Hub-and-Spoke vs Point-to-Point Distribution Networks

The network map looked tidy: every plant shipped into one central hub, every distribution centre was fed from it, and the freight budget fell in the first quarter. By the third quarter the two biggest customers were complaining about a day of extra transit, damage claims had doubled on fragile SKUs, and one lane was sending a near-full trailer into the hub only to reload the same pallets onto another truck bound for the same city.

Hub-and-spoke and point-to-point are not rival philosophies. Each is the cheaper answer on a different set of lanes, and the crossover can be calculated from four numbers: direct trip cost, hub cost per unit, weekly volume per departure and trailer capacity. Get that arithmetic in front of the people who sign carrier contracts, and the network stops drifting toward whichever design the last consultant preferred.

• What this comparison covers: lane counts, touches per shipment, transit hours under US driving limits, the e-way bill steps at an Indian hub, a per-unit cost model and the contract clauses that change when freight passes through a sort.
• What you can reuse: a break-even formula, a worked US ($) and India (₹) lane, an incident trace from an Indian hub and a design review checklist.

Lane count, not geography, is what separates the two designs

The bottom line: point-to-point needs one lane for every origin and destination pair, while a single-hub design needs one leg per origin plus one per destination. Twelve sources serving forty delivery points means 480 possible direct lanes against 52 hub legs. Everything else in the comparison follows from that difference in how volume is pooled.

Pooling is the whole argument for a hub. A plant that sends six pallets a week to each of forty customers cannot fill a trailer to any of them, but it can fill one trailer a day to a hub. The hub sorts the mixed load by destination, and outbound trucks leave full because they carry freight from every origin. The price is extra handling, extra distance and a fixed departure schedule.

Point-to-point removes those costs but needs density on each lane. When a lane fills a trailer on its own, routing it through a hub only adds a detour, an unload and a reload. Most real networks are therefore hybrids: dense lanes run direct, thin lanes pool through one or more hubs, and the design question becomes which lanes sit on which side of the line.

Counting touches: where hub handling adds cost and damage exposure

Every touch has a labour cost and a small probability of damage, and a hub at least doubles the touches. Price the touch explicitly, rather than burying it in a blended hub rate, because it is the term that makes direct routing win on fragile, heavy or high-value freight even when trailer fill is mediocre.

A touch is any event in which a person or machine moves the unit off one conveyance or position and onto another. A direct pallet is loaded once and unloaded once. Through a cross-dock hub it is also unloaded from the inbound trailer, staged or sorted to a door, and loaded again. Parcel and small-carton networks add induction, a sorter pass and a bagging or palletising step, so a carton can see six or more touches.

• Labour per touch: measure it at the hub from forklift or sorter productivity, not from the contract rate; a pallet unload and reload often costs more than the sort itself.
• Damage per touch: if each touch carries probability p of damage, the chance a unit survives n touches is (1 − p) to the power n, so four touches roughly double the exposure of two when p is small.
• Scan per touch: each handling event is also a reconciliation point; a missed scan at the hub is where pieces go missing between two otherwise accurate manifests.
• Dwell per touch: freight that waits for a departure wave adds hours even when the handling itself takes minutes.

That reconciliation problem is covered in detail in our guide to reconciling bags and consignments at a PTL hub handover, which deals with what happens to individual pieces inside one hub. This article sits one level higher: whether the freight should pass through the hub at all, and on which lanes.

Transit time arithmetic: linehaul hours, sort windows and dwell

Hub transit equals inbound driving time plus wait for the sort plus sort time plus outbound driving time, and the wait is usually the largest controllable term. Direct transit equals driving time plus any rest the driver must take. On long US lanes the mandatory rest can make a hub relay as fast as a single-driver direct run.

US property-carrying drivers may drive at most 11 hours within a 14-hour window that starts after 10 consecutive hours off duty, and must take a 30-minute break after 8 hours of driving, under 49 CFR 395.3. A single driver on a 780-mile lane at an average of 50 miles per hour needs about 15.6 hours at the wheel, so the run must include a 10-hour rest. Door to door, that is roughly 26 hours.

Transit and cost model for one origin-destination pair (i, j)

Direct transit      t_direct = D_ij / v + r_ij
Hub transit         t_hub    = D_ih / v + w_h + s_h + D_hj / v

Direct cost / unit  C_direct = ceil(q_ij / K) * T_ij / q_ij
Hub cost / unit     C_hub    = c_in(i) + n_h * h + c_out(j)
Break-even volume   q*       = T_ij / C_hub     (one departure, q_ij <= K)

D   = road distance (miles or km)        v   = average linehaul speed
r_ij = mandatory rest added to the direct run (US: 10 h if driving > 11 h)
w_h = wait for the outbound wave (h)     s_h = sort or cross-dock time (h)
q_ij = units per departure on the lane   K   = trailer capacity in units
T_ij = cost of one direct truck          h   = handling cost per touch
n_h = touches added at the hub (usually 2)
c_in, c_out = inbound and outbound linehaul cost per unit on the hub legs
Rule: route direct when q_ij >= q* and the service gain is worth the cost gap.

