Case Study · Project Ember

Getting the dates to hold.

How a custom window and door shop stopped paying for measurements that never made it to the floor.

Client Custom window & door manufacturer, Greater Toronto Area Work Order-to-install refresh Role Operations, process, communications

The results

ENGAGEMENT WRAPPED

What it did.

Directional results over the two quarters that followed, reported by the owner rather than pulled from a system.

4% of revenue going to callbacks and rework, down from about 9%.
<15% Late jobs

Down from about a third, and the slips that remain are days rather than weeks.

$140k The job that triggered it

A multi-unit order where a revised measurement never reached the shop, costing about $22,000 to put right.

2 Repeat orders

From the builder who nearly walked after that job, placed in the following year.

01

The context

A small shop doing work nobody else will take.

A 12,000 square foot shop in the Greater Toronto Area building custom aluminum and wood-clad windows and doors, the heritage sizes and odd openings that standard product cannot cover.

Around 70% is residential replacement for high-end renovations, the rest is small-batch supply to boutique builders on four to twelve unit projects. They install most of the residential work themselves and hand builder work to the builder’s own crews. Fourteen people, roughly $3.2M a year, and an owner who knew every job by heart.

Custom windows & doors Greater Toronto Area 14 people ~$3.2M revenue
02

What was breaking

Three things, over and over.

1 job in 3

Dates that moved

About one job in three shipped or installed late. The average slip was eleven days, and a few complex heritage jobs slid six weeks or more.

9% of revenue

Work done twice

Rework and warranty callbacks ran near 9% of revenue. Most of it traced back to field measurements that were wrong, and nobody caught them before the glass was cut.

No single source

Two versions of the truth

Estimators quoted off the first site visit. Final measures lived in notebooks and email, so shop packets sometimes carried dimensions that had already been revised.

None of this was carelessness. Rush jobs jumped the queue without anything else being rescheduled, glass got ordered late, and a remeasure could be perfectly correct and still never reach the person cutting to it.

03

Why they called

A $140,000 job that arrived the wrong size.

Six units showed up on site undersized. A revised measurement from a second site visit had never made it out of the estimator’s notes and into the shop order.

The install crew was already booked and standing on site. The builder was furious. The owner absorbed about $22,000 in rush glass charges, overtime to re-fabricate, and a discount to keep the relationship alive.

I’m the only one who knows where anything actually stands, and I found out about this the same day the crew did. I can’t keep running the business like this.

That call was not about buying software. It was about making this stop, and about the owner no longer being the only system in the building.

04

The approach

Ten weeks, working the flow.

1

Map

Walk the whole path from first site visit to warranty walk-through and find where information falls out.

2

Gate

Put one hard checkpoint in front of the most expensive mistake, cutting glass to the wrong number.

3

Show

Make the status of every job visible on the floor so nobody has to ask the owner.

4

Assign

Name who owns each decision so the checks stop landing on whoever happens to notice.

05

What I built

Six pieces, none of them fancy.

Final Spec Sign-Off A one-page sheet tied to every shop packet, confirming final measurements, product type and hardware, signed by the estimator and the shop foreman. No glass order leaves without it.
A production board on the floor A magnetic board running Measured, Confirmed, Ordered, Cutting, Glazing, Ready, Install Scheduled, Complete. Colour-coded for residential against builder work, so bottlenecks are visible at a glance.
The order-to-install flow, written down Every step from site visit through warranty walk-through documented, with the handoff between field measure and shop release called out as the weak point it was.
Clear ownership The estimator owns the measurement, the foreman owns the final spec, admin confirms the date with the client before anything is ordered. No more assuming somebody else checked.
One quoting and spec template No new tool. Their existing spreadsheet, rationalised so estimators and the shop read the same fields off the same version, with a revision log so changes are visible.
A customer update cadence Three standard touchpoints, order confirmed, in fabrication, install date locked. Fewer “where are my windows” calls and no last-minute surprises.
06

The insight

The fix was a signature, not software.

The most expensive problem in the shop was solved by one sheet of paper that two people had to sign before glass got cut. Everything else, the board, the roles, the update cadence, existed to keep that one checkpoint honest. Small shops rarely need a new system. They need the handoffs to stop leaking.

07

The outcome

ENGAGEMENT WRAPPED

Dates that mean something.

Directional results over the two quarters that followed.

  • Callbacks and rework down from roughly 9% of revenue to about 4%.
  • Late jobs down from about a third to under 15%, and the slips that remain are days rather than weeks.
  • The owner checks the board twice a week instead of chasing status on the floor every day.
  • The builder who nearly walked after the $140,000 job placed two more orders in the following year.

The real win was not the percentages. It was the owner getting to run the business instead of living inside its breakdowns.

Let’s talk

Paying for the same mistake twice?

I help small shops and owner-led teams find the handoff that keeps leaking, and put something simple in its place.