Predict, Measure, Change, Repeat
Various other metrics could be used to measure
the estimator’s performance:
• Accuracy in arithmetic.
• The amount of time required to prepare a
• Completeness of estimates.
• Comparison of the estimator’s work order and
the actual work order used to ship the order.
Differences to be on the lookout for include
the number of manufacturing steps, the tooling and fixtures needed, the time required for
setup, the time required for processing each
part, and the amount of material used.
One easy measurement is to compare the number of bids presented to the number of orders
granted. It might reveal an estimator who is overly
cautious about setup or pessimistic about speeds
and feeds and, as a result, generates high cost estimates with low success rates. At the other extreme,
it might be that an estimator who enjoys 100 percent success in getting orders is a 100 percent failure in generating profit for the shop.
Comparing a project’s completed cost to the
price charged is a very useful and compelling metric. The expectation is that cost will be less than sale
price. When it isn’t, corrective action is required;
when it is, maximizing action is required.
For designing a review system for estimators, our
general advice is to focus on what the estimator
has control over—time predictions, material, and
machinery. Not all projects will be profitable. The
estimator has little control over scrap or random
acts of nature. In some cases, the corrective action
is not much more than recover and rebuild.
Training as a Consequence of Review
Here’s a prediction: The completion of a review of
the expended time and material for each manufacturing step will refine the estimator’s skill. As the
estimator studies the completed work order and
compares it to the estimated work plan, it is natural to spot errors and omissions as well as successes
As an example, consider a project that was
launched into production using tooling that was
on hand at the time. Given the track record and
forecast for this project, the shop might be wise
to invest in tooling to eliminate machine cycles.
The customer might get its metal parts sooner, the
shop’s tooling library will grow, and the throughput
of the shop will improve.
This sort of detailed review of completed work is
most informative when the estimator does it as a
self-checking process. Having other eyes duplicate
the review effort as a buddy check can be useful.
For example, the shop foreman’s view of labor and
material utilization has an emphasis on employee
behavior, whereas the estimator’s point of view is
on completeness of prediction.
As a result of this review process, several individuals can share in the blame, whether it is related to
a bad estimate or a bad overall shop performance.
What we want to emphasize are the situations in
which both sides performed extremely well. How
did that happen? What can we do to repeat that
You can nail the quotes on some of the jobs all of
the time, but you can’t nail all of the quotes all of
the time. What we want to emerge from the shared
review effort of the estimator and foreman is action
taken on trends. If the trend is toward a profitable
operation with a growing list of successful customers, then take action to do more of that. We’ll leave
it to you to imagine problems and corrective action
that might be appropriate in your specific trade.
Fortune cookie says, “Measure the results of
striving for the goal and you will succeed.” I don’t
think that necessarily means that your goal will
make you successful. Measuring and striving—
presumably with corrective aim if needed—are what
lead to success.
Pop Quiz Redux
So, how did you do on the pop quiz? We’ll offer
our version of an answer, but with a disclaimer. The
method of estimating that we are demonstrating is
based on time and motion standards for semiau-tomated machinery. How we set those standards
and selected that machinery is specific to our shop.
We’re happy to have you look over our shoulder
and kibitz. Send us e-mail, please.
When we worked on the previous pop quiz for
a small quantity with fast delivery, we made an effort to predict setup time separately from run time.
This is based on the idea that setting up a machine
is a one-time event that can be a predictable activity. Consider a press brake for bending sheet metal.
The machine requires the installation of a lower V
die, an upper punch, the setting of the stroke, and
the positioning of the backgauges. Once that is
done, the machine will cycle at full speed to complete a bend approximately every 15 seconds. The
setup time remains about the same—say, 30 minutes—regardless of how many bends are made. The
machine’s cycle speed remains about the same regardless of the setup.
This matters to the estimator. If the batch size is
small, the setup time is a dominant part of the production cost. If the batch size is larger, the run time
for each part becomes the significant production
So part of the answer in the pop quiz for a larger
batch size is to amortize the setup over the larger
quantity. To put that a different way, divide setup
time by batch size to get the cost per part for setup.
Add to that the run time per part, and you’ll find
that it costs less to build parts in larger batches. But
that’s only if you’re amortizing the setup over the
Another important consideration is material utilization. For the small batch, sawing parts from a
stick of raw material was cost-effective. For larger
batch sizes, the design in Figure 1a—the prototype
quickie—is very expensive and wastes material. Figure 1c—the thin die-stamped part—will benefit
from large-scale production. If your answer to the
pop quiz included a different material plan—and
even a different manufacturing plan—then you get
a gold star on your estimate.
One of the heartbreaks in the estimating system
we’ve described is with setup time. It is really hard
to define. Is moving material from one location to
another part of setup? How about sweeping and
cleaning up chips and scrap? Waiting while the inspector measures the first article? How about stopping production to inspect a sample? Or adding
grease or oil to the machine while it is running?
After all of those questions are answered, the
next trick is to get the production personnel to report the setup time separately from the run time.
Time clocks are not very good at that sort of data
collection. It doesn’t take long for the data collection burden to be irksome to production. It is not
practical to slow down production just so the review of the estimator is easier to do with a computer.
Instead, establish the setup time for a machine as
a one-time activity. That activity might need to be
repeated several times during a year, but it does not
have to be recorded for every job completed. These
time-standard setting events are a good way for
production personnel to train the estimator.
Gerald Davis uses CAD software to design and develop products for his clients at www.glddesigns.
com. Please send your questions and comments
over 30 years