Many RFQ scores fail because they convert a salesperson’s optimism into a number. The opportunity is “strategic,” the relationship is “strong” and the team will “find capacity.” Add the weighted fields and the request earns 82 out of 100. Nothing in the score tells estimating whether to start or engineering what it still needs to prove.
A useful score has evidence, owners and an action. It moves a supported request into the bid queue, sends an unanswered technical or commercial question to a named person, or stops the pursuit. It also changes as the customer, engineering team and suppliers replace assumptions with facts.
Decide what the score controls
Do not build a score until the team agrees what each result does. A practical first use is to allocate estimating and engineering time. The score can rank feasible RFQs, but it should not approve a certification exception, final price or customer promise.
Use four outputs. “Ready” enters the funded bid queue. “Hold” creates named actions and a review date. “Exception” goes to the person with authority over the failed condition. “No-bid” records the reason and closes the work. If every score still enters the same queue, remove the score and fix the decision process first.
| Output | What happens next | What the record must show |
|---|---|---|
| Ready | The bid leader funds the next estimate or proposal step | Evidence, score, bid budget and next gate |
| Hold | The request waits for a specific fact or action | Question, owner, due date and affected factor |
| Exception | The accountable authority decides whether work may continue | Failed rule, consequence and proposed condition |
| No-bid | Sales closes or declines the request | Reason, approver and customer response |
Run stop conditions before the weighted score
A weighted total cannot repair a bid the company may not or cannot deliver. Check legal and trade restrictions, mandatory certification, technical feasibility, minimum information, delivery credibility and commercial authority first. Route an exception when policy permits one. Otherwise stop.
This separation prevents a large deal or preferred customer from mathematically cancelling a fatal condition. The bid/no-bid framework explains the three decision gates and the owners behind these stop conditions. The RFQ score begins after the request clears that gate.
Use six factors and one clear direction
Rate each factor from 0 to 3, where a higher score always means a better pursuit. Keep the direction consistent. If “risk” earns more points when risk is high while “economics” earns more when economics are good, the total becomes easy to misread. Call the factor “scope and risk position” and give 3 points when the important uncertainties are small, supported or controlled.
PMI’s research on bid/no-bid decision criteria covers resources, company fit, the probability that the project proceeds, competition and risk. The model below turns those ideas into the questions an industrial OEM can answer from an RFQ, its systems and the people who must deliver the order.
| Factor | Weight | Evidence |
|---|---|---|
| Business fit | 15% | Named product, market, account and aftermarket priorities |
| Project reality | 15% | Budget, buying process, decision date, site, sponsor and project status |
| Buyer position | 20% | Access, installed base, prior delivery, specification position and competitors |
| Technical and delivery fit | 20% | Product envelope, engineering load, bottleneck capacity, supplier plan and requested date |
| Economics | 20% | Price range, cost range, margin, cash terms, bid cost and constrained capacity |
| Scope and risk position | 10% | Open technical, schedule, cost and contract questions; owners and mitigations |
These weights are a starting model, not an industry benchmark. Change them when actual bids show that a factor drives good or bad orders. A company with one scarce test cell may give delivery fit more weight. An OEM entering a new product market may deliberately give business fit more weight.
The buyer will also score the offer, whether or not it shares the worksheet. A NIST Manufacturing Extension Partnership guide tells buyers to examine certifications, workforce, equipment limits, production timeline, logistics, cost breakdown, warranty, intellectual property, insurance, financial health and relevant projects. Use that view when you rate buyer position. Ask which of those questions the customer cares about and what evidence makes your offer stronger than the alternatives.
Write an anchor for every number
A scale without anchors measures the reviewer more than the RFQ. Write what each point means in terms the owner can prove. Use product-family anchors where the constraint differs. A cast-equipment line may care about foundry capacity; a switchgear line may care about engineering approval and component availability.
| Technical and delivery fit | Anchor |
|---|---|
| 0 — infeasible | The requirement or date fails a stop condition; no approved exception exists |
| 1 — weak | The plan needs unconfirmed engineering, expedite, overtime, a new supplier or displacement of committed work |
| 2 — workable | The product path and current plan support the request; named risks need normal action |
| 3 — strong | Released product knowledge, current capacity and confirmed suppliers support the request with normal contingency |
Do the same for all six factors. For buyer position, 0 might mean no access beyond an unknown procurement contact and no supported differentiator. A 3 might mean direct access to the decision team, a documented performance advantage and a position in the customer specification. “Good relationship” is not an anchor.
Keep “unknown” separate from zero
Zero means the evidence is bad. Unknown means the team does not have the evidence. Do not score an unknown as 0, because the total then claims the factor was evaluated. Do not score it as the midpoint, because that rewards incomplete requests. Leave it open and calculate coverage separately.
A simple coverage measure is the percentage of weighted points supported by current evidence. If economics and delivery fit remain open, the score may cover only 60% of the model. The bid leader sees both numbers: 74 out of 100 on the answered factors, with 60% coverage. Policy can then require, for example, full coverage on delivery and economics before the request becomes ready.
Every unknown should become a question with an owner and due date. If the customer deadline forces an assumption, record who accepted it and rerun the affected factor when the answer arrives.
