🔁 Backhaul / Deadhead Reducer

Purpose

Turn a week's load list, committed outbound lanes, and available equipment into a ranked set of backhaul-pairing and triangulation opportunities — each with estimated empty miles saved, revenue added, driver-hour feasibility, and ready-to-use sourcing actions (post to load board, call target broker, pull from committed shipper pool).

When to Use

Use this skill during the Friday / Monday planning huddle, after the routing-guide rerate, whenever a new lane is added with an unfavorable headhaul ratio, when fuel prices spike and deadhead dollars become painful, or any time the operations team notices a persistent empty-mile pattern in a region (e.g., trucks consistently running 200+ empty miles out of a Kansas City drop). It is equally useful for a small carrier with 15 trucks and for a broker trying to place a partner's return leg profitably.

Required Input

Provide the following:

Outbound load list — For each committed load this week: pickup city/state, delivery city/state, equipment type, delivery day, estimated unload complete time, and driver/truck assignment if set
Equipment and HOS snapshot — Available driver hours at the drop city for each truck (11/14/70 remaining), home-time or reset windows, equipment restrictions (reefer, flatbed, hazmat endorsement), and any domicile anchor each truck must return to by a given date
Backhaul sources — Preferred load boards (DAT, Truckstop, 123Loadboard), broker partners for the destination region, committed shipper backhaul pool (shippers with a return-lane agreement), and private shipper contacts
Triangulation constraints — Maximum acceptable deadhead miles per leg, minimum target RPM for a backhaul, and any lane blacklist (regions the fleet avoids)
Market context (optional) — Recent DAT RateView or Greenscreens benchmarks on the candidate backhaul lanes, load-to-truck ratios in the drop region

Instructions

You are a dispatch planner focused on reducing empty miles without breaking HOS or committed delivery windows. Your job is to take the outbound picture and produce a prioritized backhaul plan the dispatcher can execute the same day.

Before you start:

Load config.yml and read the backhaul.* defaults profile (detailed in Configuration Reference below). When present it supplies the values that are stable week to week — deadhead tolerance and minimum all-in RPM (by equipment class where the fleet sets them differently), all-in cost per mile by equipment class, the domicile and home-time policy, the preferred boards and broker partners by region, the committed-pool shippers with their standing return-lane rates, the lane blacklist, the benchmark source, and the fleet type. A week's-input value always overrides the matching config default; record which values came from config in the internal-notes block
Reference knowledge-base/terminology/ for correct terms (deadhead, backhaul, triangulation, headhaul, RPM, HOS 11/14/70, domicile, tender)
Read backhaul.fleet_type (regional / OTR / dedicated) — the triangulation logic differs by type and is applied in step 4. If config does not set it, infer from the load list and state the inference

Process:

Group drops by region and day — Cluster the outbound list into drop regions (3-digit ZIP or commercial market) and by delivery day so pairs of trucks finishing in the same area can share backhaul opportunities
Identify the deadhead exposure — For each drop, estimate empty miles to the next likely origin (domicile from backhaul.domicile, next committed pickup, or nearest known freight source). Rank drops by deadhead dollars at risk (empty miles × the all-in cost per mile for that truck's equipment class from backhaul.cost_per_mile, falling back to the fleet default where a class isn't itemized)
Score backhaul candidates — For each high-exposure drop, propose 2–3 candidate backhaul lanes using the boards and broker partners in backhaul.sources for the drop region plus the committed-pool shippers in backhaul.committed_pool (apply their standing return-lane rate where the drop matches a pool lane). Score each candidate on:
- Deadhead-in miles (drop → pickup), measured against backhaul.deadhead_tolerance
- Pickup window feasibility given HOS and unload finish time
- Estimated all-in RPM vs. the backhaul.min_rpm floor for that equipment class
- Destination usefulness (does it set up the next committed move or land at domicile?)
- Source reliability (committed shipper > preferred broker > spot board)
- Lane is not on backhaul.lane_blacklist
Build triangulation plans where they pay — When a single backhaul does not get the truck home or to the next committed move, propose a two-stop triangulation (A → B → C → home) only if total empty miles drop, both legs meet the minimum RPM, and HOS holds. Apply the triangulation logic for the fleet's backhaul.fleet_type:
- Regional — home-time and domicile return govern. Prefer triangulations that land the truck at backhaul.domicile inside its home-time window even at a slightly lower blended RPM; do not chain a second triangulation that pushes the truck past its reset/home-time anchor
- OTR — domicile pull is weak; maximize blended RPM and forward progress. Chaining a third leg or a relay is in-bounds as long as each leg clears backhaul.min_rpm and HOS holds, even if it keeps the truck out longer
- Dedicated — the dedicated-lane commitment is protected first. Only fill a backhaul that returns the truck in time for its next dedicated tender; never propose a triangulation that risks the dedicated start, regardless of RPM
Flag feasibility killers — Call out anything that would void the plan: driver out of hours, equipment mismatch, reset required before the candidate pickup, weather or port congestion affecting the lane
Recommend and assign actions — For the top 5–8 opportunities, produce a one-line action the dispatcher can execute today: post to DAT with specific rate floor, call Broker X about lane Y, pull committed-pool return from Shipper Z, or hold for a better candidate tomorrow
Summarize the weekly impact — Roll up estimated empty miles saved, backhaul revenue added, and net margin improvement across the full plan. Break out which trucks gain the most

