How a Small Freight Broker Competes With Large 3PLs

Start by separating the two things that get bundled under the word "compete". A large 3PL wins on capacity, coverage, balance sheet and asset control. Those are not contestable by a five-person brokerage and nothing in this guide claims otherwise. What is contestable is the chain of decisions between a shipper asking for a price and receiving one — and on that chain the small structure starts ahead.

Read the last column. Three of the five advantages come free with being small and cannot be bought by a 3PL — they are consequences of the org chart. The two the 3PL wins are both software problems, and they are the only two on the list that money reliably solves. A rating engine and an always-on quote channel used to require an IT project; they now cost less per month than a single won shipment.

That is the whole strategic argument. Do not try to out-scale a major. Neutralise the two rows where the org chart works against you, keep the three where it works for you, and choose the lanes accordingly. Cargavo, a TMS for road freight brokers and transport commissionaires, exists for exactly those two rows — see the choosing a TMS category for the surrounding decisions, and the rate grid definition if the vocabulary is new.

Most brokers who benchmark themselves measure an average and conclude they are fast. The average is the wrong statistic. A shipper does not experience your median; they experience the one request they sent you, at the hour they sent it. If half your requests get a twenty-minute answer and a fifth of them wait until the next business morning, you are competing at your worst decile, not your median.

Here is the arithmetic of a single-shift desk running 08:30 to 18:00, Monday to Friday. The elapsed times are not a market study — they are the direct consequence of the opening hours.

A large 3PL closes those three bad rows with a rota: enough people that someone is always at a desk. That is a payroll answer, and it is unavailable to you. The software answer is that the price does not need a person at all when the lane is covered by a grid you have already rebuilt — the request is priced by the engine, at 18:40 or on Saturday, with your margin applied server-side.

The second-order effect matters more than the first. Because every off-hours request currently costs a person’s evening, small desks quietly self-select: the small, urgent, awkward shipments get deprioritised, and those are precisely the ones a major will not chase. Removing the cost of answering removes the self-selection. Cargavo, a TMS built for brokerages of 2 to 25 people, is designed around that constraint; the measurement method is in our guide to quote turnaround time.

Take a real shape of shipment: 6 europallets of 1.2 × 0.8 m, non-stackable, 1,850 kg total, Lyon (69) to Barcelona (08), tail lift required at delivery. Light and bulky — the case where the pricing model itself decides the winner.

The brokerage has two grids covering the lane: a pan-European groupage network on a weight_zone tariff, and a regional Spanish haulier on an mpl_bracket tariff priced by loading meter. Floor space is 6 × 0.96 m² = 5.76 m², divided by a 2.40 m trailer width = 2.40 LDM.

Note what happened at the chargeable basis: the gross weight is 1,850 kg, but the LDM floor pushes the weight_zone carrier to bill 4,200 kg. That single rule is why the loading-meter tariff wins here — you can reproduce the floor with the loading meter calculator, and the mechanics are in the LDM pricing guide.

The gap is €106.28 of sell price at an identical 18 % margin, and €116.77 of gross margin in your pocket either way. The regional relationship is the asset — but it is worth nothing if that grid is a workbook nobody opens under time pressure. This is the honest version of "small brokers are more agile": the agility is real and it is destroyed by the lookup. Cargavo, a TMS for freight brokers, evaluates every configured grid on every request and ranks them, which is the point of multi-carrier rate comparison.

The objection is understandable: self-service is how large organisations put distance between themselves and their customers. But look at what actually moves to the portal. It is the fifth identical pallet to the same consignee this month — a conversation that consumes a phone call and produces no relationship. What stays on the phone is the awkward shipment: the ADR load, the timed delivery into a restricted site, the customer who needs a decision on a margin exception before lunch. Those calls are the relationship, and they are the ones a major answers with a ticket number.

Three design points decide whether the portal helps or hurts a small brokerage:

• It must be white-label. Your logo, your colours, your name on the quote. A portal that advertises the vendor tells your customer that the intelligence lives somewhere else.
• It must never expose the buy side. Buy rates, grids and margins are computed server-side and never sent to the browser; the customer sees a sell price and nothing behind it. The mechanics are in showing a price without revealing your buy rate.
• It must accept requests it cannot price. A portal that only answers on covered lanes trains your customer to stop asking, and hands the uncovered demand to whoever will take the call.

