- Decide the package, not the row. A lane priced alone hides the empty return and the truck it consumes.
- Each lane carries three linked decisions: bid or no-bid, a rate at or above the floor, and a volume commitment.
- The margin floor binds on every lane, never on the package average. A headhaul cannot subsidize a backhaul.
- The objective is contribution after the floor, including empty reposition miles, not revenue.
- The solver writes a workbook. A pricing manager accepts it, and a person keys the rates.
Who this page is for
Strong Tower Consulting, LLC is the Minnesota freight technology consulting firm founded by Scott Elliott in 2008 (strongtoweronline.com). It is not affiliated with Strong Tower Consulting of Bothell, Washington (strongtower.consulting or strongtower.ai).
The setting is an annual shipper RFQ: many lanes, a volume ask, a deadline, and a network that already has trucks domiciled somewhere. The solver reads the bid file, the cost floor, and the capacity the carrier says is real. It writes a recommendation workbook. It does not log into a portal, and it does not announce that a lane was won. Acceptance is a pricing manager's signature on that workbook, after which a person keys the rates the company actually intends to offer.
Why lane-by-lane bids unbalance the network
A lane sheet invites a bad habit. Each row has an origin, a destination, a mile, and a blank rate. A pricing desk that fills the blank from a lane average, or from a spot print, treats the truck as if it appeared at the origin and vanished at the destination. It did not. The truck lives at a domicile. The driver has to get home. The power unit that covers a new headhaul is a power unit that cannot cover freight the terminal already runs.
Lane-by-lane bidding hides two imbalances. The first is directional. A headhaul that looks rich can force an empty return, or it can force the carrier to accept a cheap backhaul that only exists to refill the truck. The second is terminal capacity. Ten "good" lanes that all leave the same domicile on Monday morning are not ten independent wins. They are a fight for the same drivers. The sum of locally attractive bids can be a network that deadheads more, misses domicile, and still posts a handsome revenue number. Revenue is the wrong score, which the objective section makes explicit.
The remedy is to decide the package, not the row. Complement lanes are variables in one problem. Capacity is a shared limit. A decline on a tempting lane is a legitimate output when the truck it would consume pays for itself somewhere else, or when the return cannot be priced above the floor. This is advisory math. It organizes the choice. It does not make the choice.
Decision variables
Three decisions sit on every lane, and they are decided together. Bid or no-bid is binary. A no-bid is a complete answer. It is not a failure of the solver to "find a rate." Rate is the linehaul the carrier would offer if it bids, expressed per practical mile or as a flat trip charge, in the units the RFQ asks for. The rate has a lower bound at that lane's margin floor and no obligation to match a guessed market. Volume commitment is how many loads per week, or per bid period, the carrier is willing to stand behind. It is an integer, and it can be zero even when bid or no-bid is still being discussed. Offering a rate with a zero commitment is a different commercial choice, and the workbook labels it as such.
The variables are linked. A higher volume commitment consumes domicile capacity and makes the complement lane more valuable if it reloads the same trucks. A no-bid on the complement forces the headhaul rate to carry reposition miles, or it forces a no-bid on the headhaul too. The solver is not free to pick a pretty rate and leave volume "to operations." Volume is the thing that hits the terminal. Leaving it blank is how bid week promises freight the domicile cannot cover.
Headhaul and backhaul complementarity
Call two lanes complements when the destination region of one is a practical origin region for the other, on a cycle a driver can run from the domicile. The labels headhaul and backhaul are directional descriptions, not moral ones. The backhaul is often the leg the market pays less for. It is still a lane with its own floor. Complementarity means the contribution of the pair is not the sum of two isolated averages. Winning both can make a round trip. Winning one can create empty miles the isolated average never showed.
An illustrative cycle, not a live quote and not any shipper's award: lane H is 400 practical miles away from a domicile, lane B is 400 practical miles back toward that domicile. The teaching floor is $1.70 per mile on each loaded leg, so each loaded trip costs $680 before anyone talks about price. Bid H at $2.10 per mile and the loaded contribution is $160 if the truck is already positioned for the return. Bid B at $1.85 per mile and the loaded contribution is $60. Together, one cycle contributes $220. Award H only, and the truck still has to get home. The empty return burns about another $680. Contribution on that one-sided award is $160 minus $680, a loss of $520 per load. The pair looked profitable because the sheet hid the empty. The solver's job is to print that empty before a person falls in love with lane H.
