Origin storyHow I got here
I watched it happen gradually, then all at once. A local HVAC contractor in Denver — 15 years in business, 400-plus Google reviews, Better Business Bureau accredited — gets asked about by a homeowner on ChatGPT. The answer comes back with three national brands and a franchise directory. The local company, the one the homeowner would actually prefer if they knew it existed, is nowhere. The AI did not lie. It just pulled from a different signal stack than the one the local business had built its reputation on.
That bothered me. But what bothered me more was what I found when I went looking for a product to actually fix it. Every AEO tool on the market was shaped the same way: ping the API, collect a response, call it your "AI visibility score," hand you a list of pages to rewrite. Every agency was shaped the same way too: take a retainer, produce a content calendar, send a quarterly strategy deck. Different costumes, same pattern — someone watches, then hands the real work back to you. None of it was actually going to make ChatGPT cite the Denver HVAC contractor, because none of it was operating on the signal layer the AI model was ranking on.
Here is the thing nobody was saying out loud: ChatGPT does not rank on your website. It ranks on what real users on real accounts on real devices are actually doing. Real-account session activity, at scale, across geographies. A better blog post does not produce that signal. A cleaner schema tag does not produce that signal. An API polling tool does not produce that signal either — it only watches the result. Producing the signal itself is an infrastructure problem: you need thousands of real phones and laptops on residential ISPs, signed into real accounts, running real sessions against ChatGPT, Perplexity, and Gemini. Nobody was building that, because it is hard.
I tested the anti-simulation argument directly first — ran the same business against synthetic tracking tools and against real devices with real accounts in real metros. The gap was not a rounding error. One test caught a 3-versus-123 mention discrepancy on the same brand in the same week. That confirmed what I already suspected: measuring synthetic responses and handing the customer a to-do list produces neither accurate data nor actual rankings. So I built the device farm instead. 10,000+ real phones and laptops, the only infrastructure of its kind in the market.
"If you want AI to rank your brand, you have to run the signal layer AI ranks on. That means real phones, real residential ISPs, real signed-in accounts — at scale. Not an API query. Not a dashboard. Not a content calendar. The farm is the product."
So we built it. 10,000+ consumer phones and laptops, on residential connections, signed into real ChatGPT, Perplexity, and Gemini accounts, distributed across US metros. Once the farm existed, the product shape followed automatically: if we can produce the signal, the customer should not have to do anything. Name the keyword. We run the ranking. Hands-off. No content rewrites, no schema edits, no website access. You keep running your business; we handle the signal layer.
Twelve named businesses later, the results have been consistent: 340% average citation growth, 100% of clients hit their first verified citation within 14 days. We cover five AI engines. We do not round up. We do not declare success when the needle did not move. SignalAEO is a category of one — the only hands-off AI ranking service in the market — because nobody else built the infrastructure that makes it possible. I also run SEO Local, a local-SEO execution firm for contractors. Same execution-first philosophy, different product, independent team.
We set out to build a ranking outcome for businesses that could not afford to keep buying dashboards and deliverables that did not move the needle. The rest followed from that.
— Russ Thornton, Founder