Find margin leaks before they repeat

The problem

A single bad job is visible. A leak isn't: it's the same customer's engineering changes, the same part family's underestimated cycle time, or the same vendor's price creep, spread across dozens of jobs at a few points each. Averages hide it until the quarter's margin comes in low.

What to measure

Leak = the same driver across multiple jobs. Measure total erosion attributed to each customer, part family, work center, and cost category over the period, then compare to the previous period to see if fixes worked.

How MarginGuard helps

MarginGuard groups erosion by customer, part family, and category automatically, ranks the groups in dollars, and alerts when newly imported jobs match a pattern. Each fix is tracked as potential, implemented, and verified.

Next step: sign up, load the demo, or compare plans.

How to get started

1. Step 1
   Import a quarter of jobs

   Enough history for patterns to show.

2. Step 2
   Read the rollups

   Customer, part family, and category views ranked by erosion dollars.

3. Step 3
   Fix and verify

   Log each fix; confirm on the next imports that the leak shrank.

The four most common leaks

Customer leaks (scope creep, rush orders, tight tolerances not priced in). Part-family leaks (estimate standards that are consistently too low). Work-center leaks (process or tooling problems). Material leaks (price increases after quote, yield loss).

How to tell a leak from a bad week

One job running 30% over is an incident: investigate it, but don't change standards because of it. A leak has three signs: the same driver (customer, part family, work center, or category) appears on several jobs; the erosion is in the same direction each time; and it persists across more than one import. When all three are true, the fix belongs in the estimate standard, the price, or the process, not in a conversation about one job.

Fixing leaks without guessing

Each leak gets one owner and one change: raise the setup standard for a part family, re-price a customer's rush work, or re-quote material with current vendor prices. Then check the next jobs that share the driver. If erosion dropped, the fix is verified; if not, the driver was something else. Tracking fixes this way is what stops the same leak from being rediscovered every quarter.