The AEO Industry Index measures how AI answer engines recommend the players in a category. This August 2026 edition tracks commercial mechanical contractors across five AI platforms (ChatGPT, Gemini, Perplexity, Claude, and Copilot), scoring how often and how prominently each contractor is named when someone asks AI for guidance.
AEO Index · August 2026 · Updated Aug 4, 2026
The complete AI-visibility index for commercial mechanical contractors.
Two layers of AI visibility, durable memory and the live citation contest, the organic reality beneath them, and where the opportunity sits, in one view.
The signature view
The Delta Quadrant
Every contractor placed by what AI remembers (model memory) against what AI finds when it searches (live retrieval). Bubble size is mention volume. Tap any contractor to focus it; switch Layer, Platform or Signal to move the category.
Adjust this view
Layer
Platform
Signal
Dual leader
Citation-strong · reputation-weak
Reputation-strong · citation-weak
Low / low
What AI remembers → (model memory)
What AI cites now → (live retrieval)
AI has already decided which commercial mechanical contractors it recommends. Here is who, and why.
Two layers of AI visibility: durable memory, and the live citation contest.
Touch to explore
21 views · scroll down · tap any contractor to focus it · ask the index anything
Primary cohort
Extended cohort
Bubble size = mention volume · tap to focus
How this is measured
How this index is scored
Two layers, scored on the same five signals. The layers are what AI knows about a contractor without looking versus what it finds when it searches. The signals are how each layer is measured.
Methodology v2The two layers
Memory (weighted 0.32): what the model has internalized about the category, measured with web search off. The slow-moving reputation layer.
Live retrieval (weighted 0.68): what the model surfaces when it searches the web before answering. The fast, content-driven layer.
Each layer score is the weighted composite of the five signals. The blended score combines them at 0.32 memory and 0.68 retrieval, the same weighting on every Arcalea index so that two categories can be read on one scale.
The five signals
Entity Mention Frequency (EMF) (25%)
Share of tested responses that name the contractor. The foundational frequency signal.
Topical Range Score (20%)
Share of distinct question types where the contractor appears. Breadth, not volume. A contractor that leads its category will often score 1.000 here, because it shows up across every kind of question asked. The signal does its separating work further down the field.
Position Power Score (20%)
Where in the answer the contractor lands. Named first counts for more than named last.
First-Position Rate (15%)
How often the contractor is the first named of any listed. The advocacy signal.
Cross-Platform Stability (20%)
How evenly the contractor appears across engines rather than depending on one. Measured within a layer: the memory layer runs on fewer engines than retrieval, so the two are read within their own layer, not against each other.
Where they stand
Blended leaderboard
Memory and retrieval for every entity; blended is the weighted headline (w_p 0.32 / w_g 0.68). Tap a row to focus it across the whole index. The chip shows recommendation quality: how AI frames the contractor when it names it, a top pick (green) versus hedged with caveats (red). Hover for the recommended / neutral / cautioned split.
BlendedRec. quality
1EMCOR Group0.624+0.29
2Southland Industries0.535+0.29
3ASHRAE0.528
4Comfort Systems USA0.518+0.23
5ENR0.475
6McKinstry0.432+0.24
7TDIndustries0.400+0.26
8Limbach0.368+0.05
9University Mechanical0.363+0.05
10ACCO Engineered Systems0.358+0.18
11Engineering News-Record0.342
12F.E. Moran0.342+0.20
13Harris Companies0.288+0.23
14Hayes Mechanical0.266+0.67
15datacenterdynamics.com0.266
16Murphy Company0.265+0.11
17enr.com0.241
18BD+C0.232
19John W. Danforth0.227+0.50
20Building Design+Construction0.199
21Critchfield Mechanical0.182+0.00
22en.wikipedia.org0.171
23Egan Company0.150+0.00
24The Hill Group0.131+0.27
25MMC Contractors0.088+0.09
26Ivey Mechanical0.069+0.33
27MCAA0.056
28Mechanical Contractors Association of America0.046
29Data Center Dynamics0.039
Rigor you can see
Head-to-head, honestly measured
When AI compares two contractors, the naive answer is mostly an artifact of which is named first. We ask every matchup both ways; the number to trust is order-balanced. Toggle to watch the correction.
