Chief Product Officer / United States
Three defaults changed, and the revenue they carried
- Paying carrier accounts
- 12,400
- June 2026; fleets of one to twenty-five trucks
- Trucks run on the platform
- 43,000
- Customer fleets, not company assets
- Company revenue, 2025
- US$88.4M
- Subscriptions and advance fees; not my budget
Decisions & consequences
01Making the broker's payment date the default and pricing each advance in dollars
In 2022, the screen a carrier saw after submitting an invoice had a large preselected button for same-day payment. The advance cost 3.25 per cent of the invoice. Among carriers with one to three trucks, 58 per cent of invoices were advanced, and the median one-truck account paid about US$4,700 in advance fees that year. When our researchers interviewed thirty owners, fourteen could not estimate their annual total within a thousand dollars.
Advance fees were about 39 per cent of Ostervane's revenue, so this was not a cosmetic question. I proposed that payment on the broker's own terms, which carries no fee, become the default; that every advance show its cost in dollars on the button; and that a next-business-day advance be added at 1.5 per cent alongside same-day at 2.75 per cent. Our chief financial officer projected US$6.8 million less fee revenue in the first year. Two directors asked us to test the change with a third of carriers first. I argued that a preselected fee was not something to keep for two-thirds of customers while we measured, and the chief executive agreed to a full release in May 2023.
In the twelve months to May 2024, the share of small-carrier invoices advanced fell to 37 per cent, and fee revenue was US$5.9 million below the previous plan. Twelve-month retention of one-to-three-truck accounts rose from 72 to 80 per cent; I cannot attribute all of that to this change, because we also improved document capture that year. The decision established a rule that every paid option must show its cost in dollars before the carrier commits.
I would do two things differently. I would have shown each carrier its 2022 fee total at the moment of the change, rather than only counting forward. And support received 1,900 calls in the first fortnight, because I briefed the team a week before release instead of a month.
Flagging load offers that a driver's remaining hours could not reach
By late 2023, about 61 per cent of trucks on Ostervane were connected to a logging device through our integrations. When we reviewed 212,000 offers accepted by those fleets in the fourth quarter, 8.6 per cent had pickup or delivery appointments that the assigned driver's remaining hours, as reported by the device, could not reach without a required break or reset. Those loads arrived late more than three times as often as others. Dispatchers were accepting first and solving later, and drivers were left with the pressure.
There were three positions. Our head of partnerships feared brokers would see lower acceptance and send fewer offers. Our external safety adviser wanted the app to block such acceptances outright. Several owner-operators on our customer council objected to software questioning their decisions. I chose a middle course and accepted its costs: an offer with a shortfall would show it in hours and minutes, with the data's age, and a dispatcher could still accept after recording a reason, such as a different driver, a team, or a new appointment agreed with the broker. We gave up a one-tap acceptance flow that dispatchers liked.
Launch in March 2024 exposed my own mistake. In the first month, about 5 per cent of flags were wrong because logging data from rural routes arrived late, and a driver who had already taken a break still showed short. We added the time each reading was received and suppressed flags on readings older than forty minutes.
Within six months, accepted offers with a shortfall fell to 3.1 per cent. Overall acceptance of broker offers fell 1.8 percentage points, and two brokers asked us to remove the flag; we declined. Our legal position did not change: the driver's device remains the record, and we say so on every screen. I would test data freshness on rural routes before launch next time, because a warning that is wrong in its first month loses the trust it needs for the next year.
Ending location sharing at delivery and accepting a broker's lost tenders
In January 2025, a freight broker that tendered about 11 per cent of the loads offered through Ostervane asked for continuous location from drivers' phones, from acceptance of a load until twenty-four hours after delivery. It wanted to plan reloads and check detention claims. At that time, our sharing began at acceptance and ended when the carrier marked the load delivered, but some carriers forgot, so a few drivers were visible overnight without knowing it.
Our chief executive and head of sales favoured agreeing, at least for carriers who opted in. Payment fees attached to that broker's loads were worth about US$1.7 million a year. I recommended the opposite: share only from arrival at pickup until thirty minutes after delivery confirmation, end sharing automatically even if nobody marks the load, show a visible indicator and the broker's map on the driver's own screen, and never share off-duty or personal time. Instead of continuous tracking, drivers could send a city-level availability signal when they wanted a reload. I took the proposal to the audit and risk committee with our privacy counsel, because an opt-in offered by a broker who controls a carrier's next load is not a free choice.
The board supported the narrower rule. The broker moved much of its tendering to its own app, and its loads through Ostervane fell 46 per cent over the next two quarters. Some of our smallest carriers lost easy access to its freight during that period, which I do not dismiss. Nine months later, the broker returned to about 70 per cent of its earlier volume under the new window. Drivers keeping location sharing switched on during loads rose from 81 to 93 per cent, and 3,800 carriers have used the availability signal.
The change now governs every partner data request. I would do one thing differently: write and publish the location policy before a large partner asked, rather than drafting it under commercial pressure and leaving carriers to learn the rule through a dispute.