Change Orders by Job Type: Catching the Ones You Forgot to Write
· By Mike Hagberg
Your history knows how many change orders a job of each type should have by each stage — when an active job falls below its own trend, you're probably doing changed work for free.
Change Orders by Job Type: Catching the Ones You Forgot to Write
The most expensive change order is the one that never gets written. The work happens, the cost posts, the contract doesn't move — and the margin quietly absorbs it. By close-out it's indistinguishable from a bad estimate.
Here's the insight: your own history already knows what normal looks like.
Every job type has a change-order signature
Pull your completed jobs from Sage 100 Contractor and group by job type. Patterns emerge fast:
- Remodels: maybe 90% of them carry change orders, averaging 8–12% of contract value, with the first CO typically landing around 25–30% of cost burn.
- Ground-up new construction: fewer COs, later, smaller.
- Service-adjacent work: almost none.
That signature — CO frequency, CO % of contract, and when in the job COs start appearing — is a baseline you can hold every active job against.
The alert that follows
If remodels historically carry ~10% in change orders with the first one by 30% completion, and an active remodel is at 45% cost burn with zero approved COs, one of three things is true:
- It's the rare clean job (possible),
- COs are sitting unapproved in someone's truck (recoverable — today),
- Changed work is being performed for free (margin already gone — recoverable only if you catch it now).
All three deserve a five-minute conversation with the PM this week, not a post-mortem at close-out. The same logic runs in reverse: a job whose CO% has blown far past its type's norm signals scope chaos worth a different conversation.
Small print that matters
Baselines need history — DataXcel only trends a job type once at least five completed jobs back the number, so a one-off unusual job can't manufacture false alerts.
How DataXcel watches this for you
This is the new trend engine in the DataXcel CEO Briefing: completed jobs build per-type baselines (CO frequency, CO % of contract, first-CO timing, final margin), and every active job is compared against its type's trend each week. A type-X job past its trigger point with materially fewer COs than the trend gets a Project Management alert naming the job, the gap, and the supervisor — and the Job Profit-Fade & Trend dashboard shows the curve behind it.
Your completed jobs already wrote the playbook. The trend engine just reads it back — one job, one week, before the margin leaves the building.