Repeat-job variance: turn actual costs into the next quote assumption

A job that cost more than quoted does not automatically prove the estimate was wrong. Delivered quantity may have changed, an operation may have been added, or time capture may be incomplete. A useful variance review starts by establishing that estimate and actual records describe the same work.

Match the basis

Record customer part, revision, good-unit quantity, route, material, outside processes and inspection requirements. Identify changes approved after the original quote. Distinguish the original estimate from a revised production budget so scope growth remains visible. Use consistent freight, tooling and overhead treatment on both sides.

Build a table with estimated and actual costs for programming, setup, conversion, material, outside work and inspection. Add a confidence note for each actual. If material was issued to several jobs together or a timer was left running, investigate before using that row to change a pricing rule.

Work an example

A fictional ten-piece job carries $200 setup, $300 run cost and $250 material and purchased work. Actuals are $260, $280 and $290: $80 unfavorable overall. Setup accounts for $60 unfavorable, run cost for $20 favorable, and material/purchased work for $40 unfavorable. The total identifies a margin problem; the buckets identify questions.

Suppose a one-time fixture adjustment caused extra setup and a supplier minimum caused the purchased-work increase. The next quote may need the supplier minimum, while the fixture event requires a conditional note rather than a permanent setup increase. These illustrative numbers are not a benchmark or a claimed customer result.

Separate cause from classification

Use explanations such as missing allowance, changed scope, unusual interruption, rate change or unreliable actual record. Do not force one cause when evidence supports several. Ask estimator and production lead to review the same record. A favorable outcome can reflect process improvement or omitted time and deserves scrutiny too.

ProShop publishes job-costing functionality and Katana documents manufacturing-cost calculation. These establish relevant evaluation areas, not that your shop’s underlying data are complete. Test cost attribution and export with representative work before treating a dashboard as a feedback loop.

Store one actionable lesson

Close each review with an owner, evidence and a precise next-quote change. For instance: apply a supplier’s current minimum below a stated lot size and recheck it on the next RFQ. Preserve the prior value and reason. State when the lesson applies and when it does not.

Review a manageable set of repeat jobs before proposing automation. If missing scope or poor time capture recurs, address that cause first. A variance chart becomes useful when it changes a documented assumption and the following job can test whether the change helped. A correlation between high variance and one machine alone does not establish its cause.