Output requirements:

A short executive summary: total empty-mile exposure this week, proposed savings, number of actions
A ranked opportunity table: truck, drop, candidate backhaul, deadhead-in, RPM, estimated margin, action owner
Triangulation plans shown as labeled legs (Leg 1 / Leg 2 / return) with HOS and rate math
Feasibility flags listed separately so they are not lost in the table
The one-line action assignment per opportunity is in imperative voice ("Post DAT, floor $2.15/mi, 0500 MT pickup Wed") so the dispatcher can work from it directly
An internal-notes block documenting assumptions (fuel cost used, benchmark source, HOS snapshot timestamp)
Saved to outputs/ if the user confirms

Reference Example

Input (week's load list + equipment/HOS + sources):

Regional dry-van carrier, 15 trucks, Southeast domicile Charlotte NC (CLT). Friday planning huddle for next week. Four trucks finish committed outbound by Wed with no return assigned:

Truck 12 (driver L. Suarez) — delivers Halverson Manufacturing load CLT → Atlanta GA 30303, unload complete Wed ~1400 ET. HOS at drop: 6:30 drive / 8:00 window / 41 hrs of 70 remaining. Must be home CLT by Fri night (weekend home-time).

Truck 7 (driver M. Reyes) — delivers CLT → Savannah GA 31401, unload complete Wed ~1100 ET. HOS: 8:00 drive / 10:30 window / 58/70. No hard domicile return until following Mon.

Truck 19 (driver D. Tovar) — delivers CLT → Nashville TN 37210, unload complete Wed ~1600 CT. HOS: 4:00 drive / 5:00 window / 22/70 — tight. Home CLT by Fri.

Truck 3 (driver J. Park) — delivers CLT → Greenville SC 29601, unload complete Wed ~1000 ET. HOS: 9:30 drive / 11:00 window / 63/70. Flexible.

Sources: DAT + Truckstop; broker partner "Eastbound Logistics" covers GA/TN backhauls; committed-pool shipper "Piedmont Paper" tenders ATL→CLT returns at $2.05/mi. Config: deadhead tolerance 150 mi/leg, min backhaul RPM $1.90 all-in, cost/mi $1.42, lane blacklist = none. Market: DAT ATL→Southeast load-to-truck 4.1, Nashville outbound thin (1.8).

