All articles
Digital Marketing Strategy

How ChatGPT, Perplexity, and Google's AI Overview Pick Different Architects, Engineers, and Consulting Firms for the Same Search

ChatGPT, Perplexity, and Google's AI Overview do not read your firm's site the same way. Here is what each one checks before naming an architect, engineer, or consulting firm, and the one page format that works across all three.

4 min readUpdated 2026

Someone asks ChatGPT to find a structural engineer for a steel frame extension in Manchester, types the same brief into Perplexity, and searches commercial fit out architect near me on Google. Three different engines, three different shortlists, and most architecture, engineering, and consulting firms build one portfolio page and assume it covers all three.

AI Search Is Not One Search Engine

ZENETERA sees firms treat AI search as a single target, when ChatGPT, Perplexity, and Google's AI Overview do not read a site the same way or pull from the same sources. A firm that shows up in one engine's answer is often invisible in the other two. Here is what each engine actually checks before it names an architecture, engineering, or consulting firm, and the one page format that works across all three.

1. What ZENETERA Found Across 60 Firm Websites

ZENETERA ran the same set of real client briefs, such as structural engineer for a loft conversion and commercial interior fit out consultant, through ChatGPT, Perplexity, and Google's AI Overview against 60 architecture, engineering, and consulting firm websites across the markets we work in. A firm named by one engine was named by a second engine for the same brief only 29 percent of the time.

29%of firms named by one AI engine were also named by a second engine for the same client brief

2. How ChatGPT Chooses

ChatGPT increasingly browses live pages rather than relying only on training data, and it favors project pages that state a specific project type, a specific sector, and a specific scale in plain text near the top of the page. A firm that lists structural engineering as one line inside a long services list rarely gets named. A firm with a dedicated page titled structural engineering for residential extensions, naming typical project sizes and locations, gets quoted directly.

3. How Perplexity Chooses

Perplexity leans harder on citations it can show the user, so it favors pages that read like an answer rather than a brochure. A named project written as a full sentence, such as we delivered a 40 unit residential conversion in a listed building in Bristol, gets pulled into an answer with a working link attached. A portfolio page that is only a photo grid gives Perplexity nothing worth citing.

4. How Google's AI Overview Chooses

Google's AI Overview draws from the same index as ordinary search results, so a firm that is not already ranking for a query rarely appears in the AI generated answer either. Structured data matters most here. Firms that mark up their services, sectors, and accreditations as schema so Google can read them as facts, not just paragraphs, are named noticeably more often than firms with the same information written only in prose.

EngineWhat it reads firstWhat actually gets you named
ChatGPTLive page content it can browse in the momentA specific project type, sector, and scale stated in plain text near the top of the page
PerplexityCitable, answer style contentA named project written as a direct answer, not a photo grid
Google AI OverviewThe same index as normal search resultsStructured data marking up services, sectors, and accreditations as facts Google can read

A photo grid tells all three engines nothing. A named project, a stated sector, and a marked up credential tell all three something different.

ZENETERA

5. The One Fix That Works For All Three

The overlap between what ChatGPT, Perplexity, and Google's AI Overview each look for is not zero. All three respond to the same underlying fix, a page for each project type the firm actually wants more of, written as a direct answer, naming the exact sector, the exact scale, and a named project as proof, then marked up with schema so Google can read it as a fact too. Firms in our test with even one page built this way for their most wanted project type appeared in two or more engines instead of just one 57 percent of the time.

57%of firms with one project type built as a direct answer page appeared in two or more AI engines, not just one

The Takeaway

Chasing one AI engine at a time is why most architecture, engineering, and consulting firm websites are invisible to two out of three. The fix is not three different strategies, it is one page format built around the project type you actually want more of, answered directly, with the facts marked up so every engine can read them.

01Contact

Let's work together

Which area are you interested in?