Truck-Load Optimization
Can a roll-specific load optimizer deliver measurable U.S. savings, or is it a narrow feature buried under integration and safety risk?
Can one depot prove that better roll loading actually removes trips or labor, rather than simply producing a better-looking plan?
The trap is paying for sophisticated optimization while data cleanup, implementation, and human approval consume the savings.
Jake sees a trust-led wedge: prove safer, fuller roll loads at one depot, make planners faster, and expand through a flagship customer.
Sarah pressure-tests savings, site pricing, acquisition costs, and implementation effort, backing only a paid pilot tied to measurable outcomes.
Ryan focuses on the constraint engine, messy freight data, planner overrides, safety boundaries, and a narrow prototype before costly integrations.
The math that has to work
SARAH'S BACK-OF-NAPKINA depot handling 100 weekly loads could create $5,000 in monthly value under the modeled assumptions, making a $2,000 monthly site subscription defensible.
That case weakens quickly when implementation costs $10,000–$20,000 and direct customer acquisition can reach $15,000–$30,000.
This week on Tri Diligence, the idea is software that plans how to load cylindrical and irregular cargo—starting with paper rolls—into trucks and trailers. It combines load geometry with practical constraints such as weight distribution, axle limits, securement, dunnage, and unloading sequence.
The central question: is this a defensible optimization business that can reduce trips and planning time, or a narrow feature wrapped in expensive integration work and operational risk? The hosts examine the U.S. trucking opportunity, existing load-planning products, the challenge of collecting reliable shipment and trailer data, and why software recommendations cannot replace human approval of a safe load.
- Jake (the marketer) argues for a focused, trust-based offer to paper shippers and logistics operators: prove better loads at one depot before claiming to reinvent freight.
- Sarah (the backer) challenges the savings assumptions, tests site-based pricing and implementation economics, and asks whether customers will pay for measurable reductions in trips, labor, or damage.
- Ryan (the technologist) breaks down the data and constraint engine, integration strategy, planner overrides, and the role AI can play around—not instead of—the safety-critical optimizer.
The proposed pilot starts with one freight type, one depot, two trailer types, and a small number of repeatable routes. It compares planned loads with actual outcomes and tracks planning time, utilization, avoidable trips, overrides, and safety. The recommendation is to prove customer value and repeatable onboarding before building a broad platform or expanding to other cargo categories.
Transcript
Welcome to Try Diligence, the show where three people examine one business idea until it either becomes a company or a very expensive slide deck. I'm Jake, and today I'm excited about software that tells trucks how to load awkward freight, starting with paper rolls.
I'm Sarah, and I'm already worried about the slide deck. This is a business selling math to people whose trucks can't afford a bad answer.
And I'm Ryan. I build things, then ask what happens when the real warehouse ignores the thing because the roll dimensions in the system were typed in by somebody wearing gloves.
Fair. The pitch is a load optimizer for cylindrical and irregular cargo. It plans roll placement, weight, axle loads, unloading sequence, dunnage, and securement. The warehouse gets a practical visual loading plan, not a beautiful little digital sculpture.
The customer isn't everybody who owns a truck. Start with a paper mill, distributor, or large third party logistics provider with recurring roll loads and a painful exception. Half empty trailers, rejected loads, damage, or one heroic planner who apparently has the whole business stored behind their forehead.
That heroic planner is a real customer journey problem. If Maria has loaded paper rolls for twenty years, the product can't arrive saying, "Good news, computer, replace Maria." It needs to make Maria faster, safer, and impossible to lose to a competitor.
The Swedish wedge makes sense as a proving ground because Sweden produced eight million tonnes of paper and board in twenty twenty-four, according to the Swedish Forest Industries Federation. But commercially, the United States is the big benchmark. American trucking generated about nine hundred six billion dollars in freight revenue in twenty twenty-four.
And don't call that nine hundred six billion dollar market the addressable market. The serviceable market is plants and logistics operators handling repeatable difficult loads. A tiny slice of a giant freight economy is still tiny if every sale needs nine months of procurement and six months of data cleanup.
I'd target one depot with at least one hundred roll loads a week. The buyer is an operations manager who owns trailer utilization or freight spend, while planners, warehouse teams, and drivers are daily users. Everyone else is just enjoying the meeting.
