Key takeaways
- We ran our live AI visibility audit tool against 23 architecture, engineering and construction domains and could scan 20. Average visibility was 60.0% across 100 live, grounded buyer queries, and the median was 60.0%. Three domains, 13%, blocked our crawler.
- One firm was named in all five queries and five more in four of five. Sixteen of 20 were named in at least three. One, Suffolk Construction, was not named in any of its five queries.
- Comparison questions went best (85%) and category questions next (70%). Problem-first questions named a firm 45% of the time, and local-style questions only 30%, 6 of 20.
- AI leaned on encyclopedias, directories and trade-media rankings. Wikipedia was 76 of 792 citations, Yelp 28 and Building Design + Construction 22. A firm’s own website was only 1.3% of what was cited.
- Seven of the 10 citations to a firm’s own site went to a sector, service or project-type page, not a homepage. For an owner asking about a specific type of project, that is the page AI chose to quote.
- This sample is mostly very large global and national firms, chosen by us, and it mixes architects, engineers and contractors. It likely flatters small and mid-size regional firms and says nothing about them directly.
Why we audited AEC firms
Architecture, engineering and construction is bought through relationships and paperwork. An owner, developer or facility manager forms a view of who is credible, then an RFQ or RFP goes to a short list, then prequalification, references and interviews decide it. Much of the work is won before a formal request exists, and the firms that win it assume their record speaks for itself.
But the early view is starting to include an AI assistant. A developer can describe a mixed-use site, or a facility manager a warehouse budget, and read the firms it names. So we asked a simple question. When a buyer describes a project type without naming a firm, which AEC firms get named, and does being famous guarantee a place on the list?
60.0%
average AI visibility across 20 established AEC firms, measured by live grounded search on 100 buyer queries
Deligatr AEC Firm AI Visibility Audit, October 3, 2026
That is the headline, and it is flattered by who is in the sample. The more useful result is where the 40% went missing.
How the audit works
The tool reads each firm's homepage, classifies it in the words a buyer would use, writes five buyer-intent queries (one local-style, two category, one problem-first, one comparison) and runs each as a live, grounded AI search with the firm's own name left out. It is one run, five queries per firm, on one day. AI answers change between runs, so read every score as a snapshot. The full method is in our audit documentation, and all twelve industry audits are on our AI visibility research page.
A firm is either in the answer or it is not.
The distribution: a strong middle, and a tail that matters
Firms by AI visibility score
20 AEC firms, each tested on 5 live grounded buyer queries. Number of firms at each score.
Deligatr AEC Firm AI Visibility Audit, October 3, 2026
Only HDR was named in all five queries. Clark Construction, DPR Construction, Gensler, Gilbane and AECOM were named in four. Ten firms landed on exactly three, which makes 60% the typical result in a sample of very well-known names.
The tail is the story. Swinerton was named in one query, a comparison question about the top-ranked general contractors, and not in its other four. Suffolk Construction was not named in any, including a comparison question about the best commercial construction companies in America. These are large, established builders, and a famous name did not secure a place in the answer.
What kind of firm did best
The sample mixes three kinds of firm, and the averages differ, though the groups are small and our labels are ours.
- Architects (Gensler, HOK, NBBJ, SOM): 65.0% on average, four firms.
- Engineers and architecture-engineering firms (AECOM, Burns & McDonnell, HDR, Jacobs, Kimley-Horn, Stantec): 63.3% on average, six firms.
- Contractors and construction managers (Clark, DPR, Gilbane, Hensel Phelps, McCarthy, Mortenson, Skanska, Suffolk, Swinerton, Turner): 56.0% on average, ten firms.
We would not read much into the ranking of three small groups, and Gilbane and Skanska also do development. The point is narrower: the weakest two scores in the sample were both contractors, and both were asked generic questions about commercial builders, where the answers named the same few national names and a spread of local firms.
What kind of question loses the most
How often a firm was named, by query type
Share of queries in which the firm was named. 20 firms, 100 queries.
Deligatr AEC Firm AI Visibility Audit, October 3, 2026
Comparison queries named a firm 17 times out of 20, and category queries 28 times out of 40. These are the "largest" and "top" questions, which trade-media ranking pages are written to answer, and they favor well-known firms.
Local-style questions named a firm 6 times out of 20. Problem-first questions, where the buyer describes a project and asks who to hire, named a firm 9 times out of 20.
The local answers were different in kind. Across them, AI named 49 distinct firms as alternatives, and 40 of those, by exact name, appeared in no other query type. Examples include Clawson Architects, named in the local answers of three different architecture firms, and Claris Design Build, named in the local answers of two different contractors. Queries like "general contractor near me" mostly surfaced firms the tested company is not usually compared with, and some of the answers drew on Yelp, Houzz and BBB. One caveat: for 11 of the 20 firms the tool detected no location, so "near me" ran without a place.
That is relevant to a regional firm. A buyer who adds a city or a project type and a place is asking the question a smaller firm is more likely to win, if its site and profiles say clearly where it works and what it builds.
