Actual Cost Is Higher Than Estimate. Which Assumption Was Wrong?
Estimate-versus-actual review is valuable only when the variance can be traced to material, time, waste, rate, routing, scope or execution. The goal is to improve the next estimate, not simply explain the last job.
The short answer
When actual job cost is higher than estimate, the useful question is not “who went over budget?” It is “which estimating assumption, production condition or customer requirement changed?” Material, setup, run time, waste, routing, subcontracting, scope and rates should be separated so the business knows whether to update the estimate, improve execution or treat the event as a one-off exception.
Estimate-versus-actual review should improve the next job.
How this connects to the pillar
The broader closed-loop model is covered in The Quote Won the Job. Will the Job Still Make the Margin You Expected?. Actual job cost is the feedback that makes estimating smarter over time.
Build a cost bridge
Start with the estimated job cost.
Then separate actual movement into categories such as:
- material quantity;
- material price;
- setup time;
- run time;
- labour;
- waste;
- finishing;
- subcontract;
- tooling;
- packing;
- freight;
- scope change.
A bridge makes the discussion concrete.
One total variance can hide several unrelated causes.
Separate quantity and rate
Material can be expensive because too much was used or because the purchase price increased.
Labour can be expensive because more hours were needed or because the rate changed.
Subcontract cost can rise because scope changed or supplier price changed.
Separate these effects so the corrective action is appropriate.
Verify the estimate baseline
Before judging production, confirm the estimate being compared is the final approved commercial baseline.
If the customer changed quantity or specification and the estimate was not updated, production may appear unfavorable against an obsolete target.
Preserve quote revisions and approved changes.
Compare route
Did the job run through the process assumed by the estimator?
If not, why?
Possible causes include:
- machine unavailable;
- finishing moved outside;
- extra inspection;
- alternate material;
- rework;
- customer change.
Routing variance can explain cost even when each operation performed efficiently.
Review material variance
Use actual issue and scrap evidence.
Ask whether excess material came from:
- poor imposition;
- setup;
- press spoilage;
- finishing waste;
- material defect;
- rerun;
- overproduction.
This determines whether the estimating allowance or production process needs attention.
Review time variance
Separate setup from running.
A job may have accurate run speed but too many setup events.
Or setup may be normal while production speed is below assumption.
This distinction is essential for short-run work.
Review unfavorable and favorable variance
Favorable variance also teaches the business.
Perhaps a new sequence reduced setup, a material yielded better, or a machine ran faster than the old standard.
If the improvement repeats, update the estimating assumption.
Do not learn only from failures.
Avoid blame-driven review
If variance review is treated as a performance accusation, teams may spend time defending numbers rather than explaining causes.
The review should focus on whether assumptions and processes are still valid.
Accountability remains important, but the main output is a better future decision.
Use tolerance
Not every small variance needs investigation.
Set practical thresholds by value, percentage, customer, job family or recurrence.
A small repeated variance may deserve attention even when one job is immaterial.
A large one-off event may need explanation but no standard change.
Look for recurrence
Group variance by:
- product family;
- substrate;
- machine;
- process route;
- customer;
- estimator;
- quantity band.
Patterns are more powerful than single-job anecdotes.
If one family consistently uses more board than estimated, the yield model may be wrong.
Connect commercial decisions
A job can have accurate cost and still have low margin because sales deliberately priced aggressively.
That is not estimating error.
Separate cost accuracy from pricing strategy.
Management should see when low margin was intentional.
Close the loop into standards
A variance review should end with one of four outcomes:
1. no change — one-off event;
2. estimating standard update;
3. production improvement action;
4. commercial/process change.
Assign an owner where action is needed.
Without this step, the review becomes historical reporting.
Time the review while memory remains
Review important jobs soon enough that people remember what happened.
Waiting until quarter end can make the numbers accurate but the operational story weak.
A weekly or monthly exception review is often more useful than a large annual costing exercise.