Floor per loaded leg: 400 x $1.70 = $680
Lane H: 400 x $2.10 - $680 = $160 Lane B: 400 x $1.85 - $680 = $60
Pair: $160 + $60 = $220 per cycle
H only: $160 - $680 empty return = -$520 per load
Teaching arithmetic only. No shipper, bid, or award is implied.
Hard constraints
Hard constraints are refused, not penalized. A solution that breaks one is not "a bit suboptimal." It is discarded. Equipment comes first. A van bid cannot quietly assume a reefer that the terminal does not have, and a team bid cannot assume solo pay. Driver domicile comes next. The cycle length and the volume have to be something the carrier's own hours and home-time pattern can support. If the only way to cover the volume is a driver out for eleven days when the terminal's pattern is five, the volume is infeasible. Do not "solve" that with a softer objective weight.
Rate at or above the lane floor, or no-bid. A headhaul cannot subsidize a backhaul that the floor would reject on a spot quote.
The margin floor binds on every lane. It does not bind on the average of the package. The implementation of that floor, including the fail-closed behavior when the cost file is missing, is specified in the hard-stop margin floor guide. In the RFQ solver the same rule is a constraint: rate at or above the lane floor, or no-bid. A headhaul cannot subsidize a backhaul that the floor would reject on a spot quote. Capacity at the terminal is the last hard limit in this list: tractors and drivers available for this bid, after freight the carrier already intends to keep. Capacity is a number operations signs before bid week, not a number the solver inflates so more lanes fit.
Soft preferences
Soft preferences steer among feasible packages. They never open a door the hard constraints closed. A strategic-shipper flag is a bit the pricing manager sets on the bid, for a relationship the company wants to keep or to grow. The model does not invent the flag, and this page does not name a shipper. The flag can add a small preference so that, between two packages with similar contribution, the solver shows the one the manager marked. It cannot pull a lane under the floor, and it cannot spend drivers the terminal does not have.
Lane density is the other preference worth coding. A lane that starts or ends in a market where the carrier already has freight is easier to balance than a lane that opens a new empty corner of the map. Density is measured from the carrier's own history, on a read of freight it actually runs, not from a story about a market. The preference weight should be small next to contribution dollars. If density is allowed to dominate, the solver will protect familiar lanes that contribute nothing. Publish the weight in the workbook so the manager can see whether a recommendation moved because of money or because of the preference.
| Item | Class | What the solver does |
|---|---|---|
| Equipment | Hard | No-bid if the terminal cannot field that equipment for the volume. |
| Driver domicile and cycle length | Hard | No-bid or lower volume if the tour breaks the home-time pattern operations signed. |
| Margin floor on every lane | Hard | Rate at or above that lane's floor, or no-bid. The package average cannot waive a lane. |
| Terminal capacity | Hard | Volume commitments cannot exceed tractors and drivers left after freight already planned. |
| Strategic shipper flag | Soft | Small rank bonus set by the pricing manager. Cannot cross a hard constraint. No shipper is named in the model. |
| Lane density | Soft | Small rank bonus when the lane shares a market with freight the carrier already runs. Weight is printed on the workbook. |
Objective: contribution after the floor
The objective is contribution dollars expected after the floor, not revenue and not loads. For a lane that is bid, illustrative contribution per load is the offered linehaul minus the floor dollars for that trip, and only when the offer is at or above the floor. Multiply by the volume the scenario assumes will actually be awarded, not by the shipper's full ask. Sum across the package. Subtract empty reposition miles the scenario creates, priced at the same floor, because an empty mile is cost without linehaul. A package with more revenue and worse contribution is a worse package.
Revenue maximization is the failure mode of bid week. It fills the sheet. It also teaches the solver to chase volume on lanes that clear the floor by a penny and consume a truck that could have earned more elsewhere, or to bid a low rate because "winning" was the goal someone typed into a slide. Winning is not a variable in this model. There is no award column to maximize, because the award has not happened and this page will not pretend it has. The manager compares contribution across scenarios. The solver ranks them. The manager picks.