← now switch to the honest, order-balanced number
Matchups (first contractor's balanced win rate)
Citation battleground
How AI frames them
Attribute association
Which factors the models associate with each contractor. Darker cell = stronger association.
| Contractor | Data ctr / MC | Design-build | Self-perform | Prefab | Central plant | Healthcare | Industrial | Service | Safety | Geography | Energy/ESG | Reputation |
|---|---|---|---|---|---|---|---|---|---|---|---|---|
| EMCOR Group | 19% | 19% | ||||||||||
| Southland Industries | 17% | 33% | ||||||||||
| Comfort Systems USA | 17% | |||||||||||
| McKinstry | 17% | 17% | 17% | |||||||||
| TDIndustries | 27% | |||||||||||
| Limbach | 33% | 17% | 17% | 17% | ||||||||
| University Mechanical | 33% | 42% | 17% | |||||||||
| ACCO Engineered Systems | 36% | |||||||||||
| F.E. Moran | 53% | 33% | 33% | |||||||||
| Harris Companies | 20% | 20% | 20% | 20% | 60% | |||||||
| Hayes Mechanical | 100% | |||||||||||
| Murphy Company | 22% | 33% | ||||||||||
| John W. Danforth | ||||||||||||
| Critchfield Mechanical | 50% | |||||||||||
| Egan Company | ||||||||||||
| The Hill Group | 25% | |||||||||||
| MMC Contractors | 33% | 33% | 33% | |||||||||
| Ivey Mechanical |
What the machines look for
Search fan-out
One question quietly becomes many. These are the hidden sub-searches the models run to answer a category question, the content brief they wrote themselves: data center and healthcare dominate.
6×leading data center hvac and cooling contractors 2026
6×top mechanical contractors higher education central plants
6×top cleanroom mechanical contractors united states
6×enr top mechanical contractors
6×best commercial mechanical service and maintenance providers
5×design-build mechanical services data center central plants
5×best data center mechanical contractors 2026
5×selecting mechanical contractor healthcare renovation
4×best mechanical contractors hospital central utility plant
4×mechanical contractors campus district energy chilled water experience
Where citations go
Citation flow
Read left to right: the kind of question on the left, the source AI cited on the right. Aggregators and a broad long tail dominate; contractor sites are a thin slice.
Contractor sites
Aggregators
Other
The content pattern
Citation sources & page types
Across 6,648 cited URLs, 3.4% point to contractor sites, 3.6% to aggregators, and 93% to everything else. Of the contractor pages that are cited, the mix concentrates in market vertical (36%) and homepage (28%).
Citation sources by side
contractors 3.4% aggregators 3.6% other 93%
Cited contractor page types
Measured and never the cited page: community forum, editorial blog, reference.
Google's AI layer
Google AI Overviews
An AI Overview appears on 76% of the 50 category keywords, the highest-volume AI surface. Even Google's own AI leans on third parties over the contractors: contractors are cited mostly on their own names, and on non-branded discovery queries citations concentrate on a few. Tap to focus.
Citations by contractor (total / non-branded)
Southland Industries4 / 4 nb
F.E. Moran3 / 3 nb
Harris Companies1 / 1 nb
McKinstry1 / 1 nb
ACCO Engineered Systems1 / 1 nb
Top cited domains
Act 4Organic reality and opportunity
AI vs organic
AEO vs SEO
How AI visibility diverges from organic standing.
Cross-instrument · August 2026| Contractor | AI rank | Organic SoV | Divergence |
|---|---|---|---|
| EMCOR Group | #1 | 0.0% | AI-ahead |
| Southland Industries | #2 | 0.0% | AI-ahead |
| Comfort Systems USA | #3 | 0.0% | AI-ahead |
| McKinstry | #4 | 0.0% | Aligned |
| TDIndustries | #5 | 0.2% | Organic-ahead |
| Limbach | #6 | 0.0% | Organic-ahead |
| University Mechanical | #7 | 0.0% | AI-ahead |
| ACCO Engineered Systems | #8 | 0.0% | AI-ahead |
| F.E. Moran | #9 | 0.0% | AI-ahead |
| Harris Companies | #10 | 0.0% | Organic-ahead |
| Hayes Mechanical | #11 | 0.0% | Organic-ahead |
| Murphy Company | #12 | 0.0% | Organic-ahead |
| John W. Danforth | #13 | 0.0% | Aligned |
| Critchfield Mechanical | #14 | 0.0% | Organic-ahead |
| Egan Company | #15 | 0.0% | Organic-ahead |
| The Hill Group | #16 | 0.0% | Aligned |
| MMC Contractors | #17 | 0.0% | Aligned |
| Ivey Mechanical | #18 | 0.0% | Aligned |
The citation economy
AI citations
Citation rate and the domains AI leans on.