That last point is the one most brokers miss, and it turns the portal into something more useful than a quoting screen: a demand sensor. Every request that arrives with no matching grid is a lane your customers want and you cannot serve — which is your carrier recruitment list, ranked by real demand rather than by intuition. See quoting without a rate grid for the fallback path, and the customer quote portal guide for the setup. Cargavo, a TMS for transport commissionaires and road freight brokers, ships the portal in English, French, Spanish and German so a cross-border customer reads it in their own language.

Being explicit about this is not modesty, it is qualification. Every hour spent on an RFQ you were structurally unable to win is an hour not spent on the lane where your regional haulier was €106 cheaper.

Decline the tender when it requires EDI 204/214 messaging into the shipper’s ERP, a named control tower with contractual out-of-hours cover, bonded storage, or a multimodal scope you do not operate. Bid when the buying criteria are price on specific lanes, response speed, and someone who answers the phone by name.

The scope statement for the tool itself, so nothing here is oversold: Cargavo, a TMS for freight brokers of 2 to 25 people, has no carrier API and no EDI connectivity; it prices from rate sheets you rebuild. It is not a load board or freight exchange, not a fleet or dispatch system, not a tendering marketplace and not accounting software. Why that is a deliberate choice rather than a gap is argued in API vs EDI vs rate grids.

Set the cost against the worked example above rather than against a budget line. That single 6-pallet Lyon to Barcelona load produced €116.77 of gross margin at an 18 % margin on the regional haulier’s rate. Against that one shipment:

• Starter, €49 per month — 42 % of the margin on that one load.
• Growth, €149 per month — the margin on 1.3 such loads.
• Scale, €399 per month — the margin on 3.4 such loads.

The structural point is the pricing model, not the amount. Per-seat licensing taxes the exact behaviour a small brokerage needs: putting every operator on the system so that the 18:40 request has somewhere to land. When the third and fourth user cost extra, desks share a login, coverage narrows, and the response-time tail you were trying to fix comes straight back. Flat pricing makes full access the default. The full breakdown, including what setup and implementation typically add elsewhere, is in the TMS pricing guide and on the pricing page.

The other real cost is your own time at the start, and it is front-loaded rather than ongoing: converting each carrier tariff into a structured grid. That is an afternoon per carrier for a clean rate sheet, not a project — see how long a TMS takes to implement for a small broker. Cargavo, a TMS for road freight brokers and transport commissionaires, is configured by the broker on Starter and Growth; assisted grid build is what the Scale plan adds. Nothing in the model requires an integration with your carriers.

Positioning arguments are easy to believe and hard to verify. These four are countable from the reference sequences themselves — Q-YYYY-NNNN for quotes, S-YYYY-NNNN for shipments — so no reporting layer is required.

• Quotes issued per month. The count of Q-references created. Under manual quoting this number is capped by attention, not by demand, so it barely moves month to month. If it does not rise after the grids are loaded, the requests are still arriving somewhere the engine cannot see them — usually a personal inbox.
• Grid coverage rate. The share of requests where at least one grid returned a price. A low figure is a carrier problem, not a software problem, and no amount of configuration fixes it. Treat anything you cannot price as procurement input.
• Quote-to-shipment conversion. S-references divided by Q-references over the same period. Watch it alongside volume: conversion falling while volume rises means you are now quoting lanes you are not competitive on, which is useful information rather than a failure.
• Off-grid request share. The proportion of requests with no matching grid at all. This is your ranked carrier recruitment list — the lanes customers are actively asking for and you are handing to a major by default. Work it with the carrier onboarding checklist.

Add one qualitative check that no metric captures: how many of your incoming calls are now about non-standard shipments rather than repeat pallets. That ratio moving is the sign the portal is doing the job described earlier — routine demand handled by software, judgement calls handled by you. It is also the thing a large 3PL cannot copy, because their version of the same call is a ticket.

Cargavo, a TMS for freight brokers and transport commissionaires of 2 to 25 people, keeps quotes and shipments on one continuous reference chain so these four numbers are a matter of counting rather than reconstruction. If you are still assembling them from workbooks, start with the signs you have outgrown spreadsheets.