Installation scheduling from 17 hours a week to under two

What scheduling by hand was costing

Magnolia Doors builds custom iron and aluminum doors, gates, railings and glass for high-end homes and builders around San Antonio, and its own crews install everything it makes. Getting a crew to a house is the bottleneck of the whole business.

A request arrived from sales as a PDF. Someone then worked out whether the paperwork was complete, whether the materials were ready in the shop, who was already booked that day and where, what the job actually required, and what arrival window to promise the client. Five questions, answered in five different places, and no single screen that settled any of them.

At 30 to 40 installations a week, that came to 13 to 17 hours every week spent getting jobs onto a calendar. Somewhere between a third and half of a full-time position, spent gathering information the business already had.

What changed

Scheduling an installation now takes about three minutes, against 23 to 35 by hand. The weekly load drops from 13 to 17 hours to roughly one and a half to two.

15 to 19 administrative hours come back every week, about 3 to 4 hours per business day, without adding a person.

A full day of installations is approved in about a minute. Up to 20 jobs travel in a single proposal and are created together, where each one previously moved through its own sequence of checks. A whole week now fits in two or three proposals.

28.5 fewer miles in a measured week, from sequencing each day properly rather than taking the nearest job first.

Jobs reach the calendar already checked. Paperwork, materials, address and crew availability are verified before any window is proposed, so a problem surfaces while it is still a scheduling question rather than after a crew has driven out.

Nothing reaches the ERP without a person approving it. Speed never came at the cost of control over the calendar.

How Claude helps

Claude carries the request from arrival to a proposed booking, and stops there.

It reads the request and confirms the order in the system of record, checks that the materials are actually ready, reads the calendar for who is available, works out the plan and the arrival window, groups the day geographically, and presents the proposed schedule for review. The coordinator approves, and only then do records change. Each Wednesday Claude produces the weekly operations summary in Spanish, from Magnolia's own format.

The division of labour is the part worth copying. Fixed business rules handle everything that has to be exact, so those decisions come out identical every run. Claude handles the judgment rules cannot:

• Reading what a job actually involves from the written scope of the request, rather than relying on form checkboxes alone.
• Explaining what a job is waiting on, and who owns it, so nothing drops quietly out of the plan and every priority arrives with the reason behind it.
• Showing its reasoning, so the coordinator can see what a recommendation was based on and weigh it rather than take it on faith.
• Reconciling records that disagree across the systems the business runs on.

And the one call that stays with a person by design: confirming a job with the client is a claim about a conversation, so it belongs to whoever had it.

Keeping the coordinator in control

Every booking is proposed before it is made. The coordinator sees exactly what would change, field by field, current value against new one, and approves it. The planning stages have no ability to write to the ERP at all, and the step that does is confined to the installation calendar and nothing else in the business.

That is what makes the speed usable. A schedule that lands in one approval is only an improvement if the person approving it can see what they are agreeing to.

About these numbers

The volumes, the batch behaviour, the mileage and the verification results are measured against Magnolia's production system. The per-event minutes are a reasoned estimate, built from the work the process demonstrably requires rather than from timing a person doing it, and the weekly hours follow from them. A short before-and-after timing study would turn that estimate into Magnolia's own hard number, and it is worth doing.

Business impact

Magnolia Doors turned up to 17 hours of weekly scheduling into a workflow that handles each installation in about three minutes, and got back the equivalent of a third to half a position without hiring. The team still works inside the ERP it already had, rather than maintaining a second scheduling system alongside it.

The value is not only the hours. Jobs now reach the calendar with their materials confirmed, their address resolved, their day sequenced and their exceptions named, and the coordinator approves each one. That combination is what makes the same approach worth extending to the next process.