BOM Health Score: The Five Metrics We Track on Every Customer's Bill of Materials

Most BOMs arrive at our quoting desk as a CSV with manufacturer part numbers, descriptions, and per-board quantities. That's enough to price a build; it isn't enough to tell anyone whether the programme will still be buildable in eighteen months. Before any NPI design review, we score the BOM on five axes and put the numbers on a single sheet that goes to the customer's procurement lead and engineering manager together.

The five metrics, in the order we calculate them:

• Sole-source count — lines with exactly one approved manufacturer part number. Anything above 15% of total line count gets flagged red.
• Lifecycle-status mix — what fraction of lines are Active, NRND (not recommended for new designs), or Last-Time-Buy as called out by the manufacturer. We pull this nightly from Silicon Expert and IHS feeds.
• Lead-time exposure — the dollar-weighted average lead time across the BOM, plus the count of lines exceeding 26 weeks.
• Qualified-alternates ratio — for each line, do we have a customer-approved second source with a verified footprint match. Numerator: lines with approved alts. Denominator: total lines.
• Per-line margin sensitivity — if any single line doubles in price, what does it do to landed unit cost. We rank the top 10 most-sensitive lines and flag them for hedging or alternate qualification.

"A healthy BOM isn't the one with the cheapest parts. It's the one where I can still build twenty thousand boards if any two suppliers blink." — Pioneer Horizon sourcing team

The whole scorecard runs in under an hour for a BOM of 200–400 lines once it's loaded into our procurement database. The hour buys back weeks of back-channel firefighting at the next allocation event.

Sole-source count and lifecycle-status mix tend to drift together — a BOM that's been carried unchanged for three years almost always sees both numbers worsen. Lifecycle drift is the slower poison. A part that was Active when the design was released can quietly slip to NRND a year later without anyone on the engineering side noticing until the distributor stock count starts dropping.

What we flag as red

• Sole-source > 15% of total line count, or any sole-source line above $5 unit price.
• NRND fraction > 8% — manufacturer-declared end-of-design-in is the loudest available signal that obsolescence is coming.
• Any Last-Time-Buy line still in active production without a qualified successor in CAD.

What we flag as amber

• Sole-source lines between 8% and 15%.
• Any line where the original manufacturer has been acquired in the last 24 months (post-merger SKU pruning is real and predictable).
• Distributor stock below 26 weeks of forecast demand across the top three franchised channels combined.

On a recent industrial-controls customer, the BOM scored 22% sole-source on first audit. After eight weeks of alternate-qualification work — most of it MLCC and resistor housekeeping plus three real MCU substitutions — we pushed it to 9%. The next allocation wave hit four months later. Their build schedule didn't slip; two competitors on similar designs lost a quarter of fab capacity. The eight-week investment paid for itself before the year was out. For deeper coverage of lifecycle signals, see our predicting obsolescence article.

A naive average lead time across a BOM tells you almost nothing. A 400-line BOM with a single 52-week MCU and 399 stock items will average around 1.4 weeks, which is the wrong number to plan around. We weight lead time two ways and report both.

Dollar-weighted average

Per-line cost × lead time, summed and divided by total BOM cost. This gives you a planning horizon for working-capital purposes — how long, on average, your money is in motion before product can ship. For a healthy electronics BOM in 2026 we'd expect 6–10 weeks; above 14 weeks the financing cost starts to matter.

Critical-path lead time

The single longest-lead line that doesn't have an in-stock alternate. This is the number that drives production planning. If that line is 38 weeks, your build cadence is gated to 38-week buying cycles regardless of what the rest of the BOM looks like.

Lines we count as exposed

• Lead time > 26 weeks with no qualified alternate in stock.
• Lead time > 16 weeks combined with sole-source status.
• Any line where distributor pricing has changed by more than 20% in the last 90 days (volatile pricing is a leading indicator of allocation).

On the snapshot side, we maintain a running MCU allocations watchlist that feeds into this scoring automatically. If a part on your BOM gets added to that list, your scorecard refreshes the same week.

The qualified-alternates ratio is the metric most BOMs score worst on, and it's the metric most worth fixing. "Qualified" doesn't mean "the engineer thinks it would probably work." It means footprint-matched, datasheet-compared on the parameters that drive the design, and ideally with a functional build behind it.

Our qualification tiers

• Tier A — fit-form-function equivalent, same footprint, same critical parameters. Drop-in. We hold Tier A on every passive line and on as many actives as practical.
• Tier B — same footprint, parameter-compatible within tolerance. Requires a small engineering review at substitution time but no schematic change.
• Tier C — different package or different supplier silicon. Requires a layout-aware review and possibly a board spin. We document these so the work isn't redone from scratch under pressure.

Margin sensitivity — the ten lines that move the unit cost most

For each line, we run a 2× price perturbation and recompute landed unit cost. The ten lines with the largest absolute impact get a separate scorecard. These are the lines worth hedging on — long-term agreements with one franchised distributor, a buffer-stock policy at our warehouse, or in extreme cases a customer-funded last-time-buy.

On a typical 300-line BOM, the top ten sensitivity lines drive about 65% of unit-cost variance. Spending procurement attention proportionally is the cheapest risk reduction available.

The scorecard isn't a one-time document; if it's only run at NPI it stops mapping to reality within a year. We refresh on three triggers and ship the new version with the next production-build documentation pack.

Refresh triggers

1. Quarterly cadence — every BOM in active production is rescored on a 90-day cycle regardless of changes.
2. Manufacturer lifecycle event — any PCN, NRND notice, or LTB notice for a part on the BOM triggers an immediate rescore and an email to the engineering and procurement leads on that programme.
3. Pre-build refresh — within seven days before any production build of more than 1,000 boards, the procurement team runs a delta scorecard against the last archived version. Anything that has moved gets a written note explaining the change and the mitigation.

The single-page format

• Five scores at the top, each with a colour band and a 90-day delta arrow.
• Red-flag line table — at most twelve lines, sorted by combined risk score.
• Action register — what we owe the customer, what the customer owes us, and what's blocking each.

The format is deliberately blunt. Procurement directors don't read fifty-page risk reports; they read a green/amber/red row and ask the right follow-up question. The scorecard's only job is to surface the right follow-up question fast.

If you'd like us to score one of your live BOMs against this rubric — no quote attached, no commitment — share it with our sourcing team and we'll come back inside five working days with a marked-up scorecard and a top-ten remediation list.