Use live capacity, not last quarter’s impression
Delivery fit expires quickly. An open production slot can disappear tomorrow. A supplier can move its lead time after another customer places an order. The score should show when the capacity, material or supplier fact was checked and who supplied it.
Do not ask sales to guess whether operations can fit the order. Pull the current backlog, engineering load, material position and bottleneck schedule where those records exist. Then ask the accountable planner or engineer to resolve the exceptions. This makes the score a shared decision instead of a sales forecast wearing factory language.
Add economics without pretending the estimate is final
At the first gate, economics may be a range. Use the best current cost basis, name the uncertain drivers and show how they could move the result. Do not award full points because the target price clears a margin floor before anyone has tested the custom scope.
The GAO Cost Estimating and Assessment Guide calls for sensitivity analysis on the assumptions and parameters that can move an estimate. It also says the range should come from supported data rather than an arbitrary plus-or-minus percentage. For an OEM bid, test the cost drivers that are both large and likely to change: material quantity, bought-out equipment, labor hours, test scope, freight, schedule and performance assumptions.
For queue ranking, calculate expected bid value as the probability of win multiplied by expected order contribution, less bid cost. Show the inputs and their ranges. Then compare the result with the scarce resource the bid and order will consume. A large expected contribution can still be a poor use of the only test slot available that quarter.
Worked example: a packaging line RFQ
An existing food customer asks for a packaging line before a planned shutdown. The project is funded and the OEM has access to operations and procurement. Most of the design uses released modules, but the washdown requirement needs an engineering exception. The site has not supplied the final line-speed profile, and controls engineering is near capacity.
At intake, the team can score five factors. It leaves scope and risk open because the line-speed profile controls equipment selection. It also gives delivery fit 1 because the plan depends on unconfirmed controls capacity. The request stays on hold even if the answered factors produce an attractive partial score.
| Factor | Weight | Rating | Weighted points | Evidence |
|---|---|---|---|---|
| Business fit | 15 | 3 | 15.0 | Named target account and released product family |
| Project reality | 15 | 3 | 15.0 | Funded shutdown project with decision date |
| Buyer position | 20 | 2 | 13.3 | Direct access; competitor still active |
| Technical and delivery fit | 20 | 1 | 6.7 | Washdown exception and unconfirmed controls slot |
| Economics | 20 | 2 | 13.3 | Cost range supports the floor; bought-out conveyor price open |
| Scope and risk position | 10 | Unknown | — | Final line-speed profile missing |
The customer supplies the line-speed profile. Engineering confirms that the released drive package works, and planning assigns a controls engineer. Technical and delivery fit rises to 2. Scope and risk position becomes 2. The complete score is 77. The request clears a sample 70-point queue threshold, with two conditions carried into costing: the washdown exception and the open conveyor price.
The missing line-speed evidence, the owners and the second review determined when the request entered full estimating. The score exposed the gap and showed what changed when the team found a credible path.
Give each factor to the person who knows it
Sales owns project reality and buyer position. Product or application engineering owns technical fit. Operations owns current capacity and the delivery plan. Estimating and finance own the cost, contribution and commercial thresholds. The bid leader resolves the whole case and records the decision.
Do not average away disagreement. If sales rates delivery fit 3 and operations rates it 1, show both inputs and send the factor to the operations owner. The conversation matters more than a smoother number. Preserve the earlier rating when the owner changes it so the team can later see which facts improved or deteriorated.
What Bourne does with the score
Bourne reads the checked RFQ and assembles the evidence from customer history, similar bids, released product data, current capacity, supplier responses and commercial rules. It calculates the agreed factors, leaves unsupported inputs open and routes each question to the person who owns it.
The bid leader sees the score, evidence coverage, stop conditions and changes since the last review in one place. They can fund the bid, hold it, decline it or record an override. A ready request enters product costing with the accepted assumptions and risks attached. The queue shows why one RFQ moves before another.
Calibrate the score against wins and delivered orders
Take a year of submitted, declined, won and lost requests. Recreate each score using only the evidence available at the decision date. Then compare the result with bid hours, win or loss, quoted margin, delivered margin, schedule performance, change orders and warranty cost.
Look at score bands instead of one average. If bids scoring 80–100 do not win more often than bids scoring 60–79, the buyer-position factors may be weak. If high-scoring wins deliver poor margin, the economics or risk anchors are too generous. If many profitable orders began below the threshold, inspect which factor rejected them before lowering the bar.
| Measure | Question it answers |
|---|---|
| Evidence coverage at each gate | Are people making the score from facts or defaults? |
| RFQs by ready, hold, exception and no-bid | Does the model change work allocation? |
| Win rate by score band | Does the external evidence separate stronger pursuits? |
| Bid hours by score band | Is the team spending effort on the requests it ranked highest? |
| Quoted versus delivered margin | Does the score find internal cost and execution risk? |
| Override rate and outcome | Do exceptions create good orders or bypass the model? |
Change one rule at a time and date the version. Otherwise the business cannot tell whether a new weight, a market shift or a different mix of RFQs changed the result. The score should get better because the company learns from its own bids and orders, not because the spreadsheet gained more fields.
Arda Bulut
Arda Bulut is the co-founder and CTO of Bourne and HockeyStack. He leads engineering at Bourne, building the platform people use to create AI products, agents and automations.