Logistics and distribution
Can freight brokers license load and rate histories to AI developers?
By SourceX Editorial · Reviewed by Noah Loul ·
Short answer
Freight brokers can license load and rate histories to AI developers when the rights checks pass: shipper contracts allow de-identified use, carrier agreements and software terms do not restrict it, and third-party benchmark data is removed. Shipper, consignee, carrier and driver identities come out, locations are generalized and recent rates are held back before release.
Key takeaways
- Yes, if the rights checks pass; those checks set the scope before any record is prepared.
- Benchmark and load board data licensed from third parties stays out of every package.
- Shipper and carrier identities are removed, and lanes are generalized until none points back to one customer.
- Older, aggregated rate data raises fewer competition questions than current rates on identifiable lanes.
The short answer: yes, if the rights checks pass#
Freight brokers can license load and rate histories when they hold the rights to use the records and identities are removed before release. The developer receives permission to use a de-identified copy for named purposes and a fixed period; the broker keeps ownership of its TMS records and approves what goes into the copy.
Most of the work is in the checks, not the export. Shipper contracts, carrier broker agreements, TMS and load board terms, and competition considerations each shape what can be included. A package that skips them creates risk that outweighs whatever it earns.
A license is also not a live data feed. Packages are historical, prepared once or on a set schedule, and delivered as files the broker has reviewed. Open loads, current customer lists and today's rates stay out.
Which broker records make up a load and rate history?#
A broker's load and rate history is the set of records that tie a shipment to its price on both sides and to what happened in transit. Developers find the linked version far more useful than any single table, because it shows quote, cover, execution and outcome together.
Spot market benchmarks appear in many broker systems because reps use them to price. They belong to the subscription provider, not the broker, which is why that row is excluded by default.
| Record | Typical system | Useful fields | Licensing note |
|---|---|---|---|
| Load record | TMS such as McLeod, Aljex, Turvo or Revenova | Lane, equipment, dates, weight, status history | Core of any package |
| Customer quote and rate | TMS, quoting tool, email | Quoted rate, accessorials, win or loss | Shipper contracts may restrict |
| Carrier rate confirmation | TMS, document storage | Carrier pay, terms, accessorials | Carrier agreement terms apply |
| Tender and booking history | TMS, load board postings | Posted rate, offers, time to cover | Load board terms may restrict |
| Check calls and tracking | TMS notes, visibility tools | Status, delay reason, ETA changes | Driver details removed |
| Exceptions and accessorials | TMS, email, claims files | Detention, TONU, layover, claim outcomes | High value, needs review |
| Spot market benchmarks | DAT, Truckstop or similar subscriptions | Market rates by lane | Third-party data, excluded |
What do AI developers use broker histories for?#
AI developers use broker load and rate histories to train and evaluate models that quote freight, predict time to cover, flag likely exceptions and draft carrier and customer messages. The value comes from many real decisions with their outcomes, not from any single rate.
Evaluation is an underrated use. A developer testing a quoting or tracking agent needs real held-out cases to measure against, and a broker's quotes, covers and exceptions provide exactly that. Evaluation licenses can involve smaller, tightly scoped packages, which can make the rights review simpler.
Developers rarely want a broker's whole database. They want a documented slice: specific record families, a defined period, consistent fields and a clear description of how the records were prepared.
Which rights checks decide the scope?#
The rights checks that decide scope are shipper contracts, carrier agreements, software and data subscriptions, record-keeping duties and competition considerations. Each check can remove a customer, a record family or a field; together they define what the license can cover.
Silent contracts are common. Many shipper agreements say nothing about de-identified operational data, and counsel decides whether silence allows use, calls for consent or warrants exclusion.
| Check | What to look for | Common result |
|---|---|---|
| Shipper contracts and RFP terms | Rate and volume confidentiality, data use limits, ownership of shipment data | Restrictive customers excluded; others de-identified |
| Carrier broker agreements | Confidentiality of carrier pay, limits on using carrier information | Carrier identities removed; pay kept only in de-identified form |
| TMS and visibility vendor terms | Limits on extracting or reusing platform data | Export only through permitted routes |
| Load board and benchmark subscriptions | Bans on redistributing market data | All third-party benchmark fields removed |
| Broker record-keeping duties | Records you must keep and produce | Originals untouched; licensing uses copies |
| Competition considerations | How current and identifiable shared rates are | Recent periods held back, thin lanes aggregated |
What gets removed or generalized before release?#
Before release, everything that identifies a shipper, consignee, carrier or driver is removed, and anything that could point back to one customer's freight is generalized. The aim is to keep the pricing logic and the outcome while making the parties unidentifiable.
