009 Agency does Generative Engine Optimization for visualisation studios: we make ArchViz, CGI and game-art studios visible, cited and chosen inside AI answers across ChatGPT, Claude, Gemini, Perplexity, Grok and DeepSeek — and we re-measure it every two weeks against a logged baseline. There is no Google demand for a “visualisation studio SEO agency”, which is why this page exists anyway: the buyers do not search for an agency, they ask an assistant which studio to hire, and the answer is being written without you.
Visual studios are the most homogeneous bench in our dataset: five audits — four architectural-visualisation studios in London, Budapest, Bradford and Dubai, and one global game co-development group — with the same finding in every one. The work is images; the engines read text; the portfolio is proof that nothing can parse. Everything else on this page follows from that.
why studios lose in AI answers specifically
The general failure pattern holds everywhere: known but not recommended. In our published dataset 93% of brands were described accurately by AI and only 4.5% were named when a buyer asked who to pick. Studios show the widest version of that gap we have measured.
known to the world, unknown to the buyer
A top-tier ArchViz studio with clients like BIG, Zaha Hadid and Kengo Kuma was named on 7 of 12 “best studios in the world” prompts — and on 0 of 12 prompts phrased the way a developer actually asks: a marketing CGI package for a scheme in New York, London or Dubai. The engines answered the developer with other studios, one of them for a Dubai project when the client had Dubai Holding in its portfolio. A game co-development group with credits on Alan Wake 2 and Jedi: Survivor had the best recall in our whole set, 12 of 14, and was cited as a source on 0 of 10. Known, admired, and not the answer.
the shelf is built from studio-published roundups and awards, not from traffic
Asked on 5 September which studios a London developer should hire for a marketing CGI package, ChatGPT named The Boundary, DBOX, Hayes Davidson, Visualhouse, Forbes Massie, Recent Spaces, Cityscape Digital and PixelFlakes, citing Maverick Frame’s roundup, a press-release wire and Wikipedia. Perplexity named The Boundary, Hayes Davidson, Maverick Frame, Recent Spaces, Forbes Massie and Arqui9, citing Maverick Frame twice and a chain of small studio blogs. In an earlier audit, ChatGPT’s “best 3D rendering companies” answer named twenty prestige studios from the CGarchitect awards, Architizer and reality-atlas — and not one of the studios that lead the Google results for the same query. The SEO tier and the AI tier are disjoint sets. One studio in our set traffic-outranked five studios Perplexity recommended, by up to thirteen times, and was not recommended.
your own roundup feeds the competitor, and your entity is not on file
One studio published “best 3D rendering companies” on its own blog, listed itself last, with no criteria and no links. Perplexity cited that article as its source and named two other studios from it; Claude cited it three times and used it to describe two more. Another studio’s branded answer in ChatGPT was built from ten sources, all of them the studio’s own site, with the verdict “self-published, not independently proven”; the engine’s search log showed it looking for the studio on Clutch, DesignRush and Trustpilot and finding nothing. A third had its founders on the site, at the bottom of the contact page under a mega-menu, and ChatGPT told the buyer it could not confirm who owned the company.
what the shelf looks like in our audits
| studio type | what we measured | what the engines did |
|---|---|---|
| ArchViz, London HQ, six languages | category recall across US, UK, Germany and Dubai; brand verdict | 2 of 6 engine-market pairs; the one Claude hit came from a single third-party listicle. Branded answer built entirely from the studio’s own site: “not independently proven”. A live, indexed testimonials page of Lorem ipsum with invented names. |
| ArchViz, Bradford and Dubai | “best 3D rendering companies”, Dubai studio prompt, brand facts | ChatGPT named 20 prestige studios from awards and roundups, none from the SEO tier; the studio was on Perplexity via its own blog. The engine told the buyer, unprompted, that the founding date was “unclear” (2016 on site, 2009 on LinkedIn). |
| ArchViz, clients HOK and DAMAC | hiring prompt, brand verdict, AI Overview citations | Google’s AI Overview cited the studio 30 times; the assistants cited it 0 times and named ZOA and Darcstudio from the studio’s own roundup. Founders present on the site, unread; “could not confirm owners or address”. |
| ArchViz, Budapest, clients BIG and Zaha Hadid | 12 prompts: world’s best studios vs developer briefs in NYC, London, Dubai | 7 of 12 on “best in the world” via reality-atlas and Maverick Frame; 0 of 12 on developer briefs; the one citation was the studio’s training academy. Redesign shipped without redirects, schema 0, Lorem ipsum in the footer. |
| game co-development, 1,100 staff | 14 recall and 10 citation prompts | recall 12 of 14, the best in our dataset; citation 0 of 10. Brand split across three domains; the homepage returned 500 during the audit; Claude’s crawler was blocked at the edge. |
| developer brief, London (live, 5 sep) | “which studios for a marketing CGI package” | ChatGPT: The Boundary, DBOX, Hayes Davidson, Visualhouse, Forbes Massie, Recent Spaces, Cityscape Digital, PixelFlakes. Perplexity: The Boundary, Hayes Davidson, Maverick Frame, Recent Spaces, Forbes Massie, Arqui9. Maverick Frame’s roundup cited by both. |
what the first 90 days look like
The order is the same as everywhere, because it has to be: readable, then verifiable, then quotable.
- Audit. The prompts developers, architects and publishers actually ask — per city, per project type, per engine — run live and logged verbatim, plus what each AI crawler receives from a portfolio page and what the engine says when asked who you are. You get a health score, the shelf in your markets, and the three most expensive gaps. Complimentary. what is inside →
- Give the images words and the studio a record. Project pages with client, location, scope and outcome as text; Organization and founder records that resolve; one founding date everywhere; the redesign’s redirects; Clutch and the directories the engines look you up on. Shipped as files your web vendor applies. technical GEO →
- Publish what gets quoted. One page per city and project type a developer asks about, and a roundup done properly: criteria, links, disclosure, and you not last. The academy, the awards page and the case study are the citable assets studios already have and never mark up.
- Get onto the lists. The studio roundups, awards and architecture media the engines actually cited in your markets — Maverick Frame, reality-atlas, CGarchitect, Architizer in our measurements — identified from your own panel, not from a generic outreach list.
- Re-measure. Same panel, every review, reported as a delta. Series, not screenshots.
proof, and its limits
The studio bench is five audits and no client on retainer, so every number on this page is an audit measurement, not a program result. The measured result we can point to is from another vertical: a legal client went from 0/5 to 4/5 engines naming it on its category prompt by day 17 (July 2026), with the log published. Studios start from a better place than that client did — the world already knows the work — and from a worse one: the work is in a format the engines cannot read.
one thing we will not do
We will not mark up an award we cannot find. In one audit a “World’s Best” property award on the site returned nothing on the awarding body’s own search, while the same search found a competitor’s real win. We recorded it as unverified, not as false, and kept it out of the schema. An engine that checks the same source will do the same, and it will tell the buyer.