Two measurements in one report: when AI names you, is it right — and when it doesn't, are you even in the conversation?
GEO has two halves. One: when someone asks about Apex Environmental Partners by name, do the models describe you accurately? Two: when someone asks about environmental consulting without naming anyone, do you get surfaced at all? This audit measures both, across 5 leading AI models, in both trained and live-search modes.
Apex Environmental is described accurately but too narrowly when named (an air-quality firm, not the full EHS partner it is), and it surfaces in only about a third of unbranded category answers, where larger competitors dominate. The entity is solid; the reach and the breadth of the story are the gaps.
Every score below is measured against this reference, built from the company's own site and verified public records — not our impression. It also defines the competitor set used in the visibility ranking.
| Dimension | Intended value |
|---|---|
| One-line positioning | A globally recognized environmental, health, and safety (EHS) consulting firm for highly regulated industries. |
| Core services | Air Quality, Chemical Reporting & Management, Digital Solutions, ESG, Investigation & Remediation, Occupational Health & Safety, Waste Management, Water Quality. |
| Industries served | Pulp & paper, Food & beverage, Consumer products, Cement, Chemicals & pharma, Waste, Power, Oil & gas, Data centers, Automotive. |
| Differentiators | Regulatory + industry staff, Client-success model, Best Place to Work, Verdantix Green Quadrant Innovator, ~320 professionals, National footprint. |
| Competitor set (benchmark) | NorthPeak Consulting, Vantage Partners, Calder & Finch, Meridian Group. |
| Do NOT want | Framed as air-quality-only, regional/small, or confused with unrelated "All 4" entities. |
The nine-part branded prompt battery, run cold and with live search across 5 models, three runs each, scored against the ground truth above.
| Model | Mode | Accuracy | Completeness | Positioning | Notable behavior |
|---|---|---|---|---|---|
| ChatGPT (GPT-5) | Search on | High | Med | Med | Full service list when searching; air-only when not |
| Claude | Search on | High | High | Med | Cited owned site and LinkedIn; flagged PE ownership |
| Gemini | Search on | Med | Med | Med | Directory data; revenue figures conflicted |
| Perplexity | Search on | High | High | Good | Strongest sourcing; competitor set accurate |
| Copilot | Search on | Med | Low | Low | Air-quality framing dominated; newer disciplines missing |
We generated 20 category questions that name no brands but force brand mentions in the answer — split between Middle-of-Funnel (exploring options) and Bottom-of-Funnel (ready to buy). Each ran across 5 models. Mentions are counted in code, not by hand, so the score is measured rather than approximate.
Apex Environmental Partners appeared in 7 of 20 answers — a visibility score of 35%. For a specialist firm this is a workable base, but three competitors appear roughly twice as often, especially in early-stage exploring questions.
| # | Brand | Appears in | Share of answers |
|---|---|---|---|
| 1 | Meridian Group | 14 / 20 | 70% |
| 2 | NorthPeak Consulting | 12 / 20 | 60% |
| 3 | Calder & Finch | 9 / 20 | 45% |
| 4 | Apex Environmental (you) | 7 / 20 | 35% |
| 5 | Vantage Partners | 6 / 20 | 30% |
Apex Environmental performed better on Bottom-of-Funnel questions (specific, ready-to-buy) than Middle-of-Funnel (broad, exploring). Buyers who already know the space can find you, but you are under-represented at the discovery stage where competitors shape the shortlist.
Apex Environmental sits in the high-perception, moderate-visibility zone: when models are pointed at you they get the core right, but you are not surfacing often enough in unbranded category search, and the description narrows to air quality without a direct prompt. The two gaps share one root cause — thin, air-quality-weighted external signals — so the same fixes move both scores.
| Scenario | What it means | Where to focus |
|---|---|---|
| High perception, low visibility | Accurate when named, but rarely surfaced in category search | Off-site authority, listicles, competitor-alternative content |
| Low perception, high visibility | Surfaced often, but the description is wrong or outdated | Owned-site clarity, structured data, directory corrections |
| Low on both | Weak entity — models barely know you and get it wrong | Foundational entity building: Wikipedia, schema, consistent NAP |
| High on both | Strong AI presence | Defend and monitor; re-run quarterly to catch drift |
Apex Environmental Partners's position: Accurate entity, limited reach. The priority is breadth (tell the full EHS story on owned pages) and off-site presence (listicles, PR, category content) rather than foundational entity building, which is already in decent shape.
Each recommendation is tagged with which score it moves: Perception (P), Visibility (V), or both. Priorities reflect impact against the gaps found above.
Add concise, answer-first summaries to the homepage and each service page, structured as FAQ and How-To blocks with question-and-answer formatting that directly addresses user queries. Models cite easily-extractable owned content, which widens both the named description and the category surfacing.
