Freight brokers are growing revenue and losing margin at the same time, because carrier costs are rising faster than customer pricing can follow. In Q2 2026 that gap stopped being a forecast and started showing up in earnings. The brokers who protect profit through Q4 won't be the ones who priced best — they'll be the ones who cut the cost of servicing each load.

Here's what the Q2 numbers actually say, why the obvious fix doesn't work in a carrier's market, and the one lever most brokerages haven't pulled.

What happened to broker margins in Q2 2026?

Broker gross margins compressed across the public peer group in Q2 2026 even as load volumes grew. The clearest picture comes from the public brokerages, because they have to show their work.

J.B. Hunt's brokerage segment grew revenue 49% and loads 19% year over year — a genuinely strong quarter by volume. But purchased transportation expense rose 54%, outrunning revenue growth, and segment gross margin fell from 15.5% to 12.5%. Three hundred basis points, absorbed in a single quarter of growth.

RXO told a similar story. Companywide gross margin dropped to 13.9% from 17.8%, and brokerage gross margin slipped 70 basis points sequentially to 10.7%, with higher fuel costs a named contributor. Covenant Logistics was blunter about the mechanism: the cost of securing quality brokerage capacity outpaced its ability to secure contractual rate increases.

The pattern held across the peer group. Volume growth did not protect margin — in several cases it hid the damage until the quarter closed.

Why can't brokers simply raise prices?

Brokers cannot reprice their way out of margin compression in 2026 because the capacity shortage is structural rather than cyclical, and customer contract rates stay locked until renewal. The market moved to the carrier's side of the table, and it moved for reasons that don't reverse in a quarter.

Truckload all-in rates are running roughly 50% above year-ago levels. Tender rejections have eased from a 17.65% peak to 14.1%, but that is still nearly three times the 4.75% baseline of a year ago. At 14%, roughly one in seven contracted loads falls out of the routing guide and becomes an unplanned, urgent spot event — priced at whatever the market demands that hour.

The supply side isn't loosening either. The constraints driving this are driver scarcity, insurance, maintenance, and regulatory enforcement — cost structures, not seasonal swings. Most 2026 outlooks expect elevated rates to hold through Q4 and into 2027, with Q3 truckload rates projected to hit a four-year high.

So the buy side keeps climbing while the sell side is anchored to contracts negotiated in a softer market. Repricing helps at renewal. It does nothing for the loads moving this week.

The lever most brokerages haven't pulled: cost per load

Cost to serve is the number of human minutes required to move one load from inbound request to invoiced. It is one of three inputs to broker profit — and in 2026 it is the only one you still control:

• Buy rate (carrier cost) — Who controls it in 2026: The capacity market · Realistic near-term movement: Rising — structural, not cyclical
• Sell rate (customer price) — Who controls it in 2026: Contract cycles, procurement · Realistic near-term movement: Locked until renewal
• Cost to serve (touches per load) — Who controls it in 2026: You · Realistic near-term movement: Immediately addressable

Cost to serve rarely appears on a brokerage dashboard, which is exactly why it drifts. When volume climbs 19%, the headcount required to service it climbs too — unless the work per load falls.

That is the arithmetic behind the Q2 results. Growing 19% on a flat cost-to-serve means hiring into a compressing margin.

Which touchpoints actually compress?

Not every touchpoint is worth automating. The freight workflows that compress are high-volume, repetitive, and message-shaped — the traffic that consumes ops hours without requiring judgment:

• Inbound RFQs. A quote request arriving at 2am sits until morning. In a market where rate validity is measured in hours, the first credible quote often wins the load.
• Load tendering and rate confirmations. Structured back-and-forth across email, chat, SMS, and WhatsApp, reconciled against the TMS.
• Check calls and ETA updates. Pure status relay, endlessly repeated.
• "Where is my order?" messages. The single largest category of inbound volume for most brokerages, and the one least dependent on human expertise.
• Change-order and document reconciliation. Matching what was agreed against what arrived.

What stays with people: rate commitments outside approved bands, claims, service failures, and any conversation where the relationship is the product. That boundary should be numeric and written down — "auto-quote within X% of the rate band, escalate beyond it" — not left to judgment in the moment.

What to do before Q4

Peak season is arriving early and capacity is tightening into it. Four moves, in order:

1. Measure cost to serve. Total ops labor cost ÷ loads moved, by lane and by customer. You will find customers whose volume is unprofitable at current touch counts.
2. Count the touches. For one week, log every human interaction per load. Most brokerages find 20–40 touches where they assumed 10.
3. Automate the top two categories only. Usually inbound quoting and status/WISMO traffic. Resist the platform-wide rollout — narrow scope is what separates the automation projects that ship from the ones that get cancelled.
4. Shorten quote validity and separate fuel from linehaul. If buy-side costs move daily, a five-day quote is a free option you wrote for your customer.

The bottom line

The Q2 2026 results are not a demand problem or a sales problem. They are a spread problem: carrier costs rising faster than customer pricing, on volume that is genuinely growing. You cannot out-price a structural capacity shortage, and you cannot wait out a cost base that is rising for structural reasons.

What you can change is how many human minutes each load consumes. In a quarter where a 19% volume gain still cost 300 basis points of margin, cost to serve is the only lever fully in your hands.

Debales deploys AI agents that handle freight quoting, order processing, ETA updates, and multi-channel customer communication end to end — across email, chat, SMS, and WhatsApp, integrated with your existing TMS. [Book a demo](https://debales.ai/book-demo) to see sub-60-second quoting on your lanes.