Query for absence, not for activity.
Four questions, answerable from systems most companies already have. Each one looks for a record with a missing successor rather than for something that happened.
Estimates sent in the last ninety days with no won or lost outcome recorded. In most service businesses this is the largest single number and the easiest to act on, because these are people who asked you for a price.
Inbound calls with no outbound call to the same number afterward. This requires call logs, which is why businesses without call tracking usually cannot see this at all.
Jobs marked complete with no corresponding invoice. This is where the gap between a field system and an accounting system shows up as actual money.
Customers with a regular pattern of work that stopped. Nobody churns loudly in a service business; they just stop calling.
Quantify before fixing.
Each gap gets a number attached — count and approximate value — before anything is built. That turns a debate about priorities into arithmetic, and it usually reorders the list. The gap people assume is biggest is frequently not.
Then close it automatically.
The reason these gaps exist is that closing them depends on a person noticing, and people notice when they are not busy — which is precisely when the gaps are smallest. Follow-up that runs off the record rather than off a list will keep working during the weeks that produced the problem in the first place.
Build it connected to the record, not as a broadcast sequence. Follow-up that ignores the fact that a customer already replied is the version people find annoying, and it is a systems failure rather than a cadence one.