Construction enquiries rarely fail in one dramatic moment. They disappear between the inbox, site visit, estimating file, supplier response and the next customer call. A follow-up system makes those hand-offs visible.
The goal is not to let AI price a project. It is to ensure that evidence, ownership and the next action survive every stage.
A practical stage model
Use a small number of observable stages: new enquiry, qualification review, site visit required, estimating, internal approval, quote sent, follow-up due, won, lost or not a fit. Define the event that moves a record into and out of each stage.
Every open record needs one named owner and one next-action date. Shared ownership usually means no ownership; a general note such as 'waiting' is not a next action.
Where automation helps
Automation can capture e-mail and web enquiries, extract declared location and work type, request missing documents, create estimating tasks, remind the owner after a set interval and draft a follow-up message for approval.
Pricing, technical feasibility, capacity and bid decisions remain with qualified people. Uncertain classification and high-value opportunities should be routed to review rather than acted on silently.
How to measure the first month
Measure capture completeness, records without an owner, overdue next actions, time spent in each stage and exceptions that required manual recovery. Compare the CRM against source inboxes to find missing work.
Do not infer revenue or win-rate improvement from a working automation alone. Establish the operational baseline first; commercial comparisons become meaningful only after the process has enough consistent history.