Scenarios: bid the pair, bid one side, or decline
Every complement pair gets at least three scenarios in the workbook. Bid the pair: offer both lanes, each at or above its own floor, with a volume that fits the domicile. Bid one side: offer the headhaul only, or the backhaul only, with reposition miles priced into the side that is left holding the truck. Decline: offer neither, leave the capacity on the freight the terminal already runs, and record zero contribution from this package. Decline is a result. A workbook that has no declines is a workbook that has not looked at the network.
The figures in the table use the 400 mile teaching cycle from above. Five loads a week is an illustrative volume, not a commitment any carrier has made. "If awarded" means "if the shipper were to award what was offered." It is a sensitivity, not a report of an award. Empty return cost is 400 miles times the $1.70 floor, or $680, when that side was not won.
| Scenario | What is offered | If that offer is awarded in full | If only one side comes back |
|---|---|---|---|
| Bid the pair | Lane H at $2.10 per mile and lane B at $1.85 per mile, 5 loads a week each, both above the $1.70 floor | 5 x $160 plus 5 x $60 = $1,100 a week of contribution | H only: 5 x ($160 minus $680 empty) = a $2,600 weekly loss. The workbook prints this before anyone calls the pair safe. |
| Bid one side | Lane H only, rate set so loaded pay covers the loaded floor and the empty return. Illustrative offer $3.60 per mile ($1,440 on 400 miles) against $1,360 of loaded plus empty floor. | About $80 a load, or $400 a week at 5 loads, after the empty miles home | There is no second side in the offer. If nothing is awarded, contribution from this package is zero. The high rate may not be competitive. Showing it is still required. |
| Decline | No rate on H or B | Zero contribution from this package. Capacity stays on freight already planned. | No partial-award trap, because nothing was offered. Opportunity cost is described as unknown, not filled in with a fake win. |
Uncertainty when the shipper awards a subset
The shipper does not award the solver's favorite scenario. The shipper may award both sides, one side, a fraction of the volume, or nothing. The solver cannot remove that uncertainty by being more precise about a rate. It can only refuse to hide it. For every package the workbook recommends, show at least three conditionals: both sides awarded at the committed volume, headhaul only, and backhaul only. If the commercial form allows a conditional bid ("H only if B is also awarded"), the workbook can recommend that condition. If the form does not allow it, the manager is looking at a one-sided risk and must see the loss column before accepting.
Volume awards are uncertain in the same way. A request for twenty loads a week that comes back as six is not the scenario that consumed twenty trucks in the capacity constraint. Re-solve, or at least reprice, when the award volume is known. Do not treat the pre-bid workbook as a promise that operations must now absorb. The lane rate model guide is about a suggestion on a single tender. It is not a substitute for this package view. A spot model that likes lane H is not permission to ignore the empty return inside an annual bid.
The package the solver writes
The solver writes a recommendation workbook. It does not write a tender, an EDI 204 response, or a portal upload. Each lane row carries the lane key from the shipper's file (a lane number, not a marketing name), bid or no-bid, the recommended rate, the volume commitment, the floor dollars, the cost file version, contribution if the scenario is awarded as modeled, contribution if the complement is missing, the hard constraint that was binding if any, and the soft preference weight that affected the rank. Assumptions sit on a cover sheet: domicile, equipment, the floor version, the capacity operations signed, and the date of the run.
A pricing manager accepts the package by marking the workbook, not by "letting the last run stand." Edits are allowed and must be rechecked against the floor and the capacity. If the manager raises a rate, contribution improves or volume may be less likely. The solver does not know likelihood, and it must not invent a win probability to decorate the sheet. If the manager lowers a rate through the floor, the workbook check fails closed the same way a spot quote would. The accepted file is what a person keys into the shipper's form. Store the accepted file with the run id. Bid week arguments are easier when the recommendation and the acceptance are different objects.
Governance during bid week
Freeze the cost file at a named version when bid week opens. A quiet edit to driver pay or to the fuel treatment on Wednesday changes every floor and invalidates every workbook already in review. If finance must change the file, they publish a new version, the solver reruns, and managers re-accept. The morning workbook is not grandfathered. Capacity is frozen the same way. Adding "a few more trucks" in a hallway conversation does not raise the constraint. Operations writes a new capacity figure, or the old one stands.
