to re-quote and slow to re-tool through a supplier.
- Tolerances or material behaviour are hard for outsiders. Fibre fill, warpage control, or tight cosmetic surfaces often demand process knowledge that lives on your floor, not in a quote.
- You can improve the process. If cycle time or scrap can fall meaningfully in-house, the make case strengthens over time rather than weakening.
When Should a Manufacturer Buy a Component?
Purchasing usually wins when the component is standard, the volume is low or erratic, or the capital and attention it would take are better spent elsewhere. The common signals:
- Volume is low, uncertain, or seasonal. Below a few thousand units a year, tooling and equipment rarely amortize.
- Capital is constrained. A press, a printer, and a small clean room are a large commitment for a part that is not your core product.
- Your capacity is already committed. If adding the part means turning away higher-margin work, buying is the answer even at a higher unit cost.
- A specialist already does it better. Decorating, plating, precision machining, and certification testing all exist as supplier industries for a reason.
- You have excess capacity elsewhere that is worth more. Your people and machines may earn more on a different product.
- You want fewer things to manage. Every process you remove is supervision, training, spare parts, audits, and one less thing to go down at 2 a.m.
- The supplier offers real risk reduction. A qualified second source, or a supplier with capacity you do not have, changes the non-cost math.
How to Conduct a Make or Buy Decision Analysis
Seven steps, in this order. Skipping straight to costing is the most common failure, because most of the deciding power sits in steps one and two.
- Define the component and the scope. Write down exactly what is being compared: the part, the revision, the packaging, the inspection, the freight terms, and the delivery schedule. Two companies often disagree because they analyzed different scopes.
- Establish the demand. Build a volume range, not a single number. Low, expected, and high case, with the year each applies to and how confident you are. Everything downstream depends on this.
- Build the make cost model. Direct materials, direct labour, variable overhead, allocated fixed overhead, tooling and equipment amortization, facility and supervision, quality costs, and the management time the process consumes.
- Build the buy cost model. Supplier unit price, any tooling or setup charge you pay, freight and duties, purchase-order and receiving transaction costs, incoming inspection, inventory carrying cost, and the cost of non-conformance.
- Find the crossover volume. Solve for the volume where the two cost curves intersect. Below it, one side wins; above it, the other side does.
- Score the non-cost factors. Capacity, quality, lead time, flexibility, intellectual property, compliance, geographic risk, and strategic importance, weighted to how much they matter to your business.
- Test it and document it. Run sensitivity cases, name the assumptions, write down who agreed, and set a date to revisit. A decision nobody recorded will be relitigated the next time volume moves.
The decision rule is worth stating plainly: make the part if the relevant cost per unit at your expected volume is lower and no qualitative red flag applies. If cost favours buy but a red flag does apply — proprietary design, single-source risk, a compliance exposure — then fix the flag or accept the premium knowingly. Do not let a spreadsheet settle a question it cannot see.
How to Compare Total Cost of Ownership
Build both models the same way so the columns are comparable. List every cost head, mark it include or exclude, and write down why. The include and exclude list is the part a reviewer will actually challenge.
| Cost head | Include? | Reason |
|---|---|---|
| Direct material per unit | Include | Changes with every decision |
| Direct labour and burden | Include if the headcount changes with the decision | Fixed salaried supervision is often not relevant |
| Variable overhead | Include | Moves with machine hours and volume |
| Fixed overhead absorption | Include only to the extent capacity changes | Idle machine time is not a cost of making |
| Tooling and capital amortization | Include | Differs sharply between options |
| Freight, duties, insurance | Include | Almost always larger on the buy side |
| Inventory carrying cost | Include | Longer lead times mean more cash in work in progress |
| Quality and non-conformance cost | Include | Scrap, rework, returns, and field failures |
| Transaction and management time | Include | Purchasing, receiving, supplier qualification, expediting |
| Past capital already spent | Exclude | Sunk cost; unchanged by the decision |
| Allocated corporate overhead | Exclude unless it truly changes | Allocation usually appears in both columns and cancels out |
A worked cost-head example
Take a small molded enclosure made in-house or bought from a contract molder. All figures below are illustrative, in USD, at the volume stated.
