Research · May 2026
DC-Area Brand AI Visibility Benchmark: April 2026
7 brands. 4 AI engines. 336 prompt-response pairs. Five of seven brands are invisible to buyers asking category questions — and most have no idea.
This is the first cross-brand AI visibility benchmark for DC-area consumer and B2B brands.
We ran 7 brands through a standardized 4-engine scan in April 2026: 12 buyer-intent prompts per engine across Perplexity (sonar-pro, live web), Claude Sonnet 4 (training data), ChatGPT Search-on (gpt-4o + web_search_preview), and ChatGPT Search-off (gpt-4o, training data only). 336 total prompt-response pairs analyzed.
The headline finding: five of seven brands have zero category wins. Every appearance in AI responses is a direct name search. Buyers asking "what are the best [category] options" — the query that drives discovery — return zero Envoyra brand results for most of the companies in this study.
0 | Category wins for 5 of 7 brands — AI engines know their names but cannot classify them
Disclosure: Envoyra sells the AI visibility scanning tool used to collect this data. We have a financial interest in findings that demonstrate the value of AI visibility measurement. The methodology is disclosed in full below. Underlying scan data is available to clients and press on request.
The scores
All results are Share of Model: the percentage of 12 buyer-intent prompts where the brand appeared in the AI engine's response.
| Brand | Perplexity | Claude | ChatGPT Search-on | ChatGPT Search-off | Avg | Tier |
|---|---|---|---|---|---|---|
| Zevia | 83.3% | 83.3% | 75.0% | 83.3% | 81.2% | Market Leader |
| Better Times Market | 41.7% | 33.3% | 33.3% | 50.0% | 39.6% | Strong Challenger |
| URGO Beauty | 33.3% | 16.7% | 16.7% | 16.7% | 20.9% | Emerging |
| Persica Skincare | 16.7% | 16.7% | 16.7% | 16.7% | 16.7% | Emerging |
| Elodie's Naturals | 16.7% | 8.3% | 16.7% | 16.7% | 14.6% | Emerging |
| KarmaStaff | 16.7% | 8.3% | 16.7% | 16.7% | 14.6% | Emerging |
| Envoyra | 16.7% | 16.7% | 16.7% | 16.7% | 16.7% | Emerging |
Category averages (excluding Zevia as a nationally distributed outlier): 20.5% average across 6 DC-area brands.
Tier definitions: Foundation (<5%), Emerging (5–20%), Strong Challenger (20–35%), Market Leader (35%+).
Why Zevia is in a different category
Zevia's 81.2% average is not luck. It is the result of years of structured, specific content that gives AI engines exactly what they need to generate a confident recommendation.
When a buyer asks Perplexity "best zero sugar carbonated drinks without aspartame," Zevia appears because Zevia's content explicitly names: the sweetener (stevia), the absence (no aspartame, no artificial colors, no synthetic dyes), the use case (diabetics, health-conscious buyers, children avoiding sugar), and the comparison frame (vs. LaCroix, vs. regular diet soda). Every element an AI needs to match a buyer's query to a brand recommendation is present, structured, and findable.
The research confirms this pattern at scale. A Princeton University and KDD 2024 study introducing GEO (Generative Engine Optimization) found that adding statistics, citing authoritative sources, and structuring content with explicit evidence language boosts brand visibility in AI engine responses by up to 40%. Zevia didn't optimize for AI — but its content architecture happens to satisfy the same structural requirements.
40% | Visibility lift from structured GEO content tactics, per Princeton/KDD 2024 research
The category wins problem
The most revealing finding is not the headline score — it is what type of prompts brands win on.
For five of seven brands, every AI appearance is a branded query: "Is [brand] legit?", "[Brand] vs competitors," or direct company name searches. These queries only fire when a buyer already knows the brand exists. They contribute nothing to discovery.
Category queries — "best [product type] for [use case]," "top [industry] tools," "recommend a [category] solution" — returned zero mentions for five brands on Perplexity. This is the exact query category that drives new customer acquisition. Buyers who do not yet know your brand exist are asking these questions right now, and AI is giving them your competitors' names.
This is not a small gap. Gartner predicted in February 2024 that traditional search engine volume will drop 25% by 2026 as AI assistants become primary answer engines. Perplexity alone processed 780 million queries in May 2025, up from 230 million in mid-2024. The channel is already material — and for most DC brands in this study, the channel returns nothing.
