SW Solar — How a residential solar brand turned $5K of ad spend into $1.53M of attributed revenue by warming the market with AI citations first.
- $1.53M revenue
- $5K spend
- Solar DTC
- 90 days
Residential solar is the textbook hostile category: high ticket, long consideration window, sales cycles measured in months, and a buyer who Googles a dozen variations of 'is solar worth it' before they ever click an ad. Cold paid traffic into this market burns. The brand had a solid offer and a credible installer footprint, but their cost-per-acquired-customer through cold paid was making the unit economics impossible.
What made this worse: in the target metros, ChatGPT, Perplexity and Gemini were defaulting to three larger national installers when prospects asked 'best residential solar in [state]'. SW Solar wasn't on page one of the AI answer for any of the high-intent queries. Every prospect who used an AI tool to research before buying was being pre-sold on a competitor.
The brief was simple, the math was not: get a meaningful share of those AI recommendations, and prove paid spend could profitably harvest the warmed demand within a quarter.
- 01
Audited citation share across ChatGPT, Perplexity and Gemini for the top 80 solar-buying queries in target metros. Default rank: not on page one for any of them.
- 02
Built per-platform signal stacks targeting 'best residential solar [state]', 'solar tax credit 2026' and 'is solar worth it [region]' query clusters. Different platforms reward different signals — we stopped treating GEO as a single channel.
- 03
Engineered entity coherence: site copy, schema markup, and third-party content (review sites, regional press, knowledge bases) were rebuilt to render the same brand profile no matter where an AI model looked.
- 04
Layered branded-search paid on top to capture intent the moment AI-warmed buyers Googled the company name. The branded layer is where the $5K went — not cold prospecting.
- 05
Weekly delta reports tied creative testing themes back to citation themes — every paid ad echoed the language the AI models had already pre-sold the prospect on.
- 06
By week 12: citation share dominant in target metros and $5K of paid spend converting at the ROAS validated by our attribution model. Exact internal team allocationsPending client approval
We don't publish what we can't verify with the client. This will fill in when they sign off.
When the buyer journey is long and emotionally loaded, paid can't carry the cold introduction profitably — the math will not work. The play is to use GEO to make the brand the answer the AI assistants are already recommending, then let paid harvest the branded intent that follows. Cold paid is replaced with warm paid. The unit economics fix themselves because every click is from someone who has already been told you are the right choice by a tool they trust.
Want this for your category?
30-minute audit. We tell you what's possible in 90 days for your brand — or the honest answer that this isn't your lever.
Book a free call