Every rule is tested before release with the checks listed here: a reviewer looks for any record that, alone or combined with public information such as carrier registration data, still points to a specific shipper or carrier, and tightens the rule when one does. The table shows typical field-level treatments as a starting point, not a fixed recipe.
- Run the rules on a sample month before the full history.
- Search the sample for any lane that only one customer ships, and suppress or merge it.
- Check that no carrier can be inferred from a lane plus a date plus equipment type.
- Record each rule and its test result for the privacy record.
| Field | Typical treatment | Why |
|---|---|---|
| Shipper, consignee and bill-to names | Replaced with consistent tokens | Keeps repeat-customer patterns without names |
| Pickup and delivery addresses | Generalized to a region or market area | Exact facilities identify customers |
| Carrier name, MC and USDOT numbers | Removed; carrier replaced with a token if needed | Registration numbers are public and searchable |
| Dispatcher and driver names and phones | Removed | Personal data with no modeling value |
| Customer rate and carrier pay | Kept, de-identified; recent periods held back | The spread between them is the core pricing signal |
| Pickup and delivery dates | Kept, or shifted to week for thin lanes | Timing matters, but exact dates can single out a load |
| Notes and email text | Scanned for names, phones and reference numbers | Free text leaks identities that fields do not |
| Third-party benchmark values | Dropped entirely | Belongs to the subscription provider |
Does sharing rate data raise competition concerns?#
Rate data can raise competition questions, mainly when pricing that is recent and tied to identifiable lanes or customers could reach someone in a position to coordinate prices. A historical, de-identified training set licensed to an AI developer is a different situation from passing this week's lane rates to another broker, but antitrust counsel should make that call for each package.
Practical safeguards include holding back recent periods, aggregating thin lanes, limiting the developer's permitted use to model training and evaluation, and barring resale or redistribution of the raw records. Those terms sit in the license and are recorded in the SourceX Evidence Packet.
Illustrative: a regional brokerage scopes a rate history package#
Illustrative: a fictional regional brokerage of about 90 employees, focused on refrigerated freight, has 11 years of history in its TMS, including quotes, carrier rate confirmations, check calls and detention claims. The CEO wants to know whether any of it can be licensed.
Counsel reviews the largest shipper contracts. Most are silent on de-identified data; one prohibits any use of shipment information beyond the services, so that customer's loads are excluded. The brokerage's load board subscription bars redistribution, so posted market rates are dropped.
The resulting scope covers older quote-to-delivery chains with shippers tokenized, carriers removed, locations generalized to regions and detention outcomes kept. The CEO approves the scope before any record is prepared.
How SourceX handles broker histories#
SourceX handles broker histories through the SourceX five-step transaction: Supply, Rights, Preparation, Approval and Delivery. The fit check uses metadata such as TMS name, years of history and record families, so nothing is shared to learn whether a package makes sense.
Rights review covers shipper, carrier, vendor and subscription terms; Preparation removes identities and generalizes lanes; the broker approves the final package. A SourceX Evidence Packet records provenance, licensing rights, permitted use, the privacy record and release authorization for every delivery.
Frequently asked questions
Do we have to tell shippers we are licensing data?
Not always; it depends on each contract. Where agreements require consent or restrict use, the shipper is asked or its loads are excluded. Where they are silent, counsel advises. Separately, the CEO may decide that a few strategic shippers should hear about the program directly from their account manager.
Can carrier pay stay in if carriers are de-identified?
Often, in de-identified form. Carrier pay paired with the customer rate is central to how brokerage pricing works. Carrier broker agreements are checked for confidentiality terms, carrier identifiers are removed, and thin lanes where a carrier could be inferred are aggregated or suppressed.
Does licensing affect our record-keeping obligations?
It should not, because licensing works on copies. Original records stay intact in your TMS and archives for as long as your obligations and contracts require. The license gives the developer use of a prepared copy and does not transfer or alter the records you must keep.
What makes one brokerage's history more useful than another's?
Linkage and depth. A history where each load connects its quote, carrier cover, tracking, exceptions and settlement is more useful than separate tables. Consistent status codes, preserved emails and notes, and history that survived TMS migrations matter more than raw load counts.
Is our rate history worth anything once identities are removed?
Possibly. Developers value the relationships between lanes, timing, rates and outcomes, and those survive de-identification. Value is only known once a developer engages with a specific, documented package, so be cautious of anyone quoting a figure before scoping and rights review.
Related resources
See if your company qualifies
A short company assessment. No data uploads are needed.