A Midwest freight brokerage kept losing loads to whoever replied first. We rebuilt intake so every request is parsed, priced and routed to a named broker in minutes.
Hugo takes the project apart on screen: the bottleneck we found, the system we shipped, the number it moved.
No walkthrough yet. The written case below carries the same numbers.
Northfield Freight brokers full truckload and LTL lanes out of Columbus, Ohio. Thirty-four people, two satellite desks, around 900 quote requests a month. Revenue had grown three years running.
Every one of those requests landed in a shared inbox that four brokers watched between calls. Nothing routed itself. A request from a $2m account sat next to a one-off pallet quote, in the order it arrived.
Median first response was four hours and eleven minutes. Freight does not wait that long. Quotes sent inside thirty minutes closed at 41 percent. After three hours the same desk closed 9 percent.
Follow-up was worse. A broker would mean to chase a quote on Thursday, then a truck fell through and the note never left the notebook. Nobody could say how many quotes went cold, because nobody counted them.
Two days of interviews and one week of their own data. Here is what was actually going wrong, told the way the people living with it told us.
The inbox was the whole problem. Four brokers watched it between phone calls, so a request moved only when somebody happened to look at the right moment.
None of them were slow. A broker would open a quote, get pulled onto a load that was falling apart, then come back to nine more requests stacked underneath. Whatever sat on the screen won.
Follow-up died the same way. Chasing a silent quote is always the job you will do once the urgent thing is finished, so it stayed undone. Brokers knew they were losing freight like this. Proving it was another matter.
Built inside the tools the company already paid for. Every workflow, credential and prompt was handed over at the end.
Every inbound request is read by a model that pulls out the lane, the equipment and the pickup window. It scores the shipper against twelve months of volume, then assigns a broker with a deadline.
Gmail
Claude
n8n
HubSpot
AirtableThe broker opens one screen with lane history, current market benchmarks and the margin floor already applied. A draft quote sits underneath it, written in Northfield wording.
n8n
Google SheetsAirtableClaudeHubSpotAny quote without a reply gets a second touch on day two, a third on day five. Both are written against the original thread. One click in Slack stops the sequence.
HubSpotn8nGmail
SlackBefore figures come from the company's own records, pulled during the audit. The after column is the ninety day average once every system was live.
| Metric | Before | After |
|---|---|---|
| Median first response | 4 h 11 min | 11 min |
| Quotes sent per week | 168 | 232 |
| Open quotes with a second touch | 31% | 100% |
| Broker hours in the inbox, daily | 10 h | 2 h |
| Quote win rate | 17% | 24% |
| Monday pipeline report | 3 h by hand | Ready at 07:00 |
We thought we had a sales problem. What we had was a four hour gap between a shipper asking and us answering. Close the gap and the same brokers win more freight.
I stopped chasing my own team for updates. The board report writes itself now, and I trust the numbers in it.
Every project has them. Writing them down is how the next one gets shorter.
We switched on follow-up before the rate logic was settled, so the first two weeks produced polite emails carrying prices nobody would stand behind.
Building the rate rules first would have saved ten working days.
We also underestimated how much the satellite desks priced by feel, which cost a second round of interviews.
First system in production inside a month. Everything after that is iteration with the client in the room.
Tickets answered under 15 min: 74%
Read the caseSummary, was 2 days: 5 min
Read the caseThe AI and Operations Audit finds it, prices it, then hands you a plan in two working days. Your fee comes off the first thing you build with us.