What management should measure
Useful indicators include:
- estimate accuracy by cost category;
- recurring material variance;
- recurring setup variance;
- recurring run-speed variance;
- finishing and subcontract variance;
- jobs outside tolerance;
- time from completion to variance review;
- standards updated from repeated evidence;
- variance caused by customer change versus internal execution.
The purpose is learning velocity.
ACC and Jupiter context
Jupiter links estimating and actual job execution for printing and packaging, allowing cost assumptions and production evidence to be compared within the same job context. That supports a practical feedback loop between commercial estimating and factory performance. Explore Jupiter for printing and packaging.
Operational review point
Choose the ten jobs with the largest meaningful estimate-to-actual variance.
For each, assign one primary cause and one owner.
Then ask whether the same cause appeared in another job.
If it repeats, change the standard or process.
If it does not, preserve the event as an exception rather than distorting the model.
Compare variance against commercial margin
A cost variance is more important when it materially changes the job's contribution.
Prioritize review using both cost movement and margin effect.
A small percentage variance on a very large job can matter more than a dramatic percentage on a tiny sample run.
This keeps review proportional to business consequence.
Keep estimator feedback specific
Telling an estimator that a job “lost money” is not actionable.
Feedback should state which assumption failed: material yield, setup count, run speed, finishing, subcontract, packing or another driver.
Specific feedback makes the next estimate better and avoids defensive conversations.
Separate estimator accuracy from forecasting uncertainty
Some variance occurs because the estimator made a weak assumption.
Some occurs because the job genuinely contained uncertainty that could not be known at quote time.
Distinguishing the two keeps performance review fair.
Where uncertainty is structurally high, the commercial response may be contingency, shorter quote validity or clearer change terms rather than expecting impossible precision.
Review variance by lifecycle stage
Classify when the cause entered the job:
- quotation;
- order change;
- planning;
- purchasing;
- production;
- finishing;
- delivery.
This helps identify whether estimating accuracy is truly the problem.
A variance created after customer revision should not automatically trigger a new estimating standard.
Measure whether corrective action worked
If the team updates a waste factor or setup standard, review later jobs to see whether estimate accuracy improves.
Without follow-up, standards can change repeatedly without proving that the model is getting better.
The feedback loop should test its own effectiveness.
Use a common variance vocabulary
If estimating calls a difference “waste” while production calls it “startup” and finance records it as “material variance,” trends become difficult to compare.
Create a small shared set of cause categories.
Detailed notes can remain process-specific, but the management layer should use consistent language so recurrence can be seen across jobs.
Preserve evidence behind large variance
For material exceptions, retain enough operational context to explain the result later.
That can include change order, scrap reason, alternate route, supplier issue or production hold.
The purpose is not paperwork for its own sake.
It is to prevent the same job from being reinterpreted differently by estimating, operations and finance.
Feed lessons back before the next quote
A monthly review is too slow if the same product is quoted again tomorrow.
For repeatable high-impact variance, update or flag the relevant estimating assumption promptly.
The closed loop is valuable only when learning arrives before the next commercial decision.
Review the jobs that matched estimate too
Accurate jobs confirm which assumptions are working.
A costing model should not change only because exceptions are noisy.
Compare successful jobs alongside poor ones.
This helps distinguish a genuinely weak standard from a few unusual events and protects useful assumptions from unnecessary adjustment.
That balanced sample improves confidence that any standard change is based on a real pattern rather than an isolated job outcome.
Frequently asked questions
What level of estimate accuracy is good?
There is no universal number. Use tolerances that reflect your product mix, job complexity and commercial risk.
Should every unfavorable variance change the estimate?
No. Update standards only when evidence suggests the assumption is systematically wrong.
What if actual data is incomplete?
Fix transaction discipline first. Weak actuals cannot create strong estimating feedback.
Should finance or operations own the review?
Both should contribute. Finance understands cost structure; operations understands what physically happened.