Most small shippers do not need freight booking software. They need a rate they can trust without a phone call, a booking that lands in the carrier's system without re-typing, and a status they can give a customer without chasing a driver. Plenty of products deliver the first and quietly fail the second.
The market's clearest lesson arrived in 2023. Convoy, founded in 2015 and funded with more than a billion dollars, shut down its brokerage in October 2023, and Flexport later acquired its technology assets. A booking screen that customers liked did not survive brokerage economics. Judge anything you adopt in 2026 on cost per shipment, not on the quote page.
Four product categories get called the same thing
Vendors all describe themselves as freight platforms, which makes shortlists incoherent. These are different products with different failure modes.
| Category | What it does | What it will not do | Examples of companies in this category |
|---|---|---|---|
| Digital forwarder or broker portal | Quotes and books capacity the provider itself sells | Give neutral market rates, since the provider is a party to the deal | Flexport, founded 2013; Zencargo, founded 2014 |
| Marketplace and rate aggregator | Compares quotes from many providers on a lane | Manage execution depth when something goes wrong | Freightos, listed on Nasdaq as CRGO in January 2023 |
| Transport management system | Holds orders, carriers, rates and documents, then books against them | Find capacity, unless you already have carrier relationships | Descartes, founded 1981; MercuryGate, founded 2000; Transporeon and Alpega in Europe |
| Visibility layer | Collects telematics and carrier feeds to predict arrival | Book anything at all | Project44 and FourKites, both founded in 2014 |
Two deals, one closing in 2021 and one in 2022, show how fast this market consolidates. Uber Freight acquired Transplace in a deal reported at 2.25 billion dollars, closing in 2022, which put a brokerage and a shipper-side transport management system under one owner. Panasonic completed its purchase of Blue Yonder in 2021 at a reported 7.1 billion dollars. If a shortlist mixes two of the four rows above, the demonstrations will not be comparable, and consolidation means the categories keep blurring in marketing while staying distinct in operation.
Integration decides the value, not the interface
Every booking tool looks similar in a 45 minute demonstration, because the demonstration ends when the booking is confirmed. Your cost lives after that moment, in whether the booking reaches the carrier as data or as a human reading an email.
Road freight in North America runs on a small set of EDI transaction sets, and the numbers are worth knowing before a sales call. A tender goes out as a 204, the carrier answers with a 990, status updates return as 214, and the invoice arrives as a 210. A 997 acknowledges receipt, and warehouse flows add the 856 advance ship notice. European operations more often use EDIFACT messages such as IFTMIN for the instruction and IFTSTA for status. Newer platforms offer REST APIs instead, which are faster to build against and less widely supported by the small carriers a growing shipper actually uses. Where those APIs do exist, they are what makes automated quoting possible at all.
Ask every vendor these questions before discussing price:
- Which of my current carriers are connected today? Not connectable. Connected, in production, for another customer, with a name you can call.
- Who pays for a new carrier connection? 6 weeks per carrier is a common timeline and it rarely appears in the proposal.
- Does status come from the carrier's system or a driver app? App-based tracking collapses when the carrier subcontracts, which happens on most spot loads in Europe and North America.
- Can I export my rate table and 24 months of shipment history? A vague answer means your rate data is the vendor's asset rather than yours.
- What happens when an API call fails? The honest answer involves an email fallback, and you want to hear it before a Friday afternoon.
- How are accessorial charges coded? Free-text accessorials defeat automated invoice matching, which is where the promised saving was supposed to come from.
- Which EDI sets are live, by number? A vendor who cannot say 204, 990, 214 and 210 without checking has not run road freight at volume.
The three pricing shapes
Subscription pricing per user or per volume band is the most predictable and the easiest to compare across a shortlist of 4 vendors. Per-transaction pricing suits irregular volume and penalises growth, so model it at 3 times your current shipment count before signing. Embedded pricing, where software is free because the provider earns margin on the freight, deserves the hardest look: the tool genuinely costs nothing and the rate is not visible, and no workflow inside that platform will prompt you to test the market.
I have watched a small shipper adopt an embedded platform, cut administrative time by a real amount, and lose more than that on lane rates inside 12 months. If you choose that model, keep 2 or 3 lanes quoted elsewhere every quarter as a control, and diary the review so it survives a busy month.
What actually moves cost per shipment
Ranked by the effect I have seen rather than by how modern each sounds:
- Fixing the address book. Failed deliveries from bad addresses generate re-delivery and address-correction charges, which DHL, UPS and FedEx each bill as separate line items. This is a data cleanup, not a purchase.
- Measuring dimensions correctly. Parcel carriers in the United States apply a dimensional divisor of 139 for domestic shipments and 166 on many international services, while air freight uses the IATA volumetric standard of 6,000 cubic centimetres per kilogram. Guessing at any of those 3 figures produces the most common invoice adjustment in small operations.
- Booking earlier. Lead time buys rate, and no platform compensates for tendering at 4pm for a 6am collection.
- Consolidating shipments. 2 pallets moving separately on one lane on one day is the most reliable waste in small operations, and any system that surfaces it pays for itself.
- Watching port charges. Should your volume include containers moving from a terminal, drayage and dwell charges will dominate the invoice long before any subscription does.
- Matching invoices automatically. This needs coded charges from the carrier, which is why the accessorial question matters more than the dashboard.
- Holding one status source. Customers do not care which system knows, only that your answer does not change between two people.
A two-week shortlist process
Pick five lanes that represent most of your volume, plus one that always causes trouble. Ask each vendor to quote and book those six shipments, using your own carriers where you have them and a copy of your real rate table rather than sample data. Then measure four things that will recur every day: minutes per booking end to end, the number of fields you re-type, whether the platform status matches the carrier's own portal, and whether the invoice can be matched without a human opening it.
Two weeks of that beats any feature matrix, because it separates a product that books freight from a product that books freight into your carriers' systems. Set a pass mark before you start, something as blunt as eight minutes per booking and zero re-typed fields, so the decision does not turn into a preference contest. The second kind of product is rarer than the marketing suggests, and it is the only kind that removes work rather than moving it to a different screen.