A hub relay splits the same freight into two legs that each fit inside one duty window. If the inbound leg is 450 miles (about 9 hours), the freight waits 6 hours for its outbound wave, and the outbound leg is 390 miles (about 7.8 hours), the total is about 23 hours. Each driver also finishes near home, which matters for retention as much as for transit. A team on the direct lane would beat both, at a higher cost per mile.

India's driving-hour limits work differently and are not modelled here; the arithmetic still holds, with r set from the carrier's own relay or two-driver practice. What changes in India is the document trail at the hub, which the regulation section below covers.

A cost model per unit that shows where direct lanes start to win

Hub cost per unit is roughly constant for a lane because it rides on pooled volume, while direct cost per unit falls as lane volume rises. The lane crosses over at q*, the volume where one direct truck costs the same per unit as the hub route. Below q* the hub is cheaper; above it, direct is cheaper.

Three refinements keep the model honest. First, use the cost of the departure frequency the customer actually needs: a lane with 30 pallets a week but three required deliveries has only 10 pallets per departure. Second, charge the hub its full variable handling cost, including the extra touch at the outbound door. Third, include the detour distance on the inbound and outbound legs rather than assuming the hub sits on the straight line.

Lane-level routing in a transport management system should switch a lane from hub to direct once units per departure exceed q* = T/C_hub; on an illustrative Dallas to Atlanta lane, a $2,028 direct truck against a $93 hub cost per pallet breaks even at 22 pallets, about 85% of a 26-position trailer.

The fixed cost of the hub does not belong in this lane-level comparison. It is sunk once the hub exists, and it only matters when the question is whether to open, close or relocate a hub. That decision needs a different model, starting from where demand sits; our center-of-gravity method for warehouse location is a reasonable first pass before a full network study.

Worked example: twelve origins and forty destinations, US and India

The same formula gives opposite answers on two illustrative lanes: the US lane stays on the hub because its weekly volume is below break-even, while the Indian lane moves to direct because a smaller truck and lower hub-route efficiency pull break-even down. All rates below are illustrative planning figures, not quotes.

The US lane would flip to direct if volume grew by four pallets a week, or if a customer started paying for faster delivery. Note that the hub route is not slower here: relay drivers avoid the 10-hour rest. The Indian lane flips the other way because the hub detour is short but the wait for the outbound wave adds most of a working day.

Scaled to the full network of twelve origins and forty destinations, the planner runs the same test on all 480 pairs. In a typical consumer goods network a small share of lanes clears q* and carries most of the volume; those run direct, and the long tail of thin lanes pools through the hub. That hybrid is usually cheaper than either pure design.

Hours-of-service and e-way bill rules that shape hub placement

Neither market regulates network design directly, but each regulates what happens on the legs. In the US the binding constraint is driver hours, which favour hub spacing of one duty window or less. In India the binding constraint at the hub is paperwork: the e-way bill must follow the goods onto every vehicle.

For US planners, 49 CFR 395.3 caps driving at 11 hours inside a 14-hour on-duty window, and limits weekly on-duty time to 60 hours in 7 days or 70 hours in 8 days depending on whether the carrier operates every day. Hubs spaced so that each leg fits inside 11 hours of driving let carriers run relay networks where drivers return home most nights. Spacing them farther apart forces sleeper teams or overnight rests.

Relay hub spacing on US linehaul follows 49 CFR 395.3, which allows 11 driving hours inside a 14-hour window after 10 hours off duty; hubs placed so each leg needs under 11 hours let a 780-mile lane run as two relay legs in about 23 hours instead of 26.

In India, Rule 138 of the CGST Rules governs the e-way bill. Sub-rule (5) says that where goods are transferred from one conveyance to another, the consignor, recipient or transporter must update the vehicle details in Part B before the transfer and further movement. Sub-rule (6) lets a transporter generate a consolidated e-way bill in Form GST EWB-02 covering many consignments on one vehicle.

• First mile into the hub: the third proviso to Rule 138(3) does not require Part B for movement of up to 50 km within a State or Union territory from the consignor's place of business to the transporter's place of business.
• Linehaul out of the hub: every consignment on the outbound vehicle needs Part B showing that vehicle, either individually or through the consolidated EWB-02.
• Validity clock: Rule 138(10) gives ordinary cargo one day for the first 200 km and one more day per further 200 km or part, so hub dwell eats into the same validity as the direct run.