Who got named instead, and where the answers came from
The alternatives were the industry's biggest names. Counting name variants as one, Turner Construction was named as an alternative in 25 of 100 answers, Bechtel in 17, AECOM in 15, Gensler and Perkins & Will in 11 each, and Kiewit, HOK, WSP and DPR Construction in 10 each. Bechtel was not in our sample, and Kiewit and Perkins&Will were among the sites we could not scan, so these three were named in other firms' answers while we could not test their own.
The sources were broad. Across 792 citations on 394 domains, a few repeat names stood out.
- Wikipedia was the largest single source, at 76 citations (9.6%) across 59 pages. Entries for firms such as Page, Lauren Engineers & Constructors and NAC Architecture were each cited three or four times, so presence there is not only for the giants.
- Yelp was cited 28 times, all on local-style queries, and Houzz nine times, also on local-style queries.
- Trade-media rankings led the comparison answers. Building Design + Construction (bdcnetwork.com), home of the Giants 400 lists, was cited 22 times, mostly on comparison questions. Construction Digital, Biscred, Building Radar, Gobridgit and Mastt each appeared repeatedly with "largest" and "top" lists.
- Sector-specific lists appeared where the project type was specific. For data center questions, Data Centre Magazine's top-ten lists were cited in several answers.
1.3%
of all citations pointed to the firm's own website (10 of 792), measured by exact match on the firm's domain
Deligatr AEC Firm AI Visibility Audit, October 3, 2026
Eight of the 20 firms had at least one citation to their own site, and the other twelve had none.
The project-type page is the unit AI quotes
The ten own-site pages that were cited are the most practical finding in this audit. Three were homepages (AECOM, DPR Construction and Kimley-Horn). Seven were not:
- DPR Construction's mission-critical and healthcare pages
- McCarthy Building Companies' healthcare projects page
- Mortenson's data centers page
- Hensel Phelps' water and wastewater page
- Burns & McDonnell's substations page
- Turner Construction's project listing filtered to one market segment
Each is about one project type. The question that pulled them in was a project-type question too: who builds hospitals, who manages a data center build, who builds a wastewater plant.
The same pattern showed up on other firms' sites. In one answer to "engineering firms for data center design near me," AI cited the data center pages of five other engineering firms (Colliers Engineering & Design, Barge Design, Enercon, Karbon Engineering and RSP Engineers), alongside several trade-media and market-research lists.
We did not measure whether a project-type page raises a firm's chance of being named, and the audit cannot show cause. It does show what AI chose to quote when it quoted a firm's own site, and it did not choose the portfolio gallery.
The three we could not scan
Three of the 23 domains we tried, 13%, could not be audited: Kiewit, Perkins&Will and Whiting-Turner. Our crawler was shown a bot-verification page or an error instead of the homepage.
What that means is limited. We cannot tell from outside whether the same protection also stops the crawlers AI assistants use. If it does, the firm's content is not being read at all. If it does not, there is no problem. The only way to know is to test it.
What to do with this, if you run an AEC firm
- Build a page for every project type you want more of. One page per sector or typology, covering the owner type, size, delivery method, your role and what the project delivered. Name real projects and release them with the owner's permission.
- Make the portfolio quotable. A case study that opens with one plain paragraph (who the owner was, what was built, what you did, what happened) can be lifted into an answer. A photo gallery cannot.
- Explain delivery methods and selection honestly. Owners ask whether to use design-build or design-bid-build, and public agencies ask how selection works. A balanced guide that says when each fits is rare, so it is quotable.
- State licenses, registrations and credentials exactly. Show the licenses held, by state, the way your boards expect, and keep certifications, bonding and safety figures current and verifiable. Only licensed individuals may use the title architect in most states, so check how your people are described.
- Check the rankings and directories. Compare your listing across the trade-media rankings in your sector, the contractor and design directories, and your Wikipedia entry if you have one. Inconsistent or missing entries make you hard to place.
- Say where you work. A local-style question has no giants to hide behind. Offices, regions served and local projects, stated consistently, give AI something to place.
- Test whether your site can be read, then re-test regularly. AI answers change, so a single score tells you little.
Because the buyer universe is small and the cycle is long, many AEC firms also benefit from pairing visibility with direct outreach to a named list of owners and developers, aimed at organizations and kept out of open public solicitations. If you want this done as a program, that is what our AEO, GEO and SEO service for AEC firms is built for.
