Construction labor productivity grew about 1% a year for two decades, while manufacturing grew 3.6%.

McKinsey Global Institute's Reinventing Construction study, published in 2017, compared construction with the world economy and with manufacturing over twenty years. Construction labor productivity grew about 1 percent a year. The world economy grew 2.8 percent. Manufacturing grew 3.6 percent.

The report looks at the whole industry. I look at one small corner of it. I'm 18, I work in electrical preconstruction, and a real part of my week is counting lighting fixtures by hand. Open the set, find the fixture schedule on E-001, then go sheet by sheet through the lighting plans with a highlighter, marking every Type F so it does not get counted twice, then every F1, then every FA. When revision 3 lands, I do it again, because a single total gives no way to tell which part of it is still good.

Very little of that is judgment, and all of it has to be right. A missed fixture does not stay missed. It waits for buyout. That is the kind of task a gap like 1 percent against 3.6 percent is made of: repetitive, necessary, and done much the way it has been done for years.

So I built LumenTakeoff for that one job. Lighting fixtures first, on purpose.

What LumenTakeoff does about it

You drop the PDF, the full multi-discipline set if that is what you have. A construction PDF is not a picture. On a vector sheet the fixture symbols are paths sitting at exact coordinates, and LumenTakeoff reads them directly. A 64-page set processes in about a minute, page by page with progress. Scanned or raster pages are reported as unsupported, never guessed at. Everything runs on your machine and nothing is uploaded.

Every repeated symbol on every counted sheet becomes one family, shown with a ringed crop of one real instance and its exact instance count. Then you decide. In the naming queue one unnamed family comes up at a time, most frequent first. Type the code and the set's own fixture schedule completes it and fills in the description. Enter confirms, Esc marks it not a fixture. Confirming a name applies it to every member of the family and writes it to the firm's library, so the next job from the same engineer arrives partly named.

Each fixture card keeps four figures apart and never blends them: detected instances, production count, in bid scope, and how many times the schedule tag appears on the sheet, shown as what the document states with its discrepancy. Where the drawing does not settle scope, the family is flagged rather than assumed.

The export waits until the numbers reconcile. Lock in results runs a reconciliation check first, and export is refused rather than exporting a wrong number. The XLSX carries Coverage, Quantities, Findings and Evidence sheets, with proof images of every counted instance ringed.

The machine counts. You decide.