Guides
Job Shop Quoting and Estimating
What quoting actually costs per won job, why the estimate gets thrown away, the software categories, and how to estimate from history instead of memory.
Quick answer: quoting is the most expensive unbilled activity in a job shop and almost nobody measures it. The estimate is also the first structured record of a job, and most shops throw it away and retype a fraction of it. Fixing those two things is worth more than any estimating software you could buy.
This guide covers running the estimating function. For what number to actually put on the quote, see how to price welding and fabrication and how to calculate your shop rate.
What Quoting Costs You
The cost of quoting is carried by the jobs you lose, which is why it stays invisible.
Work it through with realistic numbers:
| Value | |
|---|---|
| Estimator loaded cost per hour | $67 |
| Average hours per quote | 4 |
| Cost per quote | $268 |
| Win rate | 1 in 5 |
| Quotes per win | 5 |
| Estimating cost per won job | $1,340 |
Now put that against the job. If the average won job is $18,000 at a 12 percent margin, gross margin is $2,160. Estimating consumed $1,340, or roughly 62 percent of it, before anyone picked up a grinder.
That is not an argument for quoting less. It is an argument for knowing the number, because two levers move it and both are available today:
| Change | Estimating cost per won job |
|---|---|
| Baseline, 4 hours, 1 in 5 | $1,340 |
| Win rate improves to 1 in 4 | $1,072 |
| Quote time halves to 2 hours | $670 |
| Both | $536 |
Halving quote time is usually easier than improving win rate, and it is entirely within your control. Most shops chase the win rate and ignore the hours.
Measure These Four Things
Almost no small shop tracks any of them, and all four come out of a quote log you could start this week.
- Quotes sent per month. The denominator for everything else.
- Win rate, overall and by customer. One customer at a 5 percent win rate is training your estimator for free.
- Hours per quote. Even roughly. Estimators resist this and it is the number that moves.
- Time from request to quote sent. On competitive work, being second is frequently the whole reason you lost.
The fourth one surprises people. When a customer needs a number and you take five days, you can be the best price and still lose, because they awarded it on day three.
The Estimate Is the First Record of the Job
This is the structural point, and it is the front-end twin of the shop floor problem covered in our metal fabrication guide.
A completed estimate already contains almost everything the job needs:
- The material list, with sizes and quantities
- The hours, broken out by operation
- The assumptions, meaning what was included and excluded
- The drawing revision it was priced against
Then the job is awarded. In most shops, someone opens a new job in the accounting package, types in a job number, a customer, and a total value, and every other piece of that structure is left behind in a spreadsheet or a PDF.
Two things follow, and both are expensive.
Estimate against actual comparison becomes impossible. You can compare the total, which tells you the job made or lost money. You cannot compare by cost line, which is the comparison that tells the estimator what to change next time. The knowledge loop never closes.
The material list gets rebuilt. Purchasing retypes what the estimator already worked out, with the transcription errors you would expect.
Whatever you use, the fix is the same in principle. The estimate has to hand its structure to the job, and handing over only the total is what breaks the loop. Odoo does this through quotation templates that carry through to the order, and dedicated estimating packages do it through an export into the production system. A spreadsheet can do it too, if the job sheet is built from the estimate sheet instead of retyped.
Estimate From History, Not From First Principles
The best estimators do not calculate every job from scratch. They remember what similar work actually cost and adjust.
That is a database, and it is sitting in one person’s head. Which means it walks out at retirement, cannot be checked, and cannot be handed to a second estimator.
Building the real version requires two things you may already have most of:
- Job cost history that is queryable by similar work, so “what did the last three tank stands cost” is a question with an answer.
- Actual costs recorded by cost line, so the answer is useful instead of being a single total.
Our job costing spreadsheet produces exactly that structure for free, and job costing covers the reporting discipline behind it.
Shops that build this quote faster and win more, for a reason worth stating. Confidence lets you price tightly. An estimator guessing adds a pad to protect themselves, and that pad loses jobs you could have profitably done.
The Software Categories
Four things get called estimating software and they do different jobs.
| Category | What it does | Where it fits |
|---|---|---|
| Structured spreadsheet | Your own logic, formalised, with material and labour tables | One estimator, low to moderate volume. Free and frequently correct |
| Takeoff software | Quantities off drawings or models, sometimes from 3D | Structural and pipe work where measurement is the slow part |
| Dedicated fabrication estimating | Material databases, labour standards, assemblies, often live steel pricing | High volume, multiple estimators, repeatable product families |
| ERP quotation | Quote built from products and BOMs, flowing into the order and the job | Where the quote-to-job handoff matters more than estimating speed |
The two ends of that table are frequently combined. Takeoff or dedicated estimating produces the number, then it lands in the business system as an order. Expecting one product to do both well is where budgets go.
Nesting deserves a specific mention, because it gets promised. General business systems do not nest. Searching the entire Odoo 19.3 documentation for nesting returns nothing. Nesting stays in dedicated software and feeds quantities back.
Before You Buy Anything
Three checks, in order.
Measure your cost of quoting. Hours per quote times quotes per win, times your loaded estimator cost. If that number is small, estimating software will not repay itself and the honest answer is a better spreadsheet.
Reverse-engineer three real quotes. Take the last three you sent and write down how every number was produced. If there is no consistent method, software will encode the inconsistency at speed. Any vendor who has not asked to see three real quotes is guessing at your business.
Trace one won job from quote to invoice. Note every point where information was retyped. That count tells you whether your problem is estimating speed or the handoff, and those need different purchases.
Our Calgary software planning guide covers sequencing the wider decision, and ERP for Alberta manufacturing companies covers what happens downstream once the quote becomes a job.
The Honest Caveat
Estimating accuracy has the same ceiling as job costing. If actual hours are recorded weekly from memory, your estimate against actual comparison tells you what people remember, and the history you build on it inherits that error.
That is worth settling before you invest in estimating tools, because better inputs to a broken feedback loop still produce a broken feedback loop.
Frequently Asked Questions
- What does it actually cost us to quote a job?
- More than most shops think, because the cost is carried by the quotes you lose. At a loaded estimator cost of $67 an hour, four hours per quote, and a one in five win rate, every won job carries about $1,340 of estimating. If your average won job is $18,000 at a 12 percent margin, that is roughly 62 percent of the gross margin consumed before a single hour is worked.
- What is estimating software and do we need it?
- It is software that builds a quote from measurable inputs instead of from a blank page, usually holding material databases, labour standards, and reusable assemblies. You need it when quote volume is high enough that speed matters, or when more than one person quotes and the numbers need to be consistent. A single experienced estimator on low volume is frequently better served by a structured spreadsheet.
- How much does estimating software cost?
- It ranges widely, from a few hundred dollars a year for a takeoff tool to five figures annually for a dedicated fabrication estimating package with material feeds. Compare that to your cost of quoting instead of to your software budget. A package that cuts four hours per quote to two pays for itself quickly at any real quote volume.
- Should the quote live in the same system as the job?
- Ideally yes, and this is the point most shops miss. The estimate is the first structured record of the job, containing the material list, the hours by operation, and the assumptions. When it lives somewhere the job cannot reach, someone retypes a fraction of it and the rest is lost, which is why estimate against actual comparison so rarely happens.
- How do we estimate more accurately?
- Estimate from your own history instead of from first principles. What did the last three jobs like this actually cost, by cost line? That is a better predictor than any standard table, and it requires job cost history that is queryable by similar work. Shops that build that database quote faster and win more, because they can price tightly with confidence.