| Cost head | Make | Buy |
|---|---|---|
| Material per unit | 1.10 | 1.15 (included in quote) |
| Direct labour per unit | 0.45 | 0.00 |
| Machine and overhead per unit | 0.60 | 0.00 |
| Tooling, one-time | 18,000 | 0 |
| Freight and duty per unit | 0.00 | 0.14 |
| Incoming inspection per unit | 0.00 | 0.08 |
| Inventory carrying per unit | 0.02 | 0.07 |
| Transaction and management per unit | 0.03 | 0.12 |
| Scrap allowance at 4 percent | 0.05 | 0.00 |
| Variable cost per unit | 2.25 | 1.56 |
Variable cost alone would send this part straight to the supplier. But the make option carries 18,000 of tooling, so the crossover is where 18,000 divided by the 0.69 per-unit advantage equals roughly 26,000 units.
| Annual volume | Make cost per unit | Buy cost per unit | Lower |
|---|---|---|---|
| 5,000 | 5.85 | 1.56 | Buy |
| 20,000 | 3.15 | 1.56 | Buy |
| 40,000 | 2.70 | 1.56 | Buy |
| 80,000 | 2.48 | 1.56 | Buy |
At these figures the supplier wins on cost across the whole range, and the make case has to rest on something else — speed of engineering change, or IP, or a capacity story that the cost model does not show. That is a perfectly good outcome. What matters is that you know which side of the line you are on before you start defending the decision in a meeting.
Run the same arithmetic on a part where the variable costs are close and the crossover lands inside your demand range, and the picture changes. That is why the crossover number, not the unit quote, is the thing to put in front of a decision meeting.
What Non-Cost Factors Should the Analysis Include?
Cost decides most comparisons that are close. Ten factors decide the ones that are not, and none of them show up in a per-unit column.
- Capacity and flexibility. Can you add the part without displacing better work, and can you change quantity without a new purchase order?
- Quality consistency. Ask what the supplier’s actual defect rate is, not what their brochure says. If defects are still hard to explain, the conversation about root cause analysis methods for manufacturing defects usually helps more than renegotiating price.
- Lead time and predictability. A three-week quote with variability is worse than a six-week quote that holds.
- Supplier reliability. Financial health, capacity commitments, and how they behaved during their last shortage.
- Intellectual property exposure. Tooling ownership, drawings, NDA enforceability, and whether the supplier serves your competitors.
- Regulatory and compliance fit. Traceability, material certifications, and audit requirements in your target markets.
- Geographic and geopolitical risk. Tariffs, import lanes, port congestion, and whether dual sourcing is realistic.
- Scalability. How each option behaves when volume doubles, or when a new region opens.
- Labour availability and skills. Hiring and training for a process you do not currently staff.
- Strategic control. Whether this part is where your differentiation lives.
Score these rather than arguing them. Give each a weight that reflects how much it matters to you, rate make and buy on a simple scale, and multiply. A weighted matrix turns a room full of opinions into a number you can disagree with precisely.
| Criterion | Weight | Make score | Buy score | Weighted make | Weighted buy |
|---|---|---|---|---|---|
| Total cost at expected volume | 35% | 3 | 5 | 1.05 | 1.75 |
| Quality control | 15% | 5 | 4 | 0.75 | 0.60 |
| Lead time | 15% | 4 | 3 | 0.60 | 0.45 |
| Flexibility for engineering changes | 10% | 5 | 2 | 0.50 | 0.20 |
| Supply security | 10% | 5 | 3 | 0.50 | 0.30 |
| Intellectual property | 10% | 5 | 2 | 0.50 | 0.20 |
| Strategic fit | 5% | 4 | 3 | 0.20 | 0.15 |
| Total | 100% | 4.10 | 3.65 |
In this example buy scores better on cost but loses badly on flexibility and IP, and the weighted total flips. Whether you agree with the weights is the real argument, and now that argument is about the weights rather than about whose spreadsheet is right.
