Sole Source: the $900k median problem

Pick a supplier tier and see what the dependency actually costs when a disruption hits. Across 3,000 synthetic disruption events, sole-sourced parts carry a median cost of $901k2.11× an approved supplier’s $427k. Toggle severity bands to test whether the premium is real, and watch why it costs more: not a slower clock, but empty shelves and lost revenue.
Illustrative, not a forecast. Built on synthetic, observational data (xpertsystems/mfg006-sample, n = 3,000). Findings are associations, not causation. Use it to size the exposure of a sourcing posture — not to predict any single disruption’s bill.
Median disruption cost
$901k
n = 374 events
vs approved baseline
2.11×
+$474k above approved’s $427k
Stockout rate
41.4%
approved: 17.8% · 2.3×
Median revenue loss
$754k
approved: $356k · 2.12×

Median disruption cost by supplier tier

Bars span the middle 50% of events (P25–P75); the tick marks the median. The dashed line is the approved baseline ($427k). Four tiers bunch around $427k–$501k — sole-source is the lone outlier at $901k. Cost is right-skewed, so we lead with medians.

Why it costs more — and the surprise

The premium is not a slower recovery clock (median 6 days either way). It is stockouts and lost revenue — the damage lands at the customer end. Your selected tier is in blue.

Stockout rate

% of events that ran the part dry

Median revenue loss

revenue that walked out the door

Median recovery (days)

the clock is flat — that’s the point

Is the premium real, or just worse luck? Test it within severity bands

Sole-source vs approved median cost, re-computed within each band. The premium holds at equal severity and grows: 2.19× (low), 2.17× (medium), 2.31× (high). Pick a band at left to highlight it.

The checkbox is theater. The dependency is the lever.

How much each “knob” actually moves the median disruption cost. The dual-source flag barely moves it (and the wrong way). Who you depend on moves it 10.7× as much.

What this says about your sourcing

Data & method

Source: xpertsystems/mfg006-sample — 3,000 synthetic supply-chain disruption events (113 columns). License CC-BY-NC-4.0 (attribution required, non-commercial). Dataset: huggingface.co/datasets/xpertsystems/mfg006-sample. Credit: xpertsystems.

Cost field is cost_of_disruption_total_usd. All headline figures are medians: cost is right-skewed (whole-dataset median $512,738 vs mean $1,912,836, a 3.73× gap), so means are outlier-inflated and shown only to expose the fat tail. A log-cost regression with a sole-source indicator gives +108.2% raw (adj R² 0.02); adding controls for severity and disruption type shrinks it to +67.5% (p = 9.7×10⁻¹⁴, adj R² 0.42). Sole-source faces more critical events (11.5% vs approved’s 3.8%), which explains part of the raw gap — not most of it. This is observational, synthetic data: association, not proof of cause.

Framing references (not the source of any number here): Sheffi, The Resilient Enterprise (MIT Press, 2005); Christopher, Logistics & Supply Chain Management, 4th ed. (Pearson, 2011); ISO 31000:2018, Risk management — Guidelines.