Run the pilot against your current freight process, not a perfect demo

Key Takeaways

A first freight trial means something only if it runs on the orders, carriers, rules and people your team uses today, compared with how that same work ran before.
A staged demo removes what creates coordinator work: incomplete orders, slow carriers and approval waits.
Lock four kinds of parity before the first load: the same order source, the same approved carriers and rate targets, the same approval limits and the same exception mix.
Use a current process vs. trial comparison sheet so every mismatch in conditions is visible before anyone reads the results.
Wilson, by Cartage, starts in assisted execution on a few lanes under the company's own approval rules. Cartage doesn't run pilots; the trial is the buyer's own test, and scope expands after the process is validated.

Why should a freight trial copy your current process instead of a clean demo?

A trial should answer one question: does freight move better on your real orders, carriers and rules than it does today? That works only when the trial and today's process run on the same inputs, so any difference comes from how the work was executed, not from easier conditions.

A vendor demo shows that a workflow exists: a complete order, a carrier that replies within seconds, no approval wait. When the buyer's own trial is set up the same way, it measures the vendor's preparation instead of the team's freight.

Consider a distributor with two plants. Plant A releases complete orders through the ERP feed. Plant B sends an emailed spreadsheet, often without a dock contact. A trial on Plant A orders alone will look excellent and say nothing about Plant B, where the coordinators actually spend their afternoons.

Four things decide whether a trial is honest: the inputs, the rules, the people and the exceptions. If any is cleaner in the trial than in daily work, the comparison is tilted. This differs from when to start the measurement clock, which is covered in measuring an AI logistics rollout once the real workflow runs. Timing decides when evidence counts; parity decides whether it is fair.

How to build the trial from your current process

Document the process as it runs, not as the SOP describes it. Sit with a coordinator for a handful of real loads and note where each order arrives, who quotes, which carrier gets the tender first, who confirms pickup and who chases the POD. Gaps between the SOP and real practice are the first findings.
Freeze the rules. Write down the approved carriers, tender order, rate targets, approval limits and who gets notified on each event. If the trial only works after a rule is loosened, record that as a result instead of quietly changing it.
Select loads by rule, not by hand. Scope the trial as every order on the chosen lanes released after an agreed date, including rush and amended orders. Nobody swaps a difficult load for an easier one.
Keep the same people in the loop. Whoever approves bookings today approves them in the trial, and the same people receive updates and invoices.
Record both columns for every load. One shows how comparable loads ran under the current process, the other what happened in the trial. Fill both from records.
Review with the coordinators at agreed checkpoints. They will spot the skipped step and the email sent on the side.

What same loads, same rules and same people look like in practice

Order intake. The trial should take orders through the path the team already has: an ERP feed, a scheduled file drop, a spreadsheet, email or an API. If the trial needs a cleaner feed than today's, that preparation belongs in the notes.

Carriers and rates. Only the carriers the company has approved, tendered in the usual order, against the same rate targets. An extra carrier makes the trial look faster for reasons unrelated to execution.

Approvals. If a manager signs off above a set amount today, the trial uses that limit too. The waiting time is real and should be measured.

Exceptions. Missed pickups, late trucks and orders changed after booking stay in the trial. Pulling them out is the most common way a trial turns into a second demo.

Communication. Carriers hear from the channels they use today, and internal updates land where the team already works, such as Slack or Microsoft Teams.

Evidence. Each trial load needs a record of messages, confirmations and changes, so anyone can rebuild the trial column without relying on memory.

Decision rules: when can you trust the trial result?

Check the conditions before reading the numbers.

Fewer exceptions than usual. If trial loads had fewer problems than the lanes normally have, keep running until real exceptions show up.
A rule changed mid-trial. Split the results before and after the change.
Shadow work. Ask coordinators whether they emailed, called or fixed data outside the system. Those touches count against the trial.
Removed loads. Record each load taken out of scope and why. Frequent removals mean the trial is being curated.
Baseline from memory. If the current process column came from estimates, treat the comparison as directional.
Partial success. A trial can pass on booking and fail on POD collection. Expand only the steps that matched or beat the current process, and keep the rest in assisted execution.

For what goes into the columns, the eight metrics to compare against a baseline are a good start.

A current process vs. trial comparison sheet

Fill in the current process column before the first trial load and the trial column as loads run. A flag means the result needs a note.

Condition	Current process (record today)	Trial (must match)	Mismatch?
Order source	ERP feed for Plant A, emailed spreadsheet for Plant B	Both sources, same formats	Flag if one plant is excluded
Order completeness	Share of orders missing ready time or dock contact	Same orders, not cleaned first	Flag if data was fixed before intake
Carriers	Approved list and tender order	Same list, same order	Flag any added carrier
Approval limits	Who approves, above which amount	Same person, same limit	Flag if limits were loosened
Order changes after booking	Count from recent loads	All changes kept in scope	Flag if amended orders were removed
Exceptions	Missed pickups, late deliveries, shortages	All exceptions kept in scope	Flag if problem loads were swapped
Evidence	Where the record of each load lives	Messages, confirmations and changes on the shipment	Flag if records must be rebuilt by hand

Illustrative worked example. Records show a typical Plant B order needs several emails and one call to confirm pickup. In the trial, the same kind of order arrives with the same missing dock contact. The trial column shows how the gap was handled, how long the booking waited for approval and whether the coordinator stepped in. What matters is that both columns describe the same order type under the same rules.

How Wilson, by Cartage, runs a trial on your current process

Wilson replaces the TMS for freight execution and connects to the company's ERP, so a buyer's trial can start from the order path the team already uses. Wilson works with SAP, NetSuite, Microsoft Dynamics 365, Oracle, Infor, Epicor, Sage, Acumatica, and other ERP environments, with the exact connection depending on the customer’s data flow and scope. Orders can also come from a scheduled file, spreadsheet or email.

Wilson tenders first to the carriers the company has approved, negotiates against the company's rate targets under configured rules and approval limits, and follows up by email, API or EDI where supported, and by phone where calls are configured. It escalates to the team when a decision falls outside the rules. Messages, confirmations and changes are recorded in shipment history, which is what fills the trial column of the comparison sheet.

Wilson starts in assisted execution. The company sets the approval boundaries, and the scope of execution can expand step by step as the team gains confidence in the workflow. Cartage doesn't run pilots; most customers start Wilson on a few lanes and expand from there. Teams configure it themselves or with the Cartage team, which helps map the first workflow during discovery. For the wider picture of the work an agent takes on, see what an AI logistics agent is.

Bring last week's messiest order from one of your lanes, with the carrier list and approval limits your team actually used, and see Wilson run that same order under those same rules.