Every firm we audited
Firm | Visibility | Appeared on | Top rival named instead |
|---|---|---|---|
| HDR | 100% | Local, Category x2, Problem, Comparison | HOK |
| AECOM | 80% | Local, Category, Problem, Comparison | Turner & Townsend |
| Clark Construction | 80% | Category x2, Problem, Comparison | Turner Construction Company |
| DPR Construction | 80% | Local, Category, Problem, Comparison | Turner Construction |
| Gensler | 80% | Category x2, Problem, Comparison | Perkins & Will |
| Gilbane | 80% | Local, Category x2, Comparison | Clayco |
| Burns & McDonnell | 60% | Category x2, Problem | AECOM |
| Hensel Phelps | 60% | Category x2, Comparison | DPR Construction |
| HOK | 60% | Category, Problem, Comparison | Gensler |
| Jacobs | 60% | Local, Category, Comparison | WSP |
| McCarthy Building Companies | 60% | Category x2, Comparison | Turner Construction Company |
| Mortenson | 60% | Local, Category, Problem | Turner Construction |
| NBBJ | 60% | Category, Problem, Comparison | Stantec |
| Skanska | 60% | Category x2, Comparison | Adler Builders |
| SOM | 60% | Category x2, Comparison | Gensler |
| Turner Construction | 60% | Category x2, Comparison | Bechtel |
| Kimley-Horn | 40% | Category, Comparison | WSP |
| Stantec | 40% | Category, Comparison | HDR |
| Swinerton | 20% | Comparison | DPR Construction |
| Suffolk Construction | 0% | None | Claris Design Build |
Not scanned (bot-verification page or error): Kiewit, Perkins & Will, Whiting-Turner.
Check your own number
Every firm above was run through the same tool below. It takes about ninety seconds, and your firm's name is never in the query.
Do AI assistants recommend you?
Five buyer-style queries for your category, run through live AI search, show who gets named instead of you. Takes about ninety seconds.
Free. The full report arrives by email so the tab can close.
Methodology and honest limitations
What we did. On October 3, 2026 we ran the live audit tool that powers our free checker against 23 architecture, engineering and construction domains: one run, five queries per firm, visibility measurement only. The step-by-step method is in our audit documentation. A firm that disputes its result can ask us to re-run it.
The sample. We chose real, identifiable firms, mostly very large global and national architects, engineers and contractors. It mixes three kinds of firm, and a few, such as Gilbane and Skanska, also develop. It is a sample we picked, not a random one and not a census. It likely flatters the average for small and mid-size regional firms, which are barely in it. A firm's score also depends on how our classifier read its homepage, and with 20 firms, one result moves the average noticeably. The grouping into architects, engineers and contractors is ours.
What it does not cover. Only the homepage was read, so a firm with a thin homepage and strong project pages may be undercounted. Five queries per firm is a small sample of real buyer language, and for 11 of 20 firms the local-style query ran without a detected location. We did not test RFP behavior, prequalification, awards or any claim a firm makes about its licenses, safety record or certifications. Results come from one run on one day, and AI answers change over time. We did not test whether any firm pays for ranking or directory placement.
On licensing and procurement. Nothing here is legal advice. Licensing, advertising and procurement rules differ by state, country and agency, and your own licensing boards, counsel and solicitation documents apply.
FAQ
How visible are established AEC firms in AI answers?
Moderately, with a clear weak spot. Across the 20 firms we could scan, average visibility was 60.0% over 100 buyer queries. Sixteen of 20 were named in at least three of five queries, but local-style questions named a firm in only 6 of 20 answers. This sample is mostly very large global and national firms, so it likely flatters small and mid-size regional firms.
Does this say how a mid-size regional firm would do?
No. We did not test regional firms. The 20 we scanned are among the best known in their fields, and a smaller firm is likely to be named less often on generic questions. The one encouraging signal is that local-style answers often named firms we did not test, which suggests there is room for firms that are clear about where and what they build.
Who did AI name instead?
The same giants, repeatedly. Counting name variants as one, Turner Construction was named as an alternative in 25 of 100 answers, Bechtel in 17, AECOM in 15, and Gensler, Perkins & Will, Kiewit, HOK, WSP and DPR Construction in 10 or 11 each.
Where does AI get its answers about AEC firms?
From encyclopedias, directories and trade-media rankings. Across 792 citations on 394 domains, Wikipedia was the largest source at 76, followed by Yelp (28) and Building Design + Construction (22), whose ranked lists were cited most on comparison questions. A firm’s own website was 1.3% of what was cited.
What kind of page on a firm’s own site did AI cite?
Mostly project-type pages. Of the 10 citations that pointed to a tested firm’s own domain, 7 went to a sector, service or project-type page, such as a healthcare, mission-critical, data center or water page. Three went to a homepage.
Why could three of the 23 sites not be scanned?
Kiewit, Perkins & Will and Whiting-Turner showed our audit crawler a bot-verification page or an error instead of the homepage. That is 13% of the domains we tried. We cannot tell from outside whether the same protection also stops the crawlers AI assistants use.
Related reading
- AEO, GEO and SEO for architecture, engineering and construction firms
- We audited 22 established law firms for AI visibility
- We audited 13 industrial manufacturers for AI visibility
- How to get cited in ChatGPT, Perplexity and AI Overviews
- What is answer engine optimization (AEO)? The complete guide
- AI visibility research: every audit in one place