Make or Buy Decision Analysis for Plastic Products and Packaging
For molded parts and packaging, the decision turns on a handful of costs that other industries never deal with. Get these right and the rest of the analysis is arithmetic.
Tooling amortization
A steel or aluminium tool is a one-time cost that only makes sense spread across parts. The amortization assumption is the single most sensitive number in the model, so state it explicitly: a 22,000 tool credited against 60,000 parts adds 0.37 per part, but against 600,000 parts it adds 0.04.
| Tool cost | Lifetime parts | Amortized per part |
|---|---|---|
| 22,000 | 60,000 | 0.37 |
| 22,000 | 150,000 | 0.15 |
| 22,000 | 600,000 | 0.04 |
Two questions decide whether your supplier’s tool amortizes over its own volume or yours. Ask whether the tool is built for your annual quantity or for the supplier’s next million pieces, and ask who owns it if you leave.
Material, scrap, and colour
Resin price moves, so model two or three material scenarios rather than one. Scrap at a 4 percent rate on a 1.10 part is not a rounding error, and a colour change on a small run can trigger a full purge that costs more than the parts in the run itself.
Inserts, secondary operations, and packaging formats
Overmoulded inserts, metal inserts, hot stamping, assembly, and printing are frequently excluded from a quote and added later as change orders. Ask for a part-level breakdown that names every secondary operation. If tolerances on a cosmetic surface are critical, review our guide to plastic part tolerance standards before comparing quotes — the tolerances in the print are what determine whether a supplier can quote at all.
Supplier minimum order quantities
This is where in-house often looks better than it is. If a molder requires 25,000 pieces per colour per run and your forecast is 8,000, you either buy 3 years of inventory or pay for a changeover. The unused material is a real cost and belongs in the buy column. Where demand is too small for production tooling, bridge tooling for low-volume production is often the honest middle path.
Packaging runs into the same arithmetic with different names: minimum print run, plate or cylinder cost, kraft or resin price, freight by weight, and pallet configuration. The principle is identical.
How Sensitivity Analysis Changes the Decision
A single cost-per-unit number is a snapshot of assumptions. Sensitivity analysis shows which assumption the decision actually depends on, and it usually shows that volume matters more than anything you negotiated.
Take the earlier example and vary the inputs:
| Scenario | Crossover volume | Decision effect |
|---|---|---|
| Base case | 26,000 parts | Buy across the forecast range |
| Volume grows to 120,000 a year | 26,000 parts | Crossover still below demand; make economics improve but do not flip on cost alone |
| Resin price falls 20 percent | 19,000 parts | Make variable cost drops; the gap narrows |
| Supplier minimum rises to 40,000 | 26,000 parts, plus holding cost on excess | Buy option worsens sharply above your forecast |
| Scrap improves from 4 to 1 percent | 21,000 parts | Process learning helps the make case |
| Downtime adds 8 percent to cycle time | 31,000 parts | Capacity risk erodes the make advantage |
| Freight and duty rise 50 percent | 18,000 parts | Buy cost rises; make starts to look competitive |
Test at minimum the things that actually move: volume, resin or material price, direct labour rates, scrap rate, freight, duty, downtime, demand growth, and the supplier’s minimum quantity. If the decision flips inside a plausible range of any one of them, you do not have a decision yet — you have a risk.
The honest output at that point is usually one of three: a pilot at low volume, a dual-source arrangement where a portion of demand stays in-house, or a written acceptance of the risk with an agreed review date. All three are better than a confident number that nobody stress-tested.