Grounded layer · August 202695.9%
responses with citations
6,651
total citations
3.4%
to a tracked contractor
midsouthmechanical.com
113
linkedin.com
98
enr.com
94
siteline.com
91
thebakergroup.com
85
datacentremagazine.com
80
metromech.com
77
reddit.com
68
anchormechanical.com
66
usengineering.com
65
en.wikipedia.org
61
sprint-mechanical.com
59
Demand beneath visibility
Branded search footprint
Branded-search presence per contractor.
DataForSEO · August 2026EMCOR Group
89
McKinstry
83
University Mechanical
82
Hayes Mechanical
78
The Hill Group
78
Southland Industries
76
AMS Industries
74
Murphy Company
73
Harris Companies
73
TDIndustries
71
Comfort Systems USA
67
John W. Danforth
66
Pan-Pacific Mechanical
66
MMC Contractors
66
Egan Company
63
Ivey Mechanical
61
The organic baseline
Organic share of voice
Share of estimated addressable organic clicks.
DataForSEO · August 2026| Entity | Organic SoV | Keywords ranking | Est. clicks/mo |
|---|---|---|---|
| TDIndustries | 0.2% | 6 | 302 |
| Critchfield Mechanical | 0.0% | 1 | 25 |
| Harris Companies | 0.0% | 1 | 19 |
| Limbach | 0.0% | 1 | 19 |
| McKinstry | 0.0% | 2 | 16 |
| Hayes Mechanical | 0.0% | 4 | 15 |
| Comfort Systems USA | 0.0% | 1 | 8 |
| Murphy Company | 0.0% | 1 | 8 |
| Egan Company | 0.0% | 1 | 2 |
| Southland Industries | 0.0% | 4 | 0 |
| ACCO Engineered Systems | not ranked | 0 | 0 |
| The Hill Group | not ranked | 0 | 0 |
| MMC Contractors | not ranked | 0 | 0 |
| F.E. Moran | 0.0% | 5 | 0 |
| John W. Danforth | 0.0% | 1 | 0 |
| John J. Kirlin | not ranked | 0 | 0 |
Keyword reality
SERP and keyword coverage
Keyword-universe SERP appearances per contractor.
DataForSEO · August 2026| Contractor | SERP appearances | KW coverage | Top-3 rate | SEO composite |
|---|---|---|---|---|
| reddit.com | 42 | 11% | 1% | 0.249 |
| ENR | 11 | 3% | 2% | 0.222 |
| F.E. Moran | 4 | 1% | 1% | 0.179 |
| csemag.com | 1 | 0% | 0% | 0.179 |
| Comfort Systems USA | 2 | 0% | 0% | 0.179 |
| University Mechanical | 8 | 2% | 0% | 0.138 |
| Egan Company | 1 | 0% | 0% | 0.129 |
| Southland Industries | 6 | 2% | 0% | 0.099 |
| ashrae.org | 2 | 1% | 0% | 0.095 |
| bdcnetwork.com | 4 | 1% | 0% | 0.095 |
| TDIndustries | 6 | 2% | 0% | 0.089 |
| Hayes Mechanical | 5 | 1% | 0% | 0.086 |
| mcaa.org | 5 | 2% | 0% | 0.079 |
| en.wikipedia.org | 7 | 2% | 0% | 0.069 |
| McKinstry | 3 | 1% | 0% | 0.061 |
| John W. Danforth | 1 | 0% | 0% | 0.004 |
"not ranked" means the contractor ranks for none of the tracked keywords, so no composite is computed. That is absence from this keyword universe, not a score of zero.
Open vs owned
Category concentration and saturation
Where the field is open versus owned.
Grounded layer · August 2026| Category | Concentration (HHI) | State |
|---|---|---|
| Comparison | 1.00 | Concentrated |
| Educational | 1.00 | Concentrated |
| Decision | 0.50 | Concentrated |
| Geographic | 0.39 | Concentrated |
| Solution | 0.27 | Contested |
| Discovery | 0.18 | Open |
| Other | 0.14 | Open |
Unclaimed territory
White space
High-volume query clusters where no cohort contractor has built a strong position.
DataForSEO · August 2026| Cluster | Volume/mo | Keywords | Avg coverage | Status |
|---|---|---|---|---|
| Local Geographic | 280,730 | 66 | 0% | Open |
| Discovery | 141,850 | 302 | 0% | Open |
| Service Lifecycle | 13,740 | 38 | 0% | Open |
| Other | 9,780 | 13 | 0% | Open |
| Industrial Process | 390 | 9 | 0% | Open |
| Capability Specialty | 200 | 7 | 0% | Open |
| Healthcare | 20 | 4 | 0% | Open |
| Data Center | 0 | 6 | 0% | Open |
| Higher Education | 0 | 3 | 0% | Open |
| Life Science | 0 | 4 | 0% | Open |
Drill down
Entity explorer
Per-contractor two-layer and organic detail. Tap any row to open the five signals behind its memory and retrieval scores.
