Review Machine
Two live steps, told as a story: a job wraps up, the customer gets asked, and the customer answers. Happy customers get walked to the Google review link. Unhappy ones get caught privately - before it becomes a one-star review.
Mark a job done
In real life this fires automatically when the invoice closes or the tech marks the job complete. The agent then reaches out to the customer while they're still happy. For the demo, the "customer" is a demo inbox so you can watch it work.
The ask is on its way.
The agent took it from here: a friendly, personal-sounding message in the owner's voice is going out to the customer right now, timed while the good experience is still fresh. No awkward asking, no forgetting.
Could not reach the agent.
Connection hiccup - try again in a few seconds.
Play the customer
Now pretend you're the customer replying to that message. Try it happy, then try it angry - the agent routes them completely differently. Happy goes straight to the Google review link. Angry gets intercepted into a private conversation, with a make-it-right reply drafted for the owner.
Routed.
The agent read the reply and did the right thing: a happy customer is being walked straight to the Google review link; an unhappy one just triggered a private escalation to the owner - with a drafted recovery reply ready for one-tap approval. The problem gets fixed instead of published.
Could not reach the agent.
Connection hiccup - try again in a few seconds.
This runs after every single job
Moving from 3.9 to 4.6 stars changes how many people call. Reviews are the number-one local ranking lever an owner actually controls - and the review count itself is the monthly report. When a new review lands, good or bad, the agent also drafts the owner's public reply in their own voice, waiting for one-tap approval.