Common Mistakes in Make or Buy Decisions
- Comparing unit price only. Fix: build both cost models with the same heads, all the way to cost per unit at expected volume.
- Leaving sunk costs in. A paid-down machine and a completed supplier audit appear in the columns and cancel out anyway, which only hides the real difference. Fix: delete every cost that does not change with the decision.
- Ignoring tooling and setup. The tool is the largest single line in most plastic decisions and it usually lives on the supplier’s side of the quote. Fix: get tooling terms, ownership, and life written into the agreement.
- Underestimating changeover and setup. Short runs look cheap until you count setup, purge, and colour changes. Fix: price the realistic run pattern, not the ideal one.
- Using a single optimistic volume. Fix: model low, expected, and high, and note where the crossover sits in each.
- Leaving quality out of the model. Scrap, rework, returns, and field failures are real costs that rarely appear in a quote. Fix: include cost of conformance and non-conformance on both sides.
- Treating risk as a single number. A risk-adjusted figure hides what you cannot quantify. Fix: name the specific exposures — single source, geography, IP, capacity — and decide what you will do about each.
- Never revisiting. Fix: set a review trigger, such as a volume change beyond a threshold, a new tariff regime, or a supplier ownership change.
Frequently Asked Questions
What is a make vs buy analysis?
A make vs buy analysis compares the total cost and total risk of producing a component in-house with buying it from a supplier. It builds one cost model for each option, excludes sunk costs that do not change with the decision, finds the volume where the two options cost the same, and scores non-cost factors such as quality, lead time, capacity, intellectual property, and supply risk before recommending make or buy.
What do you understand by a make-or-buy decision?
A make-or-buy decision is the choice between insourcing a component or service and outsourcing it to an external supplier. Insourcing keeps control, design ownership, and the ability to change production quickly, but commits capital and people. Outsourcing lowers the fixed commitment and taps supplier expertise, but adds lead time, dependency, and exposure to the supplier’s reliability. The decision is normally resolved with a documented cost and risk comparison.
What are the steps in a make or buy analysis?
Most practical processes use seven steps: define the component and scope, establish low, expected, and high demand, build the make cost model, build the buy cost model, calculate the crossover volume, score the non-cost factors with weights, and stress-test the result with sensitivity cases before documenting the decision and a review date. Skipping the scope and demand steps is where most analyses fail.
How do you calculate the break-even point for make or buy?
Subtract the lower variable cost per unit from the higher one, then divide the difference into the one-time or fixed cost of the more expensive option. The result is the annual volume at which the two options cost the same. Below that volume the low-variable-cost option wins; above it, the option carrying the fixed cost wins. Add any excess-inventory holding cost from supplier minimum order quantities to the buy side before you solve.
When does outsourcing become economically attractive?
Outsourcing becomes attractive once annual volume sits below the crossover point, demand is too uncertain to justify a capital commitment, or the process is a specialist trade rather than a core competence. It also makes sense when your capacity is fully committed, when supplier minimums fit your forecast without leftover inventory, and when dual sourcing or shorter lead times carry more value than the unit price gap.
What are the 5 Ps of purchasing?
The five Ps of purchasing management are price, quality, quantity, delivery, and place. Price covers the commercial terms, quality covers specification conformance and inspection, quantity covers order size and minimums, delivery covers lead time and schedule reliability, and place covers the supplier location, logistics, and packaging condition. A make or buy analysis should score all five on the buy side, because the cheapest quote often loses on the other four.
The First Thing to Do
Pick one part, one year, and one realistic volume. Build both cost models with the same heads and no sunk costs, then solve for the crossover volume — that number alone will tell you whether the conversation you are about to have is about cost at all.
Everything after that is judgement: whether the non-cost factors justify overriding the number, who signs off, and what event would make you revisit the decision. Write it down, because the next make or buy decision analysis on the same part will start from your notes, not from scratch.