Common Mistakes

Costly Make vs Buy and Outsourcing Mistakes That Wreck the Numbers

The make-buy and outsourcing decisions that go wrong usually fail on the same handful of errors: unabsorbed overhead, ex-works pricing dressed up as landed cost, and switching costs nobody counted.

The most expensive mistake is comparing an outside quote against variable cost only while treating in-house overhead as if it disappears. Symptom: a supplier quote of 4.20 dollars per part looks cheaper than your 5.10 dollar internal cost, so you outsource, then plant overhead per unit climbs across everything that stays. Root cause: fixed overhead of 1.40 dollars per part does not leave when volume leaves. Fix: in the Make vs Buy Cost tool, only remove the fraction of overhead you will actually eliminate. If 900,000 dollars of the 1.4 million dollar burden is truly avoidable, model 0.90 dollar per unit as saved, not 1.40, and the buy case flips.

Quoting ex-works and calling it landed cost quietly buries 15 to 40 percent of the real number. Symptom: a nearshore part at 3.10 dollars beats a domestic 3.55 dollars on the spreadsheet, but annual spend runs over budget by 220,000 dollars. Root cause: freight, duty at 2.5 to 6 percent, broker fees, and 45 days of in-transit inventory carry were never added. Fix: run the Total Landed Cost calculator with every adder. Ocean freight of 0.28 dollars, duty of 0.14 dollars, and carrying cost of 0.19 dollars per unit push the nearshore part to 3.71 dollars, above the domestic quote you rejected.

Ignoring inventory and pipeline stock on long lead times distorts the whole network model. Symptom: unit price fell 12 percent after moving offshore, yet working capital jumped and cash got tight. Root cause: a 60 day ocean lead time plus 30 day safety stock means roughly 90 days of demand sits as inventory versus 15 days locally, a sixfold increase. Fix: in Nearshore vs Domestic Cost, add carrying cost at 18 to 25 percent annually on the extra pipeline. On 12 million dollars of annual COGS, 75 extra days ties up about 2.5 million dollars, costing 500,000 dollars a year to hold.

Forgetting switching cost makes a supplier change look profitable when it is not. Symptom: the new vendor is 0.35 dollars per part cheaper, but the payback never shows up in year one. Root cause: tooling transfer at 85,000 dollars, PPAP requalification, dual-running both sources for 8 weeks, and a scrap spike during ramp were left out. Fix: the Supplier Switching Cost tool sums one-time charges. If total switching cost is 240,000 dollars and savings run 0.35 dollars on 400,000 units, that is 140,000 dollars a year, so real payback is 1.7 years, not immediate.

Unit and basis mismatches silently corrupt comparisons. Symptom: two contract quotes are off by 3x for no obvious reason. Root cause: one is priced per piece, the other per 1,000 pieces, or one includes secondary machining and the other does not. A more subtle version: comparing a fully burdened internal rate against a supplier price that excludes your own inbound freight and receiving labor of 0.22 dollars per part. Fix: in the Contract Manufacturing Comparison, normalize everything to cost per finished, delivered, inspected unit in one currency before any decision, and lock the exchange rate assumption you used.

Assuming outsourced capacity is free and instantly scalable breaks the plan under demand swings. Symptom: you outsourced the overflow, then peak season the supplier could not deliver and you air freighted at 4x cost. Root cause: the Capacity Outsourcing Gap was sized to average demand, not the P90 peak. If average is 30,000 units a month but peak hits 46,000, a gap sized at 5,000 leaves you 11,000 short. Fix: size outsourced capacity to peak minus reliable internal capacity, then verify the supplier's committed volume floor and surge terms in writing before removing your own equipment.

Treating insourcing payback as a pure price delta ignores ramp inefficiency and utilization risk. Symptom: the Insourcing Payback model showed 2.1 years, but three years in you are still behind. Root cause: the model assumed 85 percent utilization and day-one standard cost, while real ramp ran 55 percent utilization for 9 months and yields sat 6 points low. Fix: stage the payback with a learning curve. Model months 1 to 6 at 60 percent of target throughput and 1.4x scrap, and require the machine to run above 70 percent utilization or the fixed 620,000 dollar capex never earns out.

Building network decisions on stale or single-point data guarantees a wrong answer. Symptom: the Supplier Network Cost model still uses last year's 0.42 dollar per part freight and a fuel surcharge that has since moved 30 percent. Root cause: one static snapshot, no sensitivity range. Fix: run every make-buy and Outsourcing ROI case across a band, freight plus or minus 20 percent, FX plus or minus 8 percent, volume plus or minus 15 percent, and check whether the decision flips inside that band. If a 10 percent freight move reverses the recommendation, the two options are effectively tied and you should not pay a switching premium to chase it.

Published 2026-07-01.