Case study — solar, Texas
The lead engine that books solar consults on autopilot.
A Texas solar company needed consultations, not click reports. We run the whole engine — the ads, the funnel, and the AI agents that verify, engage, and book every lead. Four problems stood between their ad spend and their calendar. Here is how each one fell.
- of leads book a consult
- 50%of leads book a consult
- appointment show rate
- 94%appointment show rate
- ad spend per booked consult
- ≈$96ad spend per booked consult
1Flying blind
The problem — You can't fix a funnel you can't see. The ad platform reported clicks, the CRM showed leads, and nothing explained the gap between them. Which step lost people? Why? Nobody could say — so every “optimization” would have been a guess.
What we did — Before changing anything, we built an analytics dashboard and started logging everything: every page view, every funnel step, every drop-off and disqualification reason, tied back to ad spend. The first thing the data exposed: nearly half of paid clicks never even landed — one April week bought 223 clicks and only 116 landing views. People clicked the ad and gave up before the page finished loading. So the first fix wasn't copy or targeting. It was a page-speed rebuild: first paint cut from 5.5s to 0.8s, Lighthouse score from 55 to 80.
First paint after the rebuild. The click-to-land drop-off is gone — landings now track ad clicks one-for-one instead of losing half. Every fix below was found the same way: measure first, then solve the problem the data actually shows.
2Fake numbers
The problem — Meta delivered volume, but the sales reps kept coming back with the same report: “the number doesn't work”, “wrong number”. We measured it before fixing it — across every lead number from a 14-day window, roughly 1 in 5 wasn't a real mobile phone: invalid numbers, landlines, and burner VoIP numbers. Five of them were literally Google Voice.
What we did — A simple two-layer phone gate, inline on the contact form: a carrier line-type screen that rejects burners and landlines before a single SMS is spent, then a 6-digit SMS code the lead types back. A number that never verifies never becomes a lead — not in the CRM, not in an admin text, nowhere.
Fake numbers reaching the sales team. Every lead is now a verified, reachable human — the wrong-number call reports stopped, and the ad account optimizes on real people instead of fake form fills.
3Cold leads
The problem — Solar leads decay by the hour. Leads waited hours — sometimes days — for a callback, and by then they'd filled out three competitors' forms. In April, only 20% of captured leads ever became a booked consultation.
What we did — An AI texting agent that opens the conversation within about two minutes of the form submit — it answers first questions, coordinates a time, and books the consultation straight into the sales calendar, around the clock. No lead sits in an inbox waiting for the office to open.
The lead-to-consultation rate roughly doubled in the months after the agent launched, and has held near 50% even after the phone gate made every lead count. That's 367 consultations booked to date — about 2 a day, on autopilot.
4No-shows
The problem — People book in good faith, then life happens — plans change, work runs late, and a sales rep burns an hour on someone who was never going to pick up. In this industry, no-show rates of 30–50% are normal.
What we did — A confirmation agent that reminds every prospect before their appointment — and when something has changed, reschedules them right there in the chat. The rep's slot gets freed and rebooked instead of silently wasted.
Of 487 appointments on the calendar, only 28 ended as no-shows — fewer than 1 in 15. The reminder alone is often enough: people show up to appointments they were reminded of.
One system, from ad click to sat appointment
Over 12,000 funnel visitors. 766 leads captured — every recent one SMS-verified. 367 consultations booked, averaging under $100 in ad spend per booked consult across the campaign. The funnel, the texting agent, the confirmation agent, and the tracking all run as one system — which is why each piece makes the others better: verified leads make the ads smarter, instant follow-up makes the leads warmer, and reminders make the appointments real.
Have the same problems?
Fake numbers, cold leads, and no-shows are fixable. We'll build the engine that fixes them.