Lead Scoring Criteria for Manufacturing in 2026: The Signals That Predict a Real Order

An RFQ lands at nine in the morning. A company name nobody recognises, a first name, a free email address, and one line of text: please quote 500 units. There's no drawing attached, no material called out, and no date anyone has to hit. Someone now has to decide whether an estimator loses the afternoon to it.

That decision is lead scoring, whether or not anybody in the building uses the phrase.

The trouble is that almost everything written about lead scoring was written for software companies. A software buyer reads three articles, downloads a report, opens four emails and books a demo, and by then the system knows plenty about them. A buyer sourcing a machined bracket does none of that. They find you, they send one enquiry, and they send the same enquiry to four other shops the same week.

You get one message. Everything worth knowing is either inside it or missing from it, and the missing parts tell you as much as the rest.

Why Manufacturing RFQs Break Generic Lead Scoring

Open any standard scoring template and you'll find rows for email opens, page views, webinar attendance and content downloads. Those work when a buyer spends six weeks circling before they talk to you. Manufacturing doesn't give you six weeks. Enquiries arrive cold, from people who already know what they want and are checking whether you can make it.

What a Whitepaper Download Tells You About a Machine Shop

Almost nothing. A procurement engineer downloading your capability statement might be sourcing a million-dollar programme or writing a university assignment, and the download itself doesn't separate them.

Behavioural scoring assumes a long, visible courtship. Manufacturing gives you a short, mostly invisible one, so a score built on engagement ends up ranking the curious above the committed. That's the failure mode behind most of the manufacturing funnels that get built and then quietly ignored by sales.

The signals that matter in manufacturing sit inside the request, not around it.

The Four Signal Families That Move a Quote Forward

Every RFQ worth winning answers four questions, and the ones that never turn into orders usually leave at least two of them blank.

1. Can we actually make this? Part type, material, tolerance, finish, process.

2. Is it worth the setup? Annual quantity against a one-off, and whether repeat business follows.

3. Is anything actually happening? Delivery window, project stage, whether a budget exists.

4. Is this a real buyer? Drawings attached, specs written down, a company domain, a named role.

Those four are the spine of the model. Everything below hangs off them. If you want the generic version of the argument first, our guide to how to filter out low-intent leads covers the principle across industries.

The Manufacturing Lead Scoring Criteria That Predict an Order

Here's a starting model. Adjust the weights to your shop, because a job shop running short-run prototypes and a contract manufacturer chasing annual agreements want almost opposite things, but the shape holds.

Fit Criteria: Part Type, Material, Tolerance and Process

Fit is the cheapest thing to score and the most often skipped, because it feels obvious until you count how many quotes you lose on it.

The negative rows are doing the real work here. A tight-tolerance aerospace part from a buyer with money is still a bad lead if you can't hold the tolerance, and scoring it highly means an estimator spends a day proving so. A day you don't get back.

Volume Criteria: Annual Quantity Versus a One-Off Prototype

Quantity is the row most shops weight wrongly, usually by giving it more points than anything else. Big quantities from buyers who never order are the most expensive thing in an estimating department.

A prototype enquiry that names the production volume behind it is worth more than a large one-off, and most scoring models get that backwards.

Timing Criteria: Delivery Window and Project Stage

Timing is where the tyre-kickers separate themselves. A buyer with a real date behind them tells you the date, every time.

Completeness Criteria: Drawings, Specs and Who Is Asking

This family is the closest manufacturing has to intent data, and it costs nothing to collect.

Completeness works as a proxy because effort is reciprocal. Somebody who attaches a drawing and writes out their tolerances has already spent twenty minutes on you. People rarely spend twenty minutes on a supplier they only want a number from.

Every row here is a question you can ask at the point of enquiry. That's what lets you qualify leads inside the funnel instead of emailing back and forth for three days to find out the job was never real. If you're gating a capability guide or a materials selector to capture these answers, our roundup of lead magnets for manufacturers covers what manufacturers actually download.

Turning Manufacturing Scoring Criteria Into Point Values

A criteria list is easy. Assigning numbers to it is where most models stall, because the numbers feel arbitrary and nobody wants to defend them in a sales meeting.

Weight the Criteria That Killed Your Last Ten Quotes

Don't start from a blank page. Pull the last ten quotes you lost, and the last ten you won, and ask what was visible at enquiry stage in each one.

I'd weight the criteria that already cost you something ahead of anything else in the model, and that's an opinion rather than a finding. If four of your last ten losses were tolerance you couldn't hold, tolerance deserves a bigger negative than anything else in your model. Theory won't beat that.

The weights aren't guesses if you take them from your own lost quotes.

Negative Points for the RFQ You Do Not Want

Most published scoring models are all upside, which is why they rank everything highly and help nobody. Negative scoring is what creates separation at the bottom, and the bottom is where your estimating hours leak.

Three rules that make negatives work:

1. A hard blocker gets a number big enough to sink the lead on its own. A certification you don't hold should drag a score below your floor even if everything else is perfect.

2. Absence scores as well as presence. A missing date, a missing drawing and a free email address should each subtract, so an empty enquiry sorts itself to the bottom without anyone reading it.