There's a modeled case in the brief: one hundred loads weekly, a thousand dollar trip cost, and a five percent savings. That creates roughly five thousand dollars of monthly value before implementation. It's promising, but only if the five percent means avoided trips or measurable labor, not a more pleasing dashboard.
Exactly. Better utilization doesn't automatically remove a truck. If a route leaves five days a week regardless, fitting another roll may improve margin but not eliminate a trip. The pilot needs a baseline: planned cubic utilization, actual loads, extra trips, planning minutes, damages, and rejected plans.
The value proposition should be brutally specific: load more safely, plan faster, and know whether a shipment can leave before the trailer is sitting at a dock with everyone staring at a roll like it has betrayed them.
There are competitors, though. EasyCargo lists about seven hundred ninety-nine dollars per user annually and already offers three dimensional planning, spreadsheets, reports, and weight awareness. Goodloading handles axle loads, center of gravity, and multistop sequencing. This product can't win by drawing nicer boxes around circles.
Worse, Load Xpert says it has supported paper rolls, trailers, railcars, axle loads, and securement rules at a paper mill since two thousand six. The wedge isn't uncontested. The differentiator must be validated roll-specific constraints and a workflow that makes a real plan executable.
Which is where brand matters. Sell confidence, not optimization. Every plan should explain why: this roll orientation protects crush limits, this dunnage pattern prevents movement, this axle stays within limit, and this unload order avoids digging out the last stop with a forklift.
Technically, the first version should be a browser application, probably Shopify is obviously not relevant here, so a conventional web stack with a solver service. Import orders from spreadsheets first. Build connectors to transportation management, warehouse management, and enterprise resource planning systems only after the workflow proves value.
Thank you for not proposing custom integrations on day one. Integration is where gross margins go to have a long, expensive nap. A managed optimization service might be smarter initially: charge for setup, have a specialist review plans, learn the constraints, then automate the repeating cases.
I like that because it creates white glove onboarding without pretending it's pure software from minute one. The customer gets results immediately, and the company learns what planners actually override. Those overrides are product research wearing a hard hat.
But it must've an expiration date. If every new depot requires an engineer to interview five forklift drivers, map twelve trailer variants, and repair spreadsheet columns, this is consulting with a solver attached.
Pricing should follow value and avoid surprise bills. I'd test a site subscription of perhaps two thousand dollars monthly for one depot, plus a one time implementation fee of ten thousand to twenty thousand dollars. For a customer with five thousand dollars monthly value, two thousand is defensible.
Would you charge per shipment instead? That makes the value feel fair for smaller operators.
Not first. Per shipment produces bill anxiety precisely when volume spikes. A site plan with a throughput band is simpler. For enterprise customers, add implementation and an annual platform commitment. Planner seats are a weak unit because the savings come from loads, not mouse clicks.
The core data model isn't glamorous, but it's the company. Every roll needs diameter, width, weight, orientation permissions, crush limits, and destination. Every trailer needs interior geometry, axle positions, load limits, tie points, and securement rules. Then you add dunnage and unloading order.
That sounds like a lot to ask of customers.
It's. Which is why the narrow pilot is one freight type, one depot, two trailer types, and perhaps three common routes. No cameras, no special hardware, no mysterious computer vision tower in the loading bay. Structured exports and a human verification screen are enough.
The liability line needs to be painfully clear. In the United States, Federal Motor Carrier Safety Administration rules cover paper roll securement against rolling, shifting, and tipping. Software recommends. The carrier or shipper approves the final load and remains responsible. That isn't just legal language. It's product design.
Make approval visible. The plan has a checklist, named approver, and printable instructions for the driver. That actually helps adoption. Nobody trusts a black box that says, "Relax, the cylinders have been optimized."
Now the required AI question: what does AI do for us, and against us? For us, machine learning can flag bad input data, predict likely planner overrides, recommend familiar load patterns, and turn messy freight descriptions into structured fields. A language model can answer why a plan chose a placement, but it can't be the safety engine.
And against us?
A well-funded transportation software company can add an AI assistant to its existing warehouse or transportation management platform, ingest years of customer loads, and offer a good enough roll planner as a feature. Then this startup gets commoditized before its logo finishes drying.