Backhaul Plan — Week of 2026-06-22 (4 unassigned returns)

Domicile: Charlotte NC (CLT)   Cost/mi: $1.42   Min backhaul RPM: $1.90 all-in   DH tol: 150 mi/leg
═══════════════════════════════════════════════════════════════════════════════════════
Truck Driver    Drop          Naked DH home   $ at risk   Recommended backhaul          RPM
─────────────────────────────────────────────────────────────────────────────────────────
 12   Suarez    ATL 30303     244 mi (→CLT)   $346 empty  Piedmont Paper ATL→CLT (pool) $2.05
  7   Reyes     SAV 31401     249 mi (→CLT)   $354 empty  Triangulate SAV→Charleston→CLT $1.98
 19   Tovar     BNA 37210     408 mi (→CLT)   $579 empty  HOLD — out of hours (see flag) —
  3   Park      GSP 29601      98 mi (→CLT)   $139 empty  Run empty (under DH tol, low $) —
═══════════════════════════════════════════════════════════════════════════════════════

Executive summary. Four open returns, $1,418 total naked-deadhead exposure. Plan converts two to revenue, accepts one short empty as cheapest option, and holds one truck for a reset. Net: ~$1,490 backhaul revenue added, ~493 empty miles eliminated, 3 actions to work today.

Ranked opportunities

Truck 12 (Suarez), ATL → CLT — committed-pool return, $2.05/mi. Pull the Piedmont Paper ATL→CLT return from the committed pool. 244 loaded mi, ~3:50 drive — fits Suarez's 6:30 drive / 8:00 window with room. Lands him home CLT Wed night, well inside Fri home-time. All-in $2.05/mi = ~$500 revenue vs. $346 of empty. Source reliability: committed pool (highest). Action owner: dispatch.
Truck 7 (Reyes), SAV → CLT via Charleston — triangulation. No single SAV→CLT backhaul clears RPM on the board, but a two-stop pays: Leg 1 SAV→Charleston SC 29403 (108 mi loaded, DAT board load $2.20/mi, pickup Wed 1600 feasible), Leg 2 Charleston→CLT (207 mi, Eastbound Logistics partner load $1.85/mi or committed if available). Blended ~$1.98/mi all-in across 315 loaded mi ≈ $990 revenue, deadhead between legs ~12 mi. HOS holds (58/70, no reset needed; Mon domicile gives slack). Action owner: dispatch + broker desk.
Truck 3 (Park), GSP → CLT — run empty. Only 98 empty mi, under the 150-mi tolerance and $139 exposure. The thin GSP outbound board (no qualifying backhaul ≥ $1.90 within tolerance) means sourcing cost exceeds the empty cost. Run empty; re-check the board Tue PM for a same-day add.

Triangulation detail — Truck 7

Leg 1:  Savannah GA 31401 → Charleston SC 29403   108 mi   $2.20/mi   pickup Wed 1600
Leg 2:  Charleston SC 29403 → Charlotte NC 28202   207 mi   $1.85/mi   pickup Thu 0700
DH between legs: ~12 mi (Charleston drop → Charleston pickup)
Blended all-in: (108×2.20 + 207×1.85) / 315 = $1.98/mi   Revenue ≈ $621 + $383 = $1,004
HOS: Wed 4.0 drive used / 6.5 remaining; 10-hr reset overnight Wed; Thu fresh clock. ✓

Feasibility flags (do not lose in the table)

Truck 19 (Tovar), BNA — OUT OF HOURS. 4:00 drive / 5:00 window remaining at a 1600 CT finish, and only 22/70 on the cycle. BNA→CLT is 408 mi (~6:30) — cannot be run loaded or empty today, and a backhaul would require a 34-hr restart before pickup. Recommendation: 10-hr reset BNA Wed night, then either (a) a Thu BNA→Southeast board load if one clears $1.90 (Nashville outbound is thin at 1.8 L:T — unlikely), or (b) deadhead home Thu after restart and accept the $579. Hold the backhaul decision to Tue PM when Thu BNA boards populate; do not commit a return that the restart can't support.
Truck 7 Leg-2 source is a partner-board load, not committed — confirm Eastbound Logistics tender before releasing Leg 1, or the truck strands in Charleston. Fallback: Charleston→CLT committed pool if Eastbound can't cover.

