Chapter 13 — Key Takeaways

Detailed Estimating: CSI Divisions, Subcontractor Quotes, Equipment Costs, and the Final Number

One page. Self-contained. Come back to it the morning of a bid.

Key Takeaways

• A low number is a question, not an answer. When a bid arrives well under the field, the correct first sentence is "what did they miss?" — not "did they miss something?" The first assumes the gap and sends you to find it. On the Ridgeline library bid, a $340,000 advantage turned out to be a $383,000 hole in six minutes of scope checking.
• Divisions organize the specification; bid packages organize the market. They almost never line up, and scope disappears in the gap. Architectural precast is Division 03 but sits in the enclosure package. Medical gas is Division 22 but may sit with mechanical. Only the scope sheet reconciles them.
• An empty division gets a row and a dash. A dash means "I looked and it is zero." A missing row means nothing at all, and in six weeks nobody can tell a decided zero from a forgotten scope.
• General conditions are time-dependent, not quantity-dependent. Roughly 78% of Northgate's $2,900,000 accrues per month regardless of production. That is why every schedule slip is automatically a cost overrun — at $5,150/CD before anyone argues about cause, and $10,650/CD once liquidated damages are added.
• The scope sheet is written before the bids arrive. Write it after and you will unconsciously level against the bids you received instead of against the work the drawings require.
• Leveling routinely inverts the ranking. On the Northgate drywall package the apparent low bidder finished $196,822 above the apparent high bidder. On the masonry package the apparent high bidder was the true low by $105,520. If you have never seen a bid tab invert, you have not been leveling.
• A shared exclusion is invisible on the bid tab and lethal on the job. When all three bidders exclude the same item, adding it back changes no ranking — and the work still has to be bought by somebody. Keep a separate list.
• You are not buying a price; you are buying a company's next nine months. Capacity, backlog, bonding, experience modification rate, named personnel, and past performance are all part of the number, and a default in month eleven costs more than any spread on bid day.
• Utilization, not purchase price, decides own versus rent. Depreciation is charged per hour of use; interest, taxes, insurance, and storage accrue on the calendar. Drop fleet utilization from 1,000 to 600 hours a year and a $14,054 ownership advantage collapses to $1,346.
• Every markup sits on its own base. General conditions on time. Bond and insurance on contract value including the fee. Contingency on cost. Fee on the subtotal. Escalation on the unbought portion only. The right rate on the wrong base survives every review, because the rate is defensible and nobody re-derives the base.
• Markup and margin are different amounts of money. Northgate's 4.0% fee markup is a 3.80% margin on the GMP. Asking for "4%" without saying which one is a $75,200 misunderstanding.
• Finding savings has a name and a mechanism; buying the job does not. Curtis Boone shaved $465,000 in a forty-minute meeting and paid $998,000 back — 2.15 to 1 — turning a $784,000 fee into a $214,000 loss that requires $6.1 million of new work to recover.
• Structure the estimate so it can become the control budget. Estimate at the level of detail somebody will actually collect accurate data against. Not finer. Not coarser.

Action Items

This week, on whatever job you are on:

1. Pull your general-conditions estimate and divide it by the contract calendar days. Say the number out loud. That is what a day of delay costs you before anyone assigns blame. Add liquidated damages and say that number too.
2. Sort your general-conditions lines into one-time and time-dependent columns. Total the second column. That is your real exposure to a schedule slip.
3. Write one scope sheet for the next trade you will buy — before you look at any quote. List every inclusion, every exclusion, every by-others item. Two hours.
4. Take the last bid tab you accepted and re-level it. Add back every exclusion at your own carried value. Did the ranking hold?
5. Find every shared exclusion across your current bid tabs — items every bidder left out — and put them on a single list with a named owner and a dollar value.
6. Check one markup base. Pick your bond premium and confirm it was computed on contract value including fee, not on cost of work.
7. Ask what your last five subcontract packages' coverage was. Any package with one bidder? Write down which of the three honest responses you used.
8. Read your own basis of estimate. If there isn't one, that is this week's real assignment.

Common Mistakes (and the Fix)

Decision Framework

Before any number leaves your desk — the four-question test

1. Is every division on the summary, including the ones with a dash?
2. Is every package covered by at least two leveled bidders, or by a flagged and loaded plug? State the plug total out loud, as a percentage of the cost of work.
3. Does every markup sit on the correct base? Time · contract value including fee · cost · subtotal · unbought scope.
4. Can you defend every line to somebody who wants it to be lower? A line that cannot survive the question "why is this here" will not survive the final review.

Leveling a bid tab — the sequence

   1. Write the scope sheet          (before bids — non-negotiable)
   2. Log every quote                (time, company, number, exclusions verbatim)
   3. Check addenda acknowledged     (an unacknowledged addendum is a different building)
   4. Add back each exclusion        (at YOUR carried value, consistently)
   5. Subtract duplications          (scope carried by two bidders or by GC and sub)
   6. Adjust alternates & unit prices (to the scope actually required)
   7. Normalize bonds and insurance  (same treatment for everybody)
   8. Rank — and expect an inversion
   9. Circle the SHARED exclusions   (invisible on the tab; still must be bought)
  10. Evaluate the leveled low bidder as a company, not a number

The ethical line — one sentence

On the legitimate side, the scope or the terms change and the price follows. On the other side, only the price moves.

Everything else follows from that. Asking a bidder what is in their number: legitimate. Asking a bidder to price added scope, a value-engineering idea, a different schedule, or different payment terms: legitimate. Telling a bidder their number is above your budget: legitimate. Telling a bidder what a competitor bid, or inviting a re-bid of unchanged scope: bid shopping, whoever starts the conversation.

Own versus rent — the one question that decides it

Not "what does the machine cost." "How many hours a year will this machine actually work?" Depreciation is per hour of use; interest, taxes, insurance, and storage are per year. Low utilization moves the annual carrying cost onto fewer hours and kills the ownership advantage. This makes fleet decisions company decisions, not project decisions.

The Northgate reconciliation — for reference