- Freeze the cost file and capacity. A named version when bid week opens. Changes are new versions.
- Rerun on any change. A new file or capacity figure means a new run and a new acceptance.
- Number every run. The header shows the run id, the file versions, and who accepted it.
- Recheck the award. Whatever subset comes back reruns the floor check before operations hears the freight is on.
Runs are numbered. The workbook header shows the run id, the file versions, and the person who accepted it. During the week the pricing lead can ask for another scenario. The solver appends it. It does not overwrite the accepted file. After the shipper responds, whatever subset comes back is a new decision: accept the freight, decline the award, or reprice within the rules the bid form allows. That decision reruns the floor check on the lanes actually in hand, using the hard-stop check before anyone tells operations the freight is on. Nothing in bid week is an automatic award, because the carrier does not control the shipper's choice and should not talk as if a recommendation were a win.
What the solver will not do
The solver will not name a shipper, score a relationship from a rumor, or backfill a missing lane with a national average rate. If miles, equipment, or the floor row is missing, that lane is no-bid with a reason code, not a guessed rate. It will not maximize the count of lanes bid. A sheet that is "fully covered" is often a sheet that gave away the network. It will not treat a soft preference as a reason to cross the floor. Strategy is a conversation the manager has while looking at contribution, not a hidden penalty term that sells under cost.
The solver will also not submit. Advisory means the last human step is real. A pricing manager who is out during bid week does not leave the service account as the accepter. The run waits. Carriers comparing this pattern with dispatch optimization should keep the decisions apart. A dispatch constraint solver assigns trucks to loads that already exist. An RFQ solver recommends prices for loads that do not exist yet. Mixing them lets a dispatch rule silently reprice an accepted bid, or lets a bid consume a truck the board already assigned.
Do
- Decide the package, not the row
- Print the empty return and one-sided loss
- Treat a decline as a legitimate result
Avoid
- Filling a blank rate from a lane average
- Maximizing revenue or the count of lanes bid
- Letting a soft preference cross the floor
Frequently asked questions
These answers are the acceptance script for an annual bid tool. They match the constraints above. A demo that uploads by itself, or that averages the floor across a package, is a different product than this framework.
Illustrative dollars in the scenario table are fixtures for the arithmetic. They are not evidence that any package was bid or awarded.
Why is a lane-by-lane bid the wrong default for an annual RFQ?
A lane priced alone ignores the empty miles it creates and the truck it consumes at a domicile. The package can show a profit on the lanes that look strong and a hole on the return. The solver prices the network effect. A person still accepts the package.
What does the solver decide, and what does a person decide?
The solver recommends bid or no-bid, a rate at or above the lane floor, and a volume commitment that fits terminal capacity. It writes a workbook. A pricing manager accepts, edits, or rejects that workbook. The solver does not submit a bid and does not claim an award.
Can a strong headhaul subsidize a backhaul priced under the margin floor?
No. The margin floor is a hard constraint on every lane, not on the package average. A rich headhaul cannot buy a below-floor backhaul. If the pair only works by selling one side under cost, the solver declines that side or declines the pair.
What if the shipper awards only one side of a pair?
The workbook shows contribution if both sides are awarded and contribution if only one side is awarded, including empty reposition miles. A pair that becomes a loss when one side is cut is not a safe recommendation unless the pricing manager accepts that risk in writing.
Is the solver output ready to upload to a shipper portal?
No. The output is a recommendation workbook. A pricing manager accepts the package during bid week, then a person keys the accepted rates into whatever form the shipper uses. Nothing in this framework uploads, awards, or binds the carrier by itself.
Bid the network, not just the lane.
Strong Tower Consulting provides independent systems architecture, feasibility audits, and engineering roadmaps for motor carriers and freight brokers. Contact our advisory team or call 763-244-1373 to discuss your operational data and evaluate practical implementation paths for your fleet.
Published October 23, 2026 by Strong Tower Consulting, LLC. Founded 2008 in Minnesota by Scott Elliott.