Action assignments (work today)

Truck 12: Pull Piedmont Paper ATL→CLT from committed pool, confirm Wed 1500 ET pickup, rate $2.05/mi. Owner: dispatch.
Truck 7: Book DAT load SAV→CHS $2.20/mi (Wed 1600 pickup) AND secure Eastbound Logistics CHS→CLT $1.85/mi (Thu 0700) as paired triangulation; do not release Leg 1 until Leg 2 is tendered. Owner: dispatch + broker desk.
Truck 19: Reset BNA Wed night; re-scan Thu BNA outbound boards Tue PM before deciding load vs. empty. Owner: dispatch (Tue PM follow-up).

Weekly impact

Truck   Empty mi eliminated   Backhaul revenue   Net margin add (@ $1.42 cost/mi)
 12          244                  ~$500              ~$500 rev − $346 empty avoided = +$846 swing
  7          ~237 (vs naked)      ~$1,004            +$1,004 rev on 315 paid mi
  3            0 (empty kept)       $0               cheapest option, $139 accepted
 19          held                 TBD Tue            decision deferred
─────────────────────────────────────────────────────────────────────
Plan total: ~493 empty mi eliminated · ~$1,504 revenue added · 3 actions today, 1 deferred

Internal notes

Config-sourced values: $1.42/mi all-in cost (dry-van class) from backhaul.cost_per_mile; $1.90 all-in min RPM from backhaul.min_rpm; 150 mi/leg from backhaul.deadhead_tolerance; CLT domicile + weekend home-time from backhaul.domicile; DAT/Truckstop boards, "Eastbound Logistics" GA/TN broker partner, and the "Piedmont Paper" ATL→CLT $2.05/mi committed lane all from backhaul.sources / backhaul.committed_pool; benchmark source backhaul.benchmark_source = DAT RateView; backhaul.lane_blacklist empty. Fleet type backhaul.fleet_type = regional — which is why the plan prioritizes domicile return inside home-time (Suarez and Tovar both anchored to Fri CLT) over squeezing a higher blended RPM out of a longer chain.
HOS snapshot timestamp: Fri planning huddle, reflects projected Wed-drop hours — re-validate Tue PM against actual ELD before committing Truck 19.
Benchmark source: DAT RateView L:T ratios (ATL 4.1 favorable, BNA 1.8 thin) drove the hold-vs-source call on Truck 19.
Truck 3's empty run is a deliberate accepted cost — sourcing a sub-150-mi backhaul on a thin board costs more in dwell than the $139 saved.

Configuration Reference

config.yml — backhaul.fleet_type (regional / OTR / dedicated — selects the step-4 triangulation logic), backhaul.deadhead_tolerance (max empty miles/leg, scored in step 3), backhaul.min_rpm (all-in RPM floor, by equipment class where set), backhaul.cost_per_mile (all-in cost by equipment class, drives the step-2 dollars-at-risk ranking), backhaul.domicile (home base + home-time policy), backhaul.sources (preferred boards + broker partners by drop region), backhaul.committed_pool (committed-pool shippers with standing return-lane rates, applied in step 3 where the drop matches a pool lane), backhaul.lane_blacklist (regions/lanes excluded in scoring), backhaul.benchmark_source (DAT RateView / Greenscreens default for the market-context read). A week's-input value always overrides the matching config default; config-sourced values are listed in the internal-notes block.
Knowledge base — knowledge-base/terminology/ (deadhead, backhaul, triangulation, headhaul, RPM, HOS 11/14/70, domicile, tender)
Sibling skills — skills/operations/route-optimization-brief.md (HOS-feasibility math shared with the triangulation legs here), skills/operations/carrier-rate-comparison.md (rate benchmarking for the candidate backhaul lanes), skills/sales/spot-vs-contract-rate-negotiation-brief.md (when a recurring backhaul lane is worth converting to a committed-pool agreement)

Synthetic example — the 15-truck carrier, the four trucks/drivers (L. Suarez / M. Reyes / D. Tovar / J. Park), Halverson Manufacturing, "Piedmont Paper," and "Eastbound Logistics" are illustrative and continuous with the synthetic operator world used across the repo's examples. DAT / Truckstop are real load boards and HOS 11/14/70 rules are real; the quoted lane rates, L:T ratios, and HOS snapshots are synthetic planning values, not from a real tender.