Indian hub transfers under CGST Rule 138(5) require Part B of every e-way bill to show the outbound vehicle before the trailer leaves the hub, and Rule 138(6) allows one consolidated EWB-02 per linehaul; a 560 km Ludhiana to Jaipur consignment keeps the same three-day validity whether routed direct or through NCR.

Contract clauses that change when freight passes through a hub

A hub network moves risk to the handover points, so the carrier contract must define cut-off times, dwell limits, transit measurement and claims evidence at the hub, not only at origin and destination. A direct-lane contract can stay silent on most of these because there is only one handover.

On US interstate moves, 49 U.S.C. 14706(a)(1) makes the receiving carrier, the delivering carrier and any carrier over whose route the loss occurred liable for actual loss or injury to the property. Section 14706(e) bars carriers from setting less than 9 months to file a claim or less than 2 years to sue after a written disallowance. Adding a hub does not change who the shipper can claim against, but it changes where evidence must be captured.

Freight claims on a US hub network run under 49 U.S.C. 14706, which makes the receiving carrier, the delivering carrier or the carrier causing the loss liable and bars claim windows shorter than 9 months; an extra hub touch changes where damage evidence is captured, not whom the shipper may claim against.

In India, the Carriage by Road Act, 2007 requires a common carrier to issue a goods receipt in triplicate on loading or acceptance (section 9), limits liability for loss or damage to a prescribed amount unless the consignor pays a higher risk rate (sections 10 and 11), and limits delay liability to the freight charge within the period agreed in the goods forwarding note (section 10(2)). Check the current text, since the Act has been amended since enactment.

Fuel surcharges are the other trap. A per-mile surcharge applied to each hub leg charges for the detour, while one applied to the origin-destination distance does not. Agree which distance the surcharge uses before the rate sheet is signed, and test it on the longest detour in the network.

A missed Part B update traced from a Delhi NCR hub to the highway

The most common hub failure in India is administrative rather than physical: the consignment reaches the right outbound vehicle, but its e-way bill still names the inbound vehicle. The trace below is illustrative, with invented document numbers, and shows how a late scan at the hub produced a roadside hold.

1. 21:10, hub unload: inbound trailer PB10-AB-4471 from the Ludhiana spoke docks at the NCR hub with 38 consignments. The hub system records the inbound manifest MF-LDH-0918; 37 consignments are scanned at unload, one carton set is missed under a stretch-wrapped pallet.
2. 22:35, consolidated e-way bill: the hub clerk generates EWB-02 for outbound vehicle HR55-CD-2096 to Jaipur using the 37 scanned consignments. The missed consignment is found during loading and placed on the same vehicle, but nobody adds its e-way bill to the consolidated bill.
3. 23:05, departure: the outbound gate releases the vehicle because the load count matches the dock plan, not the e-way bill list. The orphan e-way bill still shows PB10-AB-4471 in Part B.
4. 04:20, roadside check: an officer verifies the vehicle and finds one consignment whose e-way bill names a different vehicle. The goods are held for proceedings; the other 37 consignments continue after a delay.
5. Next morning, fix: the hub adds a gate rule: the vehicle cannot be released until every consignment loaded is on the consolidated bill or shows the outbound vehicle in Part B. Detention and penalty steps are explained in our guide to e-way bill penalties under section 129.

Network rules a planning tool can enforce and calls a planner keeps

Routing thresholds, cut-off checks and document gates are arithmetic and can be enforced by the transport system every time. Forecasts, hub siting and exceptions involve judgement, so software should prepare the evidence and a named person should decide. The table names the actual tasks rather than generic categories.

Ownership matters as much as logic. The lane threshold is only as good as the trailer capacity, hub cost and lane volume behind it, and each of those has a different owner. The table below sets out who maintains each input and who approves a change.

Checklist for choosing, piloting and reviewing the network design

Run the decision lane by lane, pilot the switch on a few lanes near break-even, and review quarterly because volumes drift. A network chosen once and left alone will slowly route dense lanes through the hub and thin lanes direct, which is the worst of both designs.

Mode choice interacts with this decision: a thin lane that does not justify a direct truck may still justify LTL or part-truckload rather than the private hub. Our comparison of FTL, LTL and PTL covers that crossover. For the wider hub of guides on planning, documents and costs, see the logistics and supply chain section.

If you are re-tendering a network and want lane-level break-even, hub gate checks and contract terms built into the transport system rather than kept in a spreadsheet, you can send us a request for proposal with your lane matrix and current hub plan.