Two-layer · August 2026| Contractor | Memory | Retrieval | Blended | AI mentions | Organic SoV | SEO composite |
|---|---|---|---|---|---|---|
| EMCOR Group | 0.749 | 0.566 | 0.624 | 127 | not ranked | not ranked |
| Southland Industries | 0.633 | 0.488 | 0.535 | 93 | 0.0% | 0.099 |
| Comfort Systems USA | 0.550 | 0.502 | 0.518 | 96 | 0.0% | 0.179 |
| McKinstry | 0.547 | 0.379 | 0.432 | 51 | 0.0% | 0.061 |
| TDIndustries | 0.560 | 0.324 | 0.400 | 70 | 0.2% | 0.089 |
| Limbach | 0.483 | 0.314 | 0.368 | 48 | 0.0% | 0.000 |
| University Mechanical | 0.242 | 0.419 | 0.363 | 28 | 0.0% | 0.138 |
| ACCO Engineered Systems | 0.474 | 0.303 | 0.358 | 44 | not ranked | 0.004 |
| F.E. Moran | 0.000 | 0.503 | 0.342 | 15 | 0.0% | 0.179 |
| Harris Companies | 0.272 | 0.296 | 0.288 | 20 | 0.0% | 0.000 |
| Hayes Mechanical | 0.000 | 0.391 | 0.266 | 1 | 0.0% | 0.086 |
| Murphy Company | 0.210 | 0.290 | 0.265 | 18 | 0.0% | 0.000 |
| John W. Danforth | 0.198 | 0.241 | 0.227 | 5 | 0.0% | 0.004 |
| Critchfield Mechanical | 0.000 | 0.267 | 0.182 | 2 | 0.0% | 0.000 |
| Egan Company | 0.000 | 0.221 | 0.150 | 1 | 0.0% | 0.129 |
| The Hill Group | 0.000 | 0.192 | 0.131 | 4 | not ranked | not ranked |
| MMC Contractors | 0.000 | 0.130 | 0.088 | 3 | not ranked | 0.000 |
| Ivey Mechanical | 0.000 | 0.101 | 0.069 | 1 | not ranked | not ranked |
The freshness lever
Freshness of cited pages
How recently each contractor's most-cited pages were last updated. Pages refreshed within 90 days earn far more AI citations than stale ones.
Live fetch · August 2026| Contractor | Fresh <90d | Median age | Pages dated |
|---|---|---|---|
| Southland Industries | 100% | -2d | 2/2 |
| McKinstry | 0% | 253d | 1/1 |
| TDIndustries | 100% | 0d | 1/1 |
| ACCO Engineered Systems | 100% | -3d | 1/1 |
| F.E. Moran | 100% | -3d | 1/1 |
| Harris Companies | 100% | 25d | 1/1 |
| Murphy Company | 0% | 1445d | 1/1 |
The infrastructure lever
Citability infrastructure
Whether each contractor has the machine-readable infrastructure AI leans on to recognize and cite an entity: structured data, an encyclopedia presence, an AI-crawler guide, and named authors.
Live checks · August 2026| Contractor | Schema | Wikipedia | Wikidata | llms.txt | Authors | Infra score |
|---|---|---|---|---|---|---|
| EMCOR Group | · | ✓ | ✓ | ✓ | · | 60% |
| Southland Industries | ✓ | · | · | · | · | 20% |
| Comfort Systems USA | ✓ | ✓ | ✓ | · | · | 60% |
| McKinstry | ✓ | ✓ | ✓ | · | · | 60% |
| TDIndustries | · | ✓ | ✓ | · | ✓ | 60% |
| Limbach | · | · | · | · | · | 0% |
| University Mechanical | · | · | · | ✓ | · | 20% |
| ACCO Engineered Systems | · | · | · | · | · | 0% |
| F.E. Moran | · | · | · | ✓ | · | 20% |
| Harris Companies | ✓ | · | · | · | · | 20% |
| Hayes Mechanical | ✓ | · | · | · | · | 20% |
| Murphy Company | ✓ | · | · | · | ✓ | 40% |
| John W. Danforth | · | · | · | ✓ | · | 20% |
| Critchfield Mechanical | · | · | · | ✓ | · | 20% |
| Egan Company | · | ✓ | · | ✓ | · | 40% |
| The Hill Group | · | · | · | · | · | 0% |
| MMC Contractors | ✓ | · | · | · | ✓ | 40% |
| Ivey Mechanical | ✓ | · | · | · | · | 20% |
What to do
Recommendations
1
Publish a canonical 'how to select a commercial mechanical contractor' resource.
