AI load-board scoring and carrier recommendations in freight

Machine-learned scoring ranked loads and recommended carriers in real time on a high-volume freight platform, lifting engagement, conversion and margin together.

LogisticsAI/MLLoad board

The situation

A load board is a marketplace with a ranking problem. Show a carrier the wrong loads and they stop looking; show a shipper the wrong carriers and the load does not move. Most boards sort by what is easy to sort by — recency, lane, rate — which is not the same as what a particular carrier is likely to accept and deliver well.

The cost of that gap shows up in three places at once: carriers browse less, fewer loads convert, and the loads that do convert are matched at worse margin than they should be.

What we did

We built scoring models that rank loads for each carrier and recommend carriers for each load, scoring in real time as the board changes rather than in an overnight batch. The models learn from what actually happened — which loads were viewed, accepted, delivered on time — so the ranking improves as the marketplace runs.

Real time matters more than model sophistication here. Freight availability changes by the minute, and a recommendation computed last night is a recommendation about a market that no longer exists.

The outcome

  • Engagement up — carriers spent more time on a board that showed them relevant work.
  • Conversion up — more of the loads shown were accepted.
  • Margin up — better matching improved the economics of the loads that moved.

What made the difference

Treating ranking as a product surface rather than a reporting output. The score had to be computed where the decision was made, fast enough to be current, and fed by outcome data rather than by assumptions about what carriers want. The data pipeline work that makes this possible is what our DATANODE framework covers.

We publish the metrics and withhold the names — our clients’ competitive advantage isn’t ours to advertise. Talk to us about applying this to your marketplace.