3. Keep the negatives visible to sales. A lead that scored badly for one fixable reason is worth a two-line reply asking for the drawing, and that only happens if the reason is on screen.

The same scoring logic sits behind any assessment builder, whether it's grading a respondent or grading an enquiry.

Where the Two Thresholds Sit

You need two numbers, not one. The upper threshold is the score above which an estimator starts work today. The lower is the score below which nobody touches it and an automatic reply asks for what's missing.

Set them by volume rather than by feel. If you can quote fifteen enquiries a week and forty arrive, the upper threshold sits wherever the fifteenth-best one falls. Any threshold above your capacity is a wish.

Capturing the Signals in an involve.me RFQ Funnel

Every criterion above is worthless if the answer arrives three emails later. A contact form with a message box collects none of it, which is why the scoring model and the intake form have to be designed together in involve.me rather than one after the other.

Ask Only the Questions That Carry Points

Work backwards from the model. Each question earns its place by feeding a row in the table. Anything that doesn't feed a row is a field you're asking a busy procurement engineer to fill in for your convenience.

For most shops that's six to nine questions: what you need made, material, quantity, annual volume, tolerance, required date, project stage, drawing upload, and who you are. That reads as a lot until you compare it against the four-email exchange it replaces.

A practical constraint worth knowing before you build: in involve.me, per-answer point values come from the "Individual Score & Calculation" option, which is available on the Single/Multiple Choice, Image Choice and Dropdown elements. Questions you want to score should use those element types. A free-text field can still be collected and read, it just can't carry points.

Scoring the Answers With Score-Based Outcomes

involve.me has three funnel types, and this one wants Score-based Outcomes, where the total decides which page a respondent lands on. Set the score element to "Individual Score" so it sums the values you assigned. The other setting, "Correctly Answered Questions", counts right answers at one point each and is built for quizzes.

For anything more involved than adding up answer values, a quantity times rate calculation, a setup charge, a price break at volume, you want the calculator element instead. That's the same machinery behind a price quote calculator, and it's worth knowing the two are separate: the score element adds up answers, the calculator element runs formulas.

Showing the Buyer a Figure While You Score Them

Here's the part that makes people fill the form in. A buyer who gets an indicative figure on screen, a ballpark unit price, a lead-time estimate, a cost band, has a reason to answer nine questions accurately rather than abandoning at question three.

The trade is honest in both directions. They get a number they can take to their manager today, and you get the tolerance, the quantity and the date you'd otherwise have chased for a week. If you're weighing up how to build that part, we've weighed up the best calculator builders elsewhere on the blog.

Routing a Scored Manufacturing Lead

A score that sits in a spreadsheet changes nothing. What makes the model pay is what happens in the ninety seconds after someone hits submit. In involve.me that's an outcome page plus a workflow.

Three Outcome Pages, One for Each Band

Give each band its own page and write it as the next step rather than as a verdict.

• Above the upper threshold. Confirm you can make it, give the realistic quote turnaround, and put the calendar on the page.

• Between the thresholds. Acknowledge, ask for the one or two things that would move them up, usually the drawing or the date.

• Below the lower threshold. Send the capability guide, the materials list, the standard lead times. Nobody is offended and no estimator is involved.

The middle band is where the money is. A lead sitting between your thresholds is usually one missing attachment away from being a good one.

The Internal Alert an Estimator Will Actually Read

A "Send Internal Email" step is what puts a hot RFQ in front of a human while the buyer is still at their desk. Write the subject so it can be triaged from a phone: the score, the part, the quantity, the date. An alert that says "new form submission" gets read once and then filtered.

Then let the automated email sequences carry the rest. Build one workflow on the "Completed Submission" trigger for that funnel. The acknowledgement is the first "Send Email" node, sitting directly on the trigger. After it, a "Conditional Logic" step splits on the score and a "Wait" sets the gap before each follow-up. High scorers get a nudge toward your calendar so they book qualified meetings without another round of email. Low scorers get useful material and no salesperson.

Score Bands Need Two Conditions, Not One

Worth knowing before you build the workflow: a condition compares a variable against a single value using is, is not, greater than or less than, and there's no numeric range operator. So a middle band is two conditions joined with AND, a score greater than or equal to your lower threshold and less than your upper one, rather than one condition describing the range. A "Conditional Logic" step also always carries an else path, which is the right home for anything that somehow scored outside every band.

Validating the Model

A scoring model is a hypothesis about which enquiries become orders. Most shops write one, feel better, and never check it again. Check it.

Give it a quarter, then put your won orders next to the scores those enquiries got on arrival. The comparison takes an afternoon. If your high scorers close at roughly the same rate as everything else, the model is sorting on something that doesn't predict an order. Usually one row carries far too much weight.

Two questions come up when people ask us about scoring, and neither is about the points: whether a results page can change based on what somebody answered, and how to set up the calculation behind a score-based outcome. I expected the hard part to be choosing the criteria. In our own support data it's the outcome page, every time.

So build the three pages before you tune a single weight. The bands will move; the pages they point at won't.

We can see a lead arrive, get scored and get routed. Whether an estimator quoted it and whether that quote turned into a purchase order happens in your ERP, where we can't see it, and that's the number worth having. That close-rate comparison is the one I'd run before touching a single weight.