The easiest percentage to move

Every other progress figure on a project can be checked against something physical. Concrete either exists or it doesn't. Cable is either pulled or it isn't. Somebody can walk over and look.

Engineering has no wall. A document is forty percent complete because an engineer says so, and there is no independent record that contradicts it

WHAT CAN BE CHECKED, AND WHAT CANNOTConcrete, cable, steelsomebody can go and lookA document at 40%no wall to check
Figure 1 — The softest number on the project is also the earliest, which means the first months of reporting rest on it.

.

That makes it the softest number on the project, and it is also the earliest, which means it is the number the first several months of reporting rest on.

How weighting works

The standard approach is to weight each deliverable by expected effort and assign fixed percentages to defined stages: started, internally checked, issued for review, issued for construction.

The reason it works isn't that the stages are accurate. It is that they remove the judgement. A drawing at the review stage scores the same regardless of who is reporting it, which makes the aggregate comparable even when each individual estimate is rough.

This is the same reasoning as rules of credit for construction, which Cost & Cash Week 11 covers. Fixed steps beat continuous judgement, not because the steps are right but because they are the same for everybody.

Where it breaks

Two failures, and both are visible if you look for them.

The first is weighting by count rather than by effort. A hundred small documents and ten large ones give a percentage that moves quickly at the start, when the easy items are being cleared, and then stalls. The curve looks healthy for four months and flat for six.

WEIGHTING BY COUNT OR BY EFFORTBy countraces, then stallsBy expected hoursmoves with the work
Figure 2 — A hundred small documents and ten large ones produce a curve that is encouraging early and inexplicable later.

The second is revisions. A drawing issued for construction is at a hundred percent. It comes back with comments, or the design changes, and it has to be reworked. If the register has no way to move a deliverable backwards, the same problem from Week 11 appears here: effort is being consumed against something that already shows complete.

What can corroborate it

Since there is no physical check, the corroboration has to be procedural, and two records are usually available.

Transmittals show what was actually issued

TWO CORROBORATING RECORDSTransmittalsan event with a dateHours bookedeffort against claimNeither is conclusive. A large divergence is worth a question.
Figure 3 — Stages assigned before anything leaves the department become visible the moment the register is compared with what was actually transmitted.

, and issuing is an event with a date rather than an opinion. A register claiming a large number of deliverables at issue stage should be matched by transmittals; if it isn't, the stage is being assigned before the document leaves.

Hours booked to engineering give the other side. Effort expended against progress claimed is the engineering equivalent of the paint check in Week 8 — not conclusive, but a large divergence in either direction is worth a question.

Changing the order rather than the length

If the register is new, weight by expected hours, define four or five stages, and decide in advance what happens to a deliverable that is reissued.

Inherited: the weighting is probably by count, and changing it will restate the whole series, which nobody will thank you for mid-project. The workable move is to report the released count alongside the percentage. Two numbers, one of which can't be softened, and the second gradually becomes the one people quote.

System design

Weighted stages, and two records that can contradict them. Neither corroboration is conclusive; both are free.

RecordProduced byRequired qualityVerified againstFeeds
Deliverable weightingEngineeringBy expected hours, not by document countHistoric effortEngineering progress
Stage reachedDiscipline leadDefined stages applied the same way by everybodyTransmittal recordProgress, earned value
Hours booked to engineeringTimesheetsSplit by disciplineProgress claimedProductivity, forecast
Released countDocument controlA count, which can't be softenedTransmittalsThe figure people end up quoting

The last row is the antidote to the first three. Report it alongside the percentage and it gradually becomes the figure people quote.

Practical insight

Take the engineering register and count two things for last month: how many deliverables moved a stage, and how many transmittals went out.

If the first is much larger than the second, progress is being recorded at stages that don't involve anything leaving the department. That isn't dishonesty. It is what happens when the stages are defined internally and never checked against an external event.

Key takeaways

  • Engineering progress has no physical record to check it against. It is the softest and earliest number on the project.
  • Weighted stages work because they remove judgement, not because the weights are accurate.
  • Weighting by document count instead of effort produces a curve that races then stalls.
  • A reissued drawing consumes effort against something already showing complete, unless the register can move backwards.
  • Transmittals corroborate the issue stages. Stages assigned before anything leaves are visible this way.
  • Hours booked against progress claimed is the engineering version of the same corroboration.
  • On an inherited register, report released count alongside percentage rather than restating the series.

Records born here. Weighted deliverable register · transmittal log · released count series.

What is coming next

Drawings release work. So do purchase orders, and the second chain is longer and less visible than the first.

Next week: procurement — order status, expediting, and the promise that keeps moving a week at a time.

Enjoyed this lesson?

Join with Google to get each new lesson the moment it's published — and help me see which topics matter most to you. No spam, one email a week, unsubscribe anytime.