EvidenceSelection-phrased sub-queries (best/top lists and comparisons) are 7.5% of the composite fan-out, and the sector-capability queries buyers use to build a shortlist dominate the rest; with only 3.4% of citations pointing to a tracked contractor's own site, the how-to-choose answer is currently assembled from third parties.
DirectionBuild a buyer-facing selection guide (evaluation criteria, delivery models, red flags) so AI cites your firm as the authority on the decision itself, not just a vendor in a list.
2
Own the data-center and mission-critical proof.
EvidenceData center is the number one fan-out theme (394 of 875 sub-searches, 45%); the category leader is framed on reputation and mission-critical work.
DirectionPublish mission-critical project proof (capacity, uptime, references) so AI ties your firm to the segment buyers ask about most.
3
Become the cited source, not a mention in someone else's list.
EvidenceOnly about one in ten AI citations point to a contractor's own site; the rest go to directories and a broad long tail.
DirectionMake your capabilities, project, and outcome pages the most extractable source on your core topics, and earn placement in the industry directories the models lean on.
4
Turn capabilities into structured, citable capability pages.
EvidenceCapability and self-perform scale recur across the sub-searches; the leaders are framed on reputation and self-perform scale.
DirectionPublish structured pages for self-perform trades, prefab and modular, and central-plant HVAC with specifics (crew size, tonnage, delivery model) that AI can lift verbatim.
5
Publish sector proof for healthcare and industrial process.
EvidenceHealthcare and industrial process surface as distinct fan-out themes: buyers research by sector, not just by trade.
DirectionBuild sector landing pages with named projects and measurable outcomes per vertical so AI can match your firm to a specific sector query.
6
Build entity authority so AI recognizes your firm.
EvidenceGoogle AI Overviews appear on 76 percent of category searches but concentrate on a few recognized names; visibility compounds for known entities.
DirectionStrengthen Wikipedia and Wikidata presence, keep firm data consistent across the web, and earn third-party mentions so the models have more signal to attach to your firm.
These are directional, ordered by leverage: which layers to close, not which pages to edit. The page-level plan for one contractor (the specific URLs, entities, schema, and crawler fixes, sequenced with owners) is a GEO Audit.
Next step
Turn this into a plan.
This index shows where each contractor stands and which layer, model memory or live retrieval, is holding it back. A GEO Audit turns that into a prioritized action plan for one contractor: the pages, entity infrastructure, schema, and AI crawler access to fix, in order, with the measurement re-run afterward to prove the lift.
Arcalea AEO Index · neutral two-layer baseline August 2026 · 18 contractors · a Galileo-family measurement surface
FAQ
Common questions about the AEO Index
What is the AEO Industry Index?
What is the difference between the memory and live-retrieval layers?
The index measures two instruments and never blends them. The parametric (memory) layer queries each model with web search off, capturing what the model has internalized about the category. The grounded (live-retrieval) layer lets the models search the live web before answering. Memory is the slow-moving reputation moat; retrieval is the fast, content-driven citation contest. The gap between the two layers is itself the finding.
Why does a contractor rank differently on memory versus live retrieval?
A contractor can hold a strong place in a model's memory yet lose the live-retrieval layer when competitors publish more citable, current content, or the reverse. The Delta Quadrant plots both layers at once, so you can see which lever a contractor must pull: build durable reputation, or win the live citation contest.
How is AI visibility different from an SEO ranking?
SEO measures where a page ranks on a results page. AEO measures whether a brand is named and cited inside the AI-generated answer, where most people never click through to a link. A contractor can rank well organically and still be absent from the AI answer, or lead the AI answer while trailing in organic search. This index reports both and shows where they diverge.
Which platforms and prompts does the index use?
Five platforms: ChatGPT, Gemini, Perplexity, Claude, and Microsoft Copilot. Each runs the same frozen prompt set. Most prompts are entity-neutral, meaning no contractor is named and the model chooses who to recommend on its own; a separate head-to-head track deliberately names two contractors per prompt to test direct comparisons. Scores aggregate across platforms and prompt categories, with per-platform detail in each section.
Can Arcalea build this index for my category?
Yes. The methodology is category-agnostic. Give us your market and competitive set and we build the same two-layer index for your industry and your brand's position within it. Use the Get Your Index button to start.