How AI Search Changes Commercial Solar Buyer Research

Why are commercial solar buyers asking AI assistants instead of just searching?

The short answer is that their questions are usually too specific for a generic search box to handle well. Commercial solar buyers rarely type "solar company Arizona" and call it research. What they're actually asking is shaped by a real project already in front of them:

• Who installs commercial solar for warehouses or manufacturing facilities?
• Which companies offer EPC (engineering, procurement, construction) services for solar?
• Who has experience with solar carports or large-scale commercial installations?
• Which companies serve the Phoenix metro area specifically?

None of these are hypothetical. They reflect how a facilities manager, business owner, or project lead actually approaches a solar decision, with a building type, a budget scale, or a service need already in mind. For that kind of specific question, an AI assistant that can hold a back-and-forth conversation is often a faster starting point than a search engine that hands back ten links and expects the buyer to do the filtering themselves.

What did KMS Signal find testing 20 buyer questions in Gemini?

To see how this plays out in practice, we ran 20 buyer-intent prompts through Gemini, built around real commercial solar scenarios in Arizona: installation, warehouses, manufacturing, EPC services, carports, and larger commercial projects. Some companies came up again and again across those prompts. Others didn't appear at all, even when their services matched the question closely.

That pattern is the whole finding, and it's worth being precise about what it does and doesn't mean. We don't know exactly why Gemini surfaced some companies and not others, since that mechanism isn't published anywhere, by Google or by anyone else. What we can say is that the repeated appearances weren't random. The companies that came up often shared a few things in common, which is where the research gets more useful.

What does it take for a company to show up in these answers?

Based on what we observed, five characteristics separated the companies that appeared often from the ones that didn't:

None of this is a checklist Gemini publishes anywhere. It's simply the pattern we noticed in the answers it gave, and like any pattern drawn from a small sample, it comes with limits worth stating clearly.

What does this research actually prove, and what doesn't it prove?

It shows what we observed in 20 specific prompts on one AI system at one point in time. It doesn't show how Gemini's recommendation system works internally, and it doesn't prove that any one of the five characteristics above directly causes a company to be mentioned more often. Two things happening together isn't the same as one causing the other, and we're not going to pretend otherwise.

What we're confident in is narrower and more useful because of it: companies with clearer, more specific, better-evidenced information about their commercial solar work showed up more consistently than companies without it. That's a pattern worth acting on even without knowing the exact mechanism behind it, and it turns out this pattern isn't unique to Arizona or to solar.

Is this an Arizona pattern, or part of a bigger shift?

It's both, and the two are worth separating. The Arizona findings above are specific to solar and specific to Gemini. But the underlying behavior, buyers turning to AI tools earlier in their research, isn't limited to this industry at all.

A 2026 survey of more than a thousand B2B software buyers by G2 found that roughly half now start their research inside an AI chatbot more often than a traditional search engine, a sharp increase from about a year earlier. Forrester separately found that the large majority of B2B decision-makers used a large language model somewhere in their 2025 purchase process. And in G2's survey, most buyers who used an AI chatbot during their research ended up choosing a different vendor than the one they'd originally planned to.

That data describes B2B software buyers, not commercial solar buyers, so it would be a mistake to treat it as proof of anything specific to this industry. What it does show is that the behavior behind our Arizona findings, using AI tools as part of vendor research and sometimes changing a decision because of what one surfaces, is a documented pattern elsewhere in B2B buying. It's not something we're speculating into existence. Which raises the practical question: what should a company actually do with that?

What should a commercial solar company actually do about this?

Start with the same five characteristics the research pointed to, and be honest about which ones are already covered. A company doesn't need to overhaul its entire web presence. It needs clear, specific, commercial-focused information, documented project evidence with real details, explicit service areas, whatever genuinely sets it apart from competitors, and a presence beyond its own website that confirms the same facts.

That last one, external corroboration, is usually the most neglected and, based on what we saw, the one with the most room to improve. A company's own website can say anything about itself. What gets referenced elsewhere is harder to fake, which may be exactly why it matters more.