The Lowest Bid Isn't the Right Bid: Where the 15–20% Variance Hides

Every bid has an exclusions section, and the low bid's exclusions section is usually why it's low. A sitework sub comes in $22,000 under the next bid on a $180,000 package. Page four, in the terms: "Excludes traffic control, dewatering, and export of unsuitable soils." The other two bids carried all three. Traffic control alone on that job runs $14,000; dewatering is weather-dependent and open-ended. The $22,000 advantage was never real — it was scope moved from the bid to your future change-order log.

Exclusions do their damage because they live in prose while the comparison lives in numbers. A leveling sheet that only captures the totals treats "excludes dewatering" and "includes dewatering" as the same bid. The fix is mechanical: pull every exclusion out of every bid, put each one on its own comparison row, and price what it would cost you to self-perform or buy separately. An exclusion with a real dollar value attached stops being fine print and starts being part of the bid.

Two bids price the same conduit. One quotes $412 per thousand feet; the other quotes $4.65 per hundred feet. Scan a forty-line bid at speed and both register as "about four dollars" — but the second bid is roughly 13% higher on that line, and nothing on the page announces it. Per foot versus per stick, per unit versus per case, per pound versus per hundredweight, each versus per box of fifty: every unit switch is a chance for two prices to look comparable while the math says otherwise.

Wire and pipe are the classic offenders because they're legitimately sold in multiple units, so neither bid is wrong — they're just not comparable until someone converts both to the same basis. Fasteners and fittings do the same thing at smaller dollars: quoted each on one bid, per box of fifty on another, per hundred on a third. On a bid with 200 lines, that conversion is exactly the kind of tedious, error-prone work that gets skipped at 4 p.m. on bid day, which is why unit variance survives more leveling sheets than any other kind. The fix costs nothing: state the unit you want quoted in the request itself, and convert anything that comes back in a different unit before a single price gets compared.

Your takeoff says 6,800 feet of pipe. Bid A prices 6,800 feet. Bid B prices 6,200 feet — at a slightly better unit price, which makes its total look great. If Bid B wins and the takeoff was right, the missing 600 feet ships later at current market price, plus a second delivery charge, and the award-day advantage evaporates. Worse is the reverse case: a bid that quietly rounds quantities up, so you award more material than the job needs and the overage becomes yard inventory you paid job money for.

Quantity variance is the easiest kind to check and the least checked, because it requires holding three documents at once — the takeoff, and both bids — and comparing line by line. The discipline is simple: every quantity in every bid gets checked against your takeoff, not against the other bids. Two bids that agree with each other and disagree with your takeoff are both wrong.

A bid is a snapshot, and some bids say so in writing. "Pricing firm for 15 days." "Wire pricing subject to market at time of shipment." "Steel subject to mill escalation." Compare a bid holding its price for 90 days against a bid holding for 15, and the totals are not the same number even when they're the same number — one of them is a commitment and the other is an opening position. A $140,000 wire package quoted with 15-day protection on a job that won't release wire for four months isn't a $140,000 bid. It's a bid for whatever copper costs in month four, with $140,000 written on it.

The leveling move is to treat price protection as a line item in its own right: window length, what triggers escalation, whether escalation is capped, and whether it runs both directions. A supplier who'll hold pricing through your actual release schedule is offering something worth paying a margin for — but you can only price that advantage if the comparison surfaces it.

Two more places the spread hides, both easy to miss because they rarely get their own row on a leveling sheet.

• Freight and fuel surcharges. One bid is FOB jobsite — delivered, done. The other is FOB supplier's yard, with freight "to be determined" and a fuel surcharge "per policy at time of shipment." On a heavy package, delivery can run 3–5% of material value. A bid that's 2% lower but excludes freight is not the lower bid; it just defers the difference to the invoice.
• Substitution language. "Or equal." "Or approved equivalent." "Manufacturer subject to availability." That language means the quoted price may not be for the quoted product. A gear bid priced around a substitute breaker line can be genuinely cheaper — or it can be a submittal fight, a rejected substitution, and a repriced package after you've already awarded. Either way, a bid that commits to the specified product and a bid that reserves the right to swap are different bids, even at the same price.

Both of these follow the same pattern as everything above: the variance lives in words, not numbers, so a totals-only comparison can't see it.

Add it up — a skipped accessory line here, an exclusion there, a unit switch, a short quantity, a soft price window, unpriced freight, an open substitution — and a 15–20% spread between bids that looked level at the bottom line stops being surprising. It's the normal condition of unleveled bids. And the award is the moment it locks in: every gap you miss becomes an RFI, a change order, or an invoice surprise, negotiated after your alternative bids have expired and your bargaining position is gone.

The fix is not more suspicion. It's leveling at line-item scope: every line in every bid mapped against the same scope list, every exclusion priced, every unit normalized, every quantity checked against the takeoff, before award. Done by hand in a spreadsheet, that's hours per package — real hours, on bid-week deadlines, which is why it usually gets done for the big lines and skipped for the rest. This is the problem CheckIT RFQ is built for: drag every bid in, and it reads every line, normalizes units and groupings, levels each bid against your scope, and flags what's missing, excluded, or priced off-market — with every variance tied to the exact line in the source bid. It's running now as an invite-only pilot; accepted pilots get 90 days free. However you do the leveling — software or a disciplined spreadsheet — the principle holds: the lowest number wins the spreadsheet, but the bid with the fewest gaps wins the job.