How do you find revenue you already missed on jobs you have completed?

The cheapest demand you will ever find

Every completed job leaves a record of what the technician saw. Some of what they saw was work you were never asked to do and never offered to do. That backlog is qualified in a way no advertising can match: you know the address, the equipment, its age, and that the customer already let you in the door.

Nobody mines it because it lives as prose. Job notes, technician comments and estimate descriptions are unstructured text, and unstructured text is invisible to every standard operational report.

Four patterns worth extracting first

• Aging equipment noted, no estimate created. The single largest pool in most companies, and the easiest to act on with a seasonal outreach.
• Options presented and declined. Recorded on the estimate, then never revisited. Many declines were timing, not refusal.
• Repeat repairs on the same system. Two or three repairs in a year is a replacement conversation the customer would probably welcome.
• Documented safety or code findings. Follow up because it matters, and because it is one of the few outreach reasons that is unambiguously in the customer's interest.

Turning notes into a work list

This is a straightforward language task applied to your own records: read every job note and estimate description, extract equipment, age, condition and any recommendation, and match it against what was actually quoted and sold. What comes back is a ranked list of addresses with a specific reason to call.

That extraction is the kind of thing operational AI is genuinely good at, because the work is high volume, repetitive and impossible to staff. It also has to be reviewed before it drives outreach, since notes contain shorthand, misspelled model numbers and technician abbreviations that need interpretation.

Handle it carefully

Two cautions. First, the list will contain findings that were discussed and declined for good reasons, and calling those customers as though nothing was said damages trust. Review before outreach, and give the technician who wrote the note a chance to weigh in.

Second, resist the urge to blast the entire list at once. Sequence it by seasonal relevance and equipment age, and give the outreach to people who can have a real conversation. Joined against membership status and service history in a customer intelligence view, it becomes a steady source of work rather than a one-time campaign that burns the list.