The three-day estimate is not an accident. It is the result of a process that accumulated over decades, and most estimating teams have learned to plan around it rather than question it. A mid-size residential or commercial project arrives on Friday afternoon, the project manager needs a number by Wednesday morning, and the team pencils in Tuesday as a draft BOQ deadline. The math works. Most of the time.
When we started building FOBECON, we wanted to understand exactly where those three days go. Not in broad strokes, but specifically: which tasks consume the most time, which are genuinely necessary, and which persist simply because nobody changed them. The answer is more specific than most people expect.
The first hours go to drawing comprehension
Before any element gets counted, someone needs to understand what they are counting. For a typical mid-size project, the drawing set might include 40 to 120 sheets: floor plans at each level, elevations, sections, structural layout, MEP schematic layouts, construction details, and a specification document sitting separately in its own PDF.
A senior estimator spends the first three to five hours of a takeoff doing nothing that looks like counting. They are building a mental model of the project: which structural system is in use, how the building steps or cantilevers, whether there are atypical elements on particular floors that will require special treatment during enumeration.
This orientation work is necessary. Element counts that contradict the structural logic produce estimates that fall apart during tender review. A column count based on an architectural drawing that hasn't been cross-checked against the structural grid is the kind of error that costs a bid its credibility. The problem is that this phase cannot be delegated easily. A junior estimator can count doors, but they need the senior's interpretation of a partially-detailed connection before they count the structural steel. The senior is occupied with comprehension while the team waits for direction.
Sheet-by-sheet counting is the longest phase
Once the scope is understood, the counting begins. For a standard BOQ, counting covers doors and door frames (checked against the door schedule, then confirmed on each floor plan), windows and glazed elements, internal partitions by linear metre broken by material type, slab areas by level excluding major openings, columns and beams enumerated from structural drawings, finishes by room type, and fixtures counted from fitout drawings or schedules where provided.
Each category requires switching between drawing types. A door count from the schedule is the starting point, but floor plans may show doors that are not in the schedule, or vice versa. This reconciliation is not an edge case. On a typical six-floor office fit-out, it is common to find 8 to 15 discrepancies between the schedule and the plan drawings. Each discrepancy requires a judgment call and a note in the count table.
The counting itself is usually done in spreadsheet columns: a row per element type, cells for each level, aggregated totals at the foot. This format has not changed substantially since spreadsheets replaced paper count sheets in the 1990s. The container has changed; the method inside it has not.
Cross-referencing multiplies the time
The sheet-by-sheet count produces numbers, but the BOQ requires those numbers to be cross-referenced against at least two other sources: the structural drawings and the specification. The structural drawings tell you whether the partition you counted as a simple stud-and-board is actually required to be a fire-rated assembly. The specification tells you the correct unit rate category and construction standard.
This cross-referencing step is where many estimates introduce silent errors. An estimator working quickly assigns a standard partition rate to a section of wall that the structural engineer flagged as a shear wall with special requirements. The count is correct. The classification is not. The bid price is wrong by a factor that might be small on one wall but becomes significant when the same classification error repeats across 40 similar elements.
Thorough cross-referencing adds a half-day to a full day for a medium-complexity project. Teams that skip it introduce risk they do not quantify. The error does not surface at submission. It surfaces after award, during procurement or construction, when the actual material cost does not match the rate assumption in the BOQ.
Revision handling is where schedules collapse
Most of the time when we hear that a takeoff took five days instead of three, revisions are the explanation. A drawing revision issued after work has started forces a comparison between the original count and the revised drawings: what changed, which counts are affected, what new elements appeared or disappeared.
Revision tracking by hand is tedious. Estimators typically download both revision sets and compare them side by side on screen or against a printed set. Small changes, like a room partition moved 300mm, are easy to miss unless you are specifically looking for them. Larger changes, like a floor plate reconfiguration, require almost a complete recount for the affected floors.
On projects where two or three revisions arrive during the takeoff window, the estimating schedule can extend by a full working day. The revision itself may take 45 minutes to process, but the back-annotation of the spreadsheet, the confirmation with the design team about whether adjacent elements were affected, and the recheck of the BOQ classification for changed elements collectively consume considerably more time than the revision note implies.
Why this has not changed in twenty years
The estimating process today looks almost identical to what it looked like in the mid-2000s, with the exception that PDFs replaced printed sets. Tools like Bluebeam and PlanSwift added markup and measurement utilities, which improved accuracy on specific tasks, but did not change the fundamental structure of the work. The BOQ format itself is stable and industry-standardised. And the output of a takeoff is a legal document that feeds into a tender submission and eventually a contract. Changing a working process for a document with legal weight moves slowly.
There is also an institutional reluctance to trust a number that was not produced by a person who can be asked to explain it. When a junior estimator produces a count, the senior checks it. When a tool produces a count, it is less clear who owns the checking responsibility. That ambiguity has historically been enough to preserve the manual process.
We are not saying this caution is irrational. It is not. The question is whether the verification step has to be as expensive as the original counting step. In most other domains where data extraction shifted from human to automated, the human role shifted from producing the data to validating it. That shift has not yet been completed in construction estimating.
Where the time is actually being spent
The time is not uniformly distributed across the three days. Drawing comprehension and cross-referencing consume roughly 30 to 40 percent of the total and are genuinely judgment-dependent. The counting itself, for most standard element types, is a mechanical process that follows clear visual rules: this symbol on this layer at this scale is a door, count it.
For a typical mid-size office fit-out in Seoul, a project with roughly 100 drawing sheets and 6 floor levels, the count phase alone accounts for 14 to 20 hours of active estimating time. Most of that is straightforward pattern recognition: finding the door symbol, logging a count, moving to the next sheet. A human is doing it because there has not been a system that can do it reliably enough to trust.
If the mechanical counting can be handled by a system that reads the same drawing layers a human would, the senior estimator's time can shift toward the judgment-intensive parts: interpreting scope, reviewing classifications, handling revisions, and validating the BOQ before submission. That reallocation moves a three-day timeline toward four hours of active review time. The drawing set still has to be read. The BOQ still has to be checked. But the time a skilled person spends is focused on the decisions that require skill.