Somewhere in your supplier list is a shop you've sent eleven RFQs and awarded zero. They keep quoting. You keep asking. Neither side has noticed.
That relationship costs both of you, and the arithmetic says how much.
Win rate, and what a zero means
win_rate = awards / quotes_submitted
| Supplier | Quotes | Awards | Win rate |
|---|---|---|---|
| Apex Precision | 14 | 6 | 43% |
| Titanium Forge | 12 | 4 | 33% |
| Vanguard Plastics | 9 | 3 | 33% |
| Summit Hydraulics | 11 | 0 | 0% |
A healthy competitive set lands roughly 25–40% each. Sustained zero means one of three things, and they need different responses:
- Wrong capability. They're being invited to jobs they can't price well. Stop inviting them for this family; keep them for what they're good at.
- Structurally uncompetitive. Higher rate, different market. Remove them.
- They stopped trying. After enough losses, quotes become polite noise — padded numbers to avoid saying no. That number is not a price, and it is quietly distorting your spread analysis.
The third case is the expensive one, because a padded quote inflates your apparent market range and makes a mediocre winner look competitive.
What the extra bidder is worth
The counter-argument is that another quote costs nothing. It isn't quite free — it costs the supplier real estimating hours, and it costs you a slower cycle — but the more interesting fact is that its value falls off a cliff.
The expected lowest of n quotes, drawn from a distribution with mean μ and standard deviation σ, is approximately:
E[min_n] ≈ μ − σ × Eₙ
where Eₙ ≈ 0.56 (n=2), 0.85 (n=3), 1.03 (n=4), 1.16 (n=5), 1.27 (n=6)
With the bracket from our spread analysis — μ = $11.47, σ = $2.64:
| Quotes | Expected best price | Gain from one more |
|---|---|---|
| 2 | $9.98 | — |
| 3 | $9.24 | $0.74 |
| 4 | $8.75 | $0.49 |
| 5 | $8.40 | $0.35 |
| 6 | $8.12 | $0.28 |
The third quote earns 50% more than the fourth. The sixth earns a third of what the third did. At 400 pieces a year, that sixth quote is worth $112 annually — against the estimating hours you asked a supplier to donate, and the days added to your cycle.
Three is usually right. Four when the spread is wide or the part is new. Six is a habit, not a strategy — and the sibling post on how many suppliers to RFQ makes the same case from the relationship side.
The number nobody tracks
Response rate is the one that predicts trouble earliest:
response_rate = quotes_returned / rfqs_sent
A supplier at 100% win rate but 40% response rate isn't a great supplier — they're a supplier who only bothers when the job suits them, and you're carrying the coordination cost of finding out which jobs those are.
How this gets solved
Every number here is a count of things you already did. The reason almost nobody has them is that the data lives in sent-mail folders, and nothing joins "we asked" to "they answered" to "we awarded".
RFQTrace records those three events per supplier as a side effect of running the RFQ — invitation sent, quote received, award made — so win rate, response rate and average lead time build themselves. No new process, no scorecard spreadsheet somebody has to maintain: just the ability to answer, before you send the next RFQ, which of these eleven shops has ever said yes.