Get featured in "Best environmental consulting" and "[Competitor] alternatives" articles, which LLMs cite frequently when answering category questions. This is the most direct lever on the visibility gaps where competitors appeared and you didn't.
Standardize listings and implement FAQ, How-To, and Organization schema so retrieval pulls one consistent story. Cross-link related pages to help models understand topic relationships. Fixes the accuracy and source-conflict issues from Part A.
Develop a verified Wikipedia presence and secure mentions in high-authority news and industry publications, key training sources for LLMs. Use uniform phrasing for the brand name and offerings everywhere to strengthen the entity in model embeddings.
Cover category subtopics comprehensively with evergreen, well-researched content, reinforced with consistent messaging across blog, social, and press. Long-lived content raises the odds of inclusion in periodic model updates.
Build a genuine presence on Reddit and similar user-generated platforms that feed LLM training and retrieval, participating with real value rather than promotion.
Track brand mentions across platforms, correct misinformation through feedback mechanisms when surfaced, and re-run this audit each quarter to measure movement on both scores and catch drift as models retrain.
This audit doesn't just diagnose. We generated three ready-to-publish assets for Apex Environmental Partners, each stating the same verified truth at a different layer so AI systems stop guessing. Together they target the exact perception and visibility gaps found above.
| Asset | Layer | What it does & where it goes |
|---|---|---|
| AI Info Page | Prose | Human-readable fact sheet with explicit guidance to AI assistants. Publish as a page on your site (recommended https://apexenvpartners.example/ai) and link it in the footer. |
| EntityMap (json + html) | Structured | 4 entities, 3 typed relationships, and on-page evidence, all attributed to Apex Environmental Partners. Publish both at your domain root. |
| llms.txt | Routing | A curated index pointing AI crawlers to your best pages, including the two assets above. Publish at https://apexenvpartners.example/llms.txt. |
The perception and visibility gaps in this report come mostly from models inferring facts that aren't stated clearly on your site. The AI Info Page states them in plain language, the EntityMap encodes them as structured entities and relationships with evidence, and llms.txt routes crawlers to both. The prose corrects the narrowing; the structure survives chunking with attribution intact; the routing gets your best pages read first.
The same entitymap.json and AI Info Page double as a verified fact sheet for AI drafting. Loaded into an AI workspace built for Apex Environmental Partners, they keep future content grounded in approved facts instead of plausible-sounding fiction, whether or not public crawlers ever read them.
All three follow emerging, self-declared conventions. No AI system is obligated to read them today, and adoption by the major AI platforms is still inconsistent, so we treat them as high-value, low-cost insurance that also serves as an internal grounding tool. Any before/after change in how AI describes you is measured against your own perception and visibility scores on re-audit, not a general claim. EntityMap v1.0 is CC BY 4.0 and published here as a generator-draft asset.
Your fix kit states the truth on surfaces you own. But AI models form their view of you from the whole web, and most of it isn't yours. This map inventories every off-site page that mentions Apex Environmental Partners and sorts each by one question: can you control the wording there? Paired with it is a library of semantic triples — clean subject-predicate-object facts — to deploy wherever you can.
A mention gets your brand named; a mention wrapped in a triple gets it understood. We derived 9 canonical triples from your positioning (the 4 that correct the gaps found above are flagged high-priority), then swept your off-site footprint so you know exactly where each one can land. When the same clean facts show up everywhere a model looks, it stops guessing and starts repeating you.
Search swept: -site:apexenvpartners.example "Apex Environmental" — 12 off-site pages.
| Bucket | Count | What it means & the play |
|---|---|---|
| Company-controlled / editable | 5 | Profiles you can claim and edit today (LinkedIn, Crunchbase, G2, and the like). Your quick wins — drop high-priority triples here first. |
| UGC / influenceable | 3 | Community surfaces (Reddit, Quora, forums). You shape these by participating, not editing. |
| Editorial / third-party | 4 | Independent pages you don't control. They tell you the narrative gap to close and who's worth a pitch. |
Start with the 5 editable profiles: claim or log in, and set the copy to your high-priority triples. That's the fastest, highest-control work in the whole plan. The full sortable inventory — every page, its bucket, a confidence flag, and a deploy note saying which triple to place — is in the accompanying workbook.
Editability is judged from domain and URL patterns, so well-known hosts are classified confidently while unfamiliar domains are flagged low-confidence for manual review. The classification is about who controls the wording, not page quality or sentiment. The SERP sweep reflects results at run time and shifts as the web changes; we refresh it on each quarterly re-audit.
This analysis simulates AI-search behavior at a point in time. Model outputs vary between runs and change as models are retrained; scores are directional indicators, not guarantees. Recommendations follow Tony Salerno Consulting's AI search visibility framework.