Unless it builds the best constraint library and proof dataset. Every approved plan, actual load outcome, damage incident, and planner override can improve a defensible benchmark. The moat isn't that circles are hard. The moat is knowing which circles behave badly in which trailers.
Data can be a moat, but only after scale. Before scale it's a promise written in very confident font. Large incumbents like Blue Yonder already own broader workflow, including constraint based load building and unloading sequence. They can bundle. This startup has to be dramatically better for a narrow pain.
Channels should reflect that. Don't buy broad search ads for "truck software" and accidentally meet every person who owns a pickup. Sell directly through paper industry relationships, freight associations, trailer makers, cargo securement consultants, and system integrators serving mills and distributors.
Integration partners could become referral partners, but be cautious. A transportation management vendor might welcome a specialized solver, or decide to build one after the first demo. The product needs clean application programming interfaces, but it shouldn't hand over its constraint engine like party favors.
Customer acquisition cost will be high because this is enterprise operations software. Assume a direct sale costs fifteen thousand to thirty thousand dollars in travel, sales time, proof work, and technical support. At two thousand dollars monthly, that means payback is ugly unless retention is strong or implementation revenue covers the effort.
Then the early motion should be founder led and concentrated. Land one flagship mill or distributor, publish verified operational results, and use that case study to sell neighboring sites. A planner saying, "We stopped building every load from scratch," beats a thousand polished ads.
Provided the case study is real. EasyCargo reports customer claims of thirty percent faster planning and at least twenty percent lower inefficient-space expense, but those are vendor reported claims. Our pilot shouldn't borrow someone else's victory lap.
Success criteria: reduce planning time by at least twenty percent, improve utilization enough to avoid or consolidate a measurable number of trips, and maintain zero safety incidents attributable to the plan. If the team can't get clean data and repeat those results within ninety days, pause.
I'd add user adoption. If planners override more than, say, one out of every four recommended plans, find out why. Maybe the solver is wrong. Maybe the loading crew has an undocumented reality, like a forklift turning radius that exists only in Dave's memory.
What has to be true for this to work? First, enough customers have recurring inefficient roll loads. Second, they can prove savings. Third, data onboarding is bounded. Fourth, legal review doesn't turn each sale into a bespoke liability negotiation. Fifth, the product expands beyond paper before the niche caps growth.
Expansion could be steel coils, cable drums, pipes, and mixed industrial freight. But not in the first pilot. Every new geometry family changes physical constraints and testing requirements. The fast way to ruin a good wedge is to announce an all cargo platform before surviving one wet Tuesday at a paper depot.
The customer relationship is also a retention engine. Monthly operations reviews can show loads planned, minutes saved, utilization, overrides, and avoided trips. That turns software from a clever tool into an operating habit. If the dashboard can't show money, it becomes a screen saver with axle loads.
For unit economics, a mature software customer could've eighty percent gross margin after onboarding, but early margins will be lower because support is hands on. I wouldn't fund a large sales team until three pilots show similar implementation effort and customers renew at the proposed price.
Key resources are a strong optimization engineer, a logistics domain lead, a product designer who understands warehouse stress, and a safety advisor. The key activity is validating plans in the field, not merely improving a mathematical objective function by a tiny percentage.
My verdict is invest, but tiny. The story is compelling when it's "safer, fuller roll loads in one depot," not "we reinvent freight." First next step: recruit one paper shipper with one hundred weekly loads and get access to six months of anonymized load records.
My verdict is wait for evidence. I'd fund a tightly priced pilot, not a platform. First next step: ask that customer to sign a paid letter of intent tied to measured outcomes. If they won't pay something, the savings may be more theoretical than operational.
My verdict is build a narrow prototype. First next step: ingest historical orders, model two trailer types, and compare the solver against actual planner decisions. No artificial intelligence magic show, no full integration, and no claim of safety autonomy.
Three votes, one cautiously moving truck. This can become valuable workflow software if it earns trust load by load and escapes the paper roll niche before it becomes a very elegant cul de sac.
And if the numbers work after implementation, not just in a spreadsheet where every trailer is full and nobody ever types the wrong roll diameter.
That's the whole game: physics, data, workflow, and a healthy respect for gravity.
Thanks for listening to Try Diligence. Load safely, measure everything, and never let a spreadsheet become the most experienced employee in the building.
A narrow, trust-first pilot can work—but only if measured savings outrun integration and liability costs.