The whiteboard in the dispatch office had its own weather system. Marker fumes, half-erased lanes, and a column of phone numbers so old that two of them belonged to drivers who had retired. This was mission control at a 60-person freight company in the Midwest, the kind of place that moves real things on real deadlines and runs the whole operation on a whiteboard, email attachments, and a woman named Marisol who had not taken a full lunch break since 2019.

This is the story of a 90-day engagement there: three builds, one near-disaster, and the numbers that got the contract renewed. Names and details are composite, the shape of it is real, and if you have ever wondered what an FDE engagement case study looks like when nobody is trying to sell you anything, pull up a chair.

The short version: ninety days embedded in a mid-size logistics operation typically produces two or three production tools, returns 40-plus hours of manual work per week, and cuts invoice lag from weeks to days. But the order of operations matters more than the builds, so let us start where every honest engagement starts. With the mess.

The Client and the Mess: Where Every FDE Engagement Case Study Starts

Sixty people, forty trucks, and a brokerage arm growing faster than the systems under it. Dispatch ran on the whiteboard plus a TMS from 2009 that everyone used as an expensive address book. Proof-of-delivery documents arrived as email attachments, sometimes as photos of paper taken in truck cabs at night, which is a lighting condition no scanner deserves.

Invoicing ran three weeks behind, and not because the bookkeeper was slow. Every invoice required a scavenger hunt: rate confirmation in one inbox, POD in another, detention notes in a text message from a driver named Big Andre. The owner, Dale, knew margin was leaking. He did not know where, which is the most expensive kind of leaking there is.

In theory, engagements like this follow five phases, and the anatomy of an FDE engagement maps them cleanly. In practice, the phases look like what follows.

Weeks 1-2: Shadow Everything, Ship Nothing (Almost)

The first two weeks of any embedded engagement feel unproductive to everyone watching. You ride along. You sit in the dispatch office from 5:45 a.m. You learn that Marisol keeps the real schedule in her head and the whiteboard is just the press release. You learn that Ray, the night dispatcher, maintains a private spreadsheet because he does not trust the day shift's data. Fair enough, Ray.

This discovery phase is where engagements are won or lost, and the first 30 days playbook exists because so many teams skip it. Skipping it means building for the org chart instead of the workflow, and the org chart never once answered the phone at 2 a.m.

The workflow map: what the whiteboard didn't show

By day eight, the map on your laptop shows 23 distinct steps between "load accepted" and "invoice sent," and exactly nine of them touch the TMS. The rest is email, texts, phone calls, and Marisol. The whiteboard, it turns out, was never the system. It was the user interface for a system that lived inside one person's skull, which is a single point of failure with a commute.

Then, day ten: the tiny script. Every afternoon, someone spent forty minutes renaming and filing POD attachments by hand. You write forty lines that watch the inbox, rename files to a standard, and drop them into the right folders. It saves maybe three hours a week. It also makes you a wizard, because that first small win is never really about the hours. It is the quick win that funds the rest of the program with credibility.

Weeks 3-6: The First Real Build

The first real build was dock scheduling. Inbound appointments lived in a shared inbox with a 24-hour response lag, carriers called to confirm times that had already changed, and the warehouse kept a clipboard backup for when the inbox lied. Which it did, daily.

Week three brought the scoping fight, and it lasted one memorable meeting. Dale wanted a full carrier portal. Marisol wanted the inbox to stop existing. Trey, the warehouse manager, wanted the clipboard to survive, because paper never crashes, a sentence he delivered with his whole chest. The compromise: one screen showing today's doors and appointments, a booking form carriers could use through a link, and yes, a print button for Trey's clipboard. Paper never crashes, and neither does goodwill.

Four weeks later it was live. Driver pushback lasted about nine days, ending the first time a carrier booked a 6 a.m. slot at 10 p.m. without calling anyone. The Friday demo that flipped the remaining skeptics was not slick. It was Marisol saying, unprompted, that she had finished her coffee while it was still hot. Adoption metrics come in many forms.

Weeks 7-9: The Near-Disaster

Now the honesty beat, because no real engagement is a highlight reel. Week seven, you launch invoice automation: POD matching, rate confirmation checks, draft invoices generated nightly. Week eight, it double-billed a major customer. Not a test customer. The customer.

The cause was mundane, as causes always are: a renamed lane had created two records that the matcher treated as one shipment. The recovery mattered more than the bug. Within two hours the automation was paused and the manual process restored. Within twenty-four, the customer's invoices were corrected through a phone call from Dale, not an email. Within forty-eight, the matcher had a duplicate-lane check, a dry-run mode, and a rule that no invoice leaves the building without a human glance for the next two weeks.

Something breaking mid-engagement is not the risk. The risk is breaking it and hiding. Owning the failure fast, in person, with the fix already running, bought more trust than six clean weeks would have. Nobody remembers the bug. Everyone remembers the phone call.

Weeks 10-13: Boring Wins and the Handoff

The last month is where an engagement quietly becomes a program. The POD capture pipeline graduated from the day-ten script into a proper flow: drivers photograph documents in an app, extraction runs overnight, and exceptions land in a review queue that takes the bookkeeper twenty minutes instead of the scavenger hunt. The Monday-numbers dashboard went up in the break room: loads, on-time percentage, invoice lag, detention dollars. It killed the meeting where people argued about whose numbers were right, because now everyone was staring at the same ones.

And then the exit, which is a deliverable, not an afterthought. Jess, the ops manager who would inherit everything, got trained on the review queue, the dashboard, and the restart procedure for the two scripts that actually matter. The handoff checklist covered the runbook, the credentials, the vendor logins, and the "if this number turns red, call this person" page. The goodbye was calm. Calm is the goal. If the goodbye is dramatic, you built a dependency, not a system.

The Scoreboard

Numbers, because Dale's renewal needed them. Hours of manual work returned per week: about 45 across dispatch, billing, and the warehouse, measured conservatively by timing the before and after. Invoice lag: three weeks down to three days, which did more for cash flow than any sales hire would have. Detention disputes dropped by half, because the PODs were suddenly findable, and findable is all most disputes ever needed.

Cost and return, stated plainly: the engagement ran about $55k all-in for the quarter. The annualized value of the time returned, the faster cash cycle, and the avoided billing errors landed north of $200k, computed with their own loaded-cost figures and their accountant's frown as the review board. That is why the renewal meeting took eleven minutes.

The last thing you pack is the laptop. The thing you actually leave behind is a Tuesday morning that runs without you. That is the whole job.