You‘ve probably heard some version of this argument. Rate data built on invoices has to be old, because invoices come after delivery. The argument states that by the time a settled transaction reaches DAT’s database, the market has moved on. But that’s not how the DAT database works, and we measured over 20 million loads to see just current we are.
Where DAT’s spot data starts
DAT spot rate data starts with the rate confirmation, the document that exists the moment a broker and a carrier agree on a price. That agreement happens before the truck loads, not after it delivers. Settlement then reconciles the record as the load closes out.
The invoice argument holds up fine against rate data built on billing records, and some products are built that way. Ours isn’t, at least not on the spot side.
Contract freight works differently. Contract rate data does come from freight bills, and it arrives on a slower timeline. That’s a function of how contract freight settles rather than a pipeline problem, and it’s why contract rate data should be seen as benchmarks rather than a read on the week.
Who sends us this data, and why they bother
Brokers, carriers, and shippers send us their completed transactions because they want to know how their own pricing compares to the rest of the market. To get the best, most accurate data for pricing, contributors are incentivized to send the most up to date and accurate data for the pool.
The uploads also run automatically out of contributors’ transportation management systems, so nobody is picking and choosing which loads to send us.
It compounds, too. Every additional contributor tightens what the system can resolve, because a lane carrying more real transactions can be reported at a narrower geography and a shorter timeframe. A feature list won’t show you that. The breadth of the contributor base is what decides whether the specific lane you care about comes back with a number worth quoting.
What the timing data shows
If the record starts at agreement rather than at billing, contribution timing should show the data arriving early, and it does.
We looked at broker spot freight, the largest rate segment DAT collects, across every credited contribution with a 2026 pickup date. That’s a little over 20 million loads. For each one we measured the gap between the estimated delivery date and the moment DAT first accepted a contribution.
About a third of broker spot loads reach us before the truck is even due to deliver.
Two-thirds are credited within 24 hours of estimated delivery. The median load lands about half a day after its estimated delivery date.
Long-haul freight shows the largest gap. On hauls of 551 miles or more, close to six in ten loads are credited before their estimated delivery date, since a longer transit gives the paperwork more room to arrive first.
The question worth asking instead
Ask what event creates the record, because that sets a hard floor on how current any rate dataset can be. It holds up no matter what a vendor claims about pipeline speed.
Factoring activity is a post-delivery financial event by definition, since a carrier can’t invoice a load that hasn’t moved. A dataset assembled from factoring can’t contain a pre-delivery rate, because that record doesn’t exist yet.
Bids and quotes are asking prices. They tell you what someone hoped to get, not what anyone agreed to.
Inferred signals like tender acceptance patterns, capacity behavior, or fuel and GDP trends can produce a perfectly defensible number, and they still can’t trace it back to two companies agreeing on a price for one specific load.
Booking-moment assessments capture the agreement, which puts them in the same event class as our spot data. We’d rather say that plainly, because it’s where an honest comparison actually starts.
So what separates one dataset from another?
If two datasets capture the same event, timing stops being the interesting question and three others take over.
How many companies stand behind the number. A rate drawn from a broad contributor base spanning brokers, carriers, and shippers describes the market. A rate drawn from an invitation-only pool of large 3PLs describes that pool and then extrapolates. Both are real data, but they aren’t the same population, and enterprise freight prices differently than small-fleet freight.
What the sample floor is. Every DAT market rate has to clear a minimum number of load reports from at least three separate contributing companies before we’ll show it, and no single company can move a lane on its own. Ask any provider what their floor is and what happens on a lane that doesn’t meet it. Weighting and interpolation can fill a gap, and a filled gap is a model output sitting inside something that looks like an observation.
Whether the record gets corrected. An assessment taken at the moment of cover is frozen there. Accessorials, detention, and the corrections that follow a load through settlement never reach it. A record that reconciles picks all of that up, and that’s what makes a number auditable and not just early.
What this doesn’t cover
This is broker spot data. Shipper contract data behaves differently and lands considerably later. Anyone quoting broker spot timing as though it describes all rate data is overreaching, us included.
Delivery dates in this analysis are estimated. DAT doesn’t capture actual delivery times reliably enough across millions of contributions to use them directly, so we derived each load’s delivery date from its mileage. Checked against the subset of loads that do carry a usable delivery date, the estimate tracked closely on short and mid-haul freight and ran slightly early on long hauls. Running slightly early means the long-haul figures here understate the real gap instead of flattering it.
What freshness does and doesn’t buy you
Freshness is one property of rate data, and treating it as the whole scorecard will lead you into false security.
A very current number built on three loads from one contributor is worth less than a three-day average built on forty loads from a dozen companies. Sample depth, contributor spread, and geographic precision all move the accuracy of a benchmark more than a few hours of recency does.
So treat freshness as one filter among several, then ask the harder questions. What event created this record? How many companies stand behind it? What geography did the tool have to expand to before it found enough data? Does the number get corrected when the load closes out?
Those are the answers you’ll want when a shipper pushes back on your quote, and they hold up better than a speed claim from anyone, us included.
