Where a Custom Builder's Week Actually Goes: The Coordination Tax

The short version: A custom builder running 2-4 jobs spends roughly 28 hours of a 50-hour week on office coordination: finding specs, checking quotes, writing logs, answering status questions, and keeping selections tied to the budget. Call it the coordination tax. SpecAlign's model, built row by row and labeled as modeled, puts about 13-14 of those hours within reach of software that reads documents instead of just storing them.

A GC sits down at 8:50 on a Sunday night with a cold coffee and a 47-page cabinet quote open on the laptop. His phone is propped against the salt shaker, showing the revised kitchen sheet A-4.2. He's trying to work out whether the island got six inches longer, because if it did, the quote's 14 base cabinets are wrong by one. Monday's daily log isn't written. Tuesday's isn't either. The drywall sub's invoice has been sitting unopened since Thursday.

Nothing on that list is building. All of it is the job.

What is the coordination tax?

The coordination tax is the time you spend keeping documents, money, and people in sync instead of building. You pay it every week, mostly after the crew leaves. It doesn't show up on a P&L, which is why it stays invisible: nobody invoices you for it, and nobody bills it to the job.

Custom work makes it worse. Census data puts owner-built and custom homes at an average of 15.2 months from start to completion, the longest of any single-family category. More months means more revisions, more selections, more subs rotating through. Each one adds a row to the tax.

Where do the 28 hours go?

SpecAlign's time model starts from a builder with 2-4 active builds and a 50-hour week, then inventories the office tasks one at a time. These are modeled hours from stated assumptions, not a survey and not customer data. Here's the table, with what a Read, Compare, Act pass does to each row.

Weekly office task	Hours	What the tax looks like	Modeled hours back
Finding and verifying spec info	5.5	Scrolling PDFs to answer "what's the spec for that room"	3.3
Quotes, POs, tracking	4.0	Retyping quote lines, checking each against memory	2.0
Client and stakeholder status	4.0	"Where are we" calls, schedule and price impact questions	1.6
Selections	3.0	Chasing decisions, syncing choices to specs and budget	1.35
Field and sub coordination	3.0	Re-sending sets, firefighting missed updates	1.5
Accounting touchpoints	3.0	Invoice vs PO, draw prep	0.75
Progress recording	2.0	Daily logs, photo documentation	1.5
Change and revision admin	2.0	Documenting changes, cascading to budget and schedule	1.0
Takeoff and estimating	1.5	Measuring quantities, re-measuring on revisions	0.75
Total	28.0		13.75

Two rows are partial. Client status hours come back only in part, because a client who wants to hear your voice will still call. And SpecAlign isn't accounting software, so the accounting row is the smallest recovery on the table: it helps you match an invoice to a PO, and it doesn't replace your bookkeeper or QuickBooks.

Why is document lookup the biggest row?

Because every other row depends on it. A quote can't be checked until you know the spec. A sub can't be told what changed until you know which sheet moved. A selection can't be tied to the budget until the room's finish schedule is current.

Dodge Construction Network's 2024 "Not by Design" report found 17 daily onsite interactions per project, 44% of them involving conflict. Most of those conflicts start as a version question: which set, which spec, which number. A change that never reached the field is the expensive version of that question. The cheap version just eats your Sunday.

SpecAlign reads the plan set into specs by room and trade, each linked back to the page it came from, so the lookup becomes a search instead of a scroll. That's the Read step. The row's modeled recovery is 60%, not 100%, because someone still has to judge the odd spec that doesn't fit a category.

Which rows come back fastest?

The ones where the work is pure re-keying. Take progress recording. A super who talks for 90 seconds into a phone gets a structured entry with crew, work performed, weather, and deliveries, which is what a voice daily log does. The model assumes 75% of the 2 hours disappears. Of the nine rows, it's the cleanest swap, because the information already exists. It's only waiting to be typed.

Quotes are the next easiest. Extracting line items from a vendor PDF and comparing them to your own history is mechanical work, and the Compare step is where a quote 17% over your last three jobs gets caught before you issue the PO. The model recovers 50% of the 4 hours.

Selections and client status move slower. They involve people, and people have their own schedules.

What does the tax cost in dollars?

At $75-100 an hour for owner time, 13.75 modeled hours a week works out to roughly $1,000-1,375 a week, or $50-66K over a 48-week year. That's an opportunity cost, not cash out of pocket, and it assumes you'd spend the recovered day on something that pays: a site walk, a bid, a client lunch that turns into the next contract.

You might also spend it at home. That counts.

Where the model runs thin

Most of the baselines are assumptions, not measurements. Only the document-lookup bucket has outside research behind it. The rest are stated estimates for a builder in the 2-4 job range, and your week will differ. A 6-job builder with an office manager has a different table than a 2-job builder who does his own estimating.

So run the exercise yourself. Take a week, one line per office task, and total it. If your number lands near 28, the model is close. If it lands near 15, you've got an office that's already doing the work, and the better question is what it costs you in salary. If it lands near 40, you're working a second job after dark.

Either way, you'll know which row to attack first. How SpecAlign works across the platform lays out every module and the AI pass attached to each, where one ships.

Pick the row that cost you last month's worst mistake. Start there.

---

Sources

• SpecAlign time-reclaimed model (2026): nine weekly office tasks for a 2-4 job custom builder on a 50-hour week; 28.0 baseline hours, 13.75 modeled hours back; modeled estimates with stated assumptions, not measured customer results
• U.S. Census Bureau, Survey of Construction (2023 data): average single-family completion 10.1 months; owner and custom-built 15.2 months
• Dodge Construction Network, "Not by Design: The True Cost of Poor Collaboration" (2024): 17 daily onsite interactions per project, 44% involving conflict
• SpecAlign product capabilities: plan-set spec extraction by room and trade with page-level provenance; voice-to-structured daily logs (Whisper transcription, structured entry); quote-line extraction with price-anomaly flags against org history

The Sunday-night scene illustrates a common pattern, not a specific job or person. Hours are modeled, and the dollar range assumes owner time at $75-100 an hour.