For two years, most businesses have used AI in marketing as a copy machine: it writes the subject line, drafts the post, makes the picture. That saves time. But this week the first published customer results arrived for a different kind of system — one that decides, customer by customer, what to say, when to say it, and when to stay quiet. The measured gains didn't come from making more content. They came from making better decisions. And the audience where those decisions pay off fastest isn't strangers. It's the customers you already have — the ones who went quiet.

The Shift

A traditional campaign is a schedule: built Monday, sent Thursday, the same message to everyone. A decisioning system is a loop. A customer opens a message, visits a pricing page, or goes quiet for two weeks — each of those is a signal. The system scores which next step is most likely to move that specific person, sends only that step, and learns from the result. No two people walk the same path. That design is built for the asset most owners overlook: the dormant list. Past customers who went quiet and quotes that never got an answer are revenue you've already paid to earn, sitting idle in your system. Reactivating them costs a fraction of acquiring a stranger — if the outreach is well-timed, personal, and legal.

The Law

Pennsylvania's Act 47 takes effect Sunday, October 18, 2026, and it's among the toughest outreach laws in the country. Marketing calls and texts are allowed only between 9 AM and 7 PM, Monday through Saturday — nothing on Sundays or legal holidays. A text counts exactly like a call, and so does a voicemail, including ringless voicemail. Automated and AI-voice marketing requires the person's prior express written consent that you can produce. And opting out is broader than STOP: QUIT, END, REVOKE, OPT OUT, CANCEL, and UNSUBSCRIBE all count — a system that only listens for STOP is not ready. Penalties run $1,000 per violation, $3,000 when the person reached is 60 or older. The wider lesson: states are writing their own rulebooks now, with different hours and consent language, so any follow-up system that crosses state lines has to know which rules apply before it reaches out.

Who's Doing It

The decisioning results stand out because each ran against a control group. A reactivation campaign aimed at quiet customers produced 137% more opens, 207% more clicks, and nearly 14% more purchases per thousand people targeted — same audience, same offers, different decisions. An onboarding journey produced more than 1,200 extra first-time purchases and a 42% higher conversion rate, with converters worth about three times the average. A churn-prevention campaign hit a 10% purchase rate against just over 1% in the control. Meanwhile, small specialists are getting found in a new channel: three of every four AI-referred sales involve specialized products outside the biggest hundred categories, because shoppers describe exactly what they need and the assistant matches them — but only to businesses whose information is clear enough to understand and recommend.

Your Move

Three steps. First, pull your dormant asset: past customers and unanswered quotes from the last twelve months, and count them. Second, before any automated outreach touches that list, run the compliance check — written consent on file, hours set by state, every opt-out word honored. Third, test honestly: reach part of the list, hold part back, and measure the difference in real revenue. That's the system we build at The Agency — a decision-driven follow-up engine for the customers you've already earned, with consent and hours handled in every state you serve, and pages clear enough for AI assistants to recommend you. Listen to the full episode: Scale with AI — October 17, 2026. Then take the Free AI Audit and see where AI can actually move your revenue.