Estimated vs actual job costing: a practical guide for job shops

The problem

Most job shops know their overall margin from the P&L, weeks after month-end. What they don't know is which jobs produced it. A shop running 25% gross margin can have a third of its jobs below 10% and a handful losing money outright. The averages hide them, and the estimates that produced them get reused on the next quote.

What to measure

Job cost variance = actual cost − estimated cost (positive means an overrun). Estimated margin = revenue − estimated cost. Actual margin = revenue − actual cost. Margin erosion = estimated margin − actual margin, which always equals the cost variance when revenue is fixed. Track it in dollars first: a 10-point miss on a $2,000 job matters less than a 3-point miss on a $90,000 job.

How MarginGuard helps

MarginGuard calculates these for every job in your ERP export, ranks jobs by dollars eroded, breaks erosion into the five categories, and rolls it up by customer and part family. A calculation that takes hours in a spreadsheet happens on upload. You can also try the free variance calculator for a single job.

Next step: sign up, load the demo, or compare plans.

How to get started

1. Step 1
   Export estimated and actual costs per job

   From your ERP, export one row per job with revenue, estimated cost, and actual cost, ideally split by category. Use the guide for your ERP.

2. Step 2
   Calculate variance and erosion

   For each job: actual − estimated by category, then estimated margin − actual margin. Sort by erosion in dollars.

3. Step 3
   Find the pattern, then act

   Group the top losers by customer, part family, work center, and category. Decide per pattern: fix the estimate, reprice the customer, or fix the process.

4. Step 4
   Repeat weekly

   Review the new top 10 every week while jobs are still fresh in people's memory, and check whether last week's fixes held.

The five variance categories

Labor (run time longer than standard, slower operators, rework), Setup (more setups than planned, long first-article), Machine/burden (wrong machine rate, routed to a costlier machine, extra hours), Material (price increases, yield and scrap, substitutions), and Outside processing (vendor price changes, expedite fees, extra operations). Labor and setup are usually estimating or process problems; material and outside processing are usually pricing or purchasing problems.

Estimate problem, price problem, or process problem?

If the same part family overruns every time, the estimate standard is wrong, so fix the standard. If one customer's jobs erode across part families (rush orders, engineering changes, tight tolerances), it's a pricing problem, so reprice or add terms. If a work center or shift overruns regardless of part, it's a process problem, so fix the process. Variance data tells you which conversation to have.

Why dollars beat percentages

Percent variance overweights small jobs. Ranking by dollars puts the jobs that actually move the P&L at the top of the list, and makes the weekly review a 30-minute meeting instead of a spreadsheet exercise.

Common data pitfalls

Open jobs without actuals look perfect, so filter to closed jobs or jobs with actuals posted. Operation-level exports double-count when summed. Mixed revenue bases (quoted vs billed) make margins incomparable. Burden rates that changed mid-year make old estimates look worse than they were, so note rate changes.

Downloads

• Job cost template (CSV)

  Estimated and actual columns by category.