780M | Perplexity queries in May 2025 — up 239% from mid-2024. Source: Business of Apps, 2026
Engine divergence: not all AI is the same market
One of the clearest findings from the 4-engine scan is how differently each engine treats the same brand.
Better Times Market scored 50.0% on ChatGPT Search-off (training data) but 41.7% on Perplexity (live web). This is the reverse of what most brands expect — Perplexity is the live-web engine and should update faster. The explanation is training data depth: Better Times Market has been covered in enough food/beverage publications to have meaningful training corpus representation, but fewer of those publications are in Perplexity's active citation pool.
URGO Beauty shows the opposite pattern: 33.3% on Perplexity versus 16.7% on Claude. URGO's niche specificity (equestrian makeup, non-comedogenic formulas for athletes) is exactly the type of differentiated query Perplexity's citation-dense responses surface, while Claude's training data contains far less equestrian-specific review content.
The practical implication: treating AI visibility as a single score is wrong. Each engine is a separate citation market with different source dependencies. Perplexity cites live web content within 3–14 days of indexing. Claude and ChatGPT Search-off reflect training data accumulated over 6–12 months. The strategies for moving each engine are different.
3–14 days | Perplexity's indexing window for new content — the fastest route to new AI citations. Source: Primary Position, 2025
What the Perplexity citation pool looks like
For the DC brands in this study that did not appear, Perplexity was citing:
In the GEO / AI visibility space: tryprofound.com (3×), semrush.com (2×), thedigitalelevator.com (2×), minuttia.com (2×). These are the brands occupying the citation pool that Envoyra and other GEO firms compete for.
In the wellness / beauty space: structured comparison sites, ingredient-specific review databases, and publication-backed brand directories. Not the brands' own websites — third-party validation of the brands.
Semrush's 2025 study analyzing 150,000+ AI citations found that Reddit, Wikipedia, and YouTube are the three most-cited domains across AI engines — with YouTube alone cited in 23.5% of AI responses. The implication for brands: the citation pool AI draws from is dominated by third-party validation sources, not brand websites. Brands that invest only in their own site are investing in the wrong layer.
The three highest-leverage actions
Based on the scan data and current GEO research, these are the three actions with the highest expected ROI for emerging-tier DC brands:
1. Get listed on structured third-party directories immediately. Crunchbase, G2, Clutch, Futurepedia, and similar platforms are indexed by Perplexity within days of creation. They provide the third-party entity validation AI engines use to confirm a brand is a legitimate category participant. Analysis from Nobori (2025) found that brands with active review platform profiles have approximately 3× higher chances of being cited by ChatGPT than brands without them. This is a zero-cost, 2-hour action.
2. Rewrite two pages with explicit use-case language. The gap between Zevia (81%) and the emerging brands (14–17%) is not content volume — it is specificity. Pick the two highest-traffic product or service pages and rewrite the first 300 words to answer: "Who specifically is this for? What specifically does it replace? What specific problem does it solve?" AI engines need a reason to recommend your brand over a generic answer. Vague positioning gives them no reason.
3. Publish one piece of original research. The Princeton/KDD GEO paper found that citing authoritative sources and including statistics in content correlates with up to 40% higher AI visibility. The simplest version: document what you observe in your own industry, put real numbers in it, and publish it on your site. Research content earns third-party citations — which is the mechanism by which AI visibility improves.
What this means for measurement
Five of seven brands in this study are measuring their AI visibility for the first time. That is the most common situation we encounter. Brands that have been active on Google and social for years have never measured whether AI recommends them to buyers.
The answer, in most cases, is that AI does not — at least not on the queries that matter for discovery. Branded queries register. Category queries do not.
This is not a permanent state. Perplexity can pick up new content within days. ChatGPT Search indexes PR Newswire and structured citation sources within weeks. Claude and ChatGPT Search-off move on 6–12 month training data cycles. The brands that act now are not just fixing a gap — they are building the citation infrastructure that compounds over time.
Brands that wait are not standing still. They are watching the gap between their AI footprint and their competitors' widen every month.
Methodology
Scan date: April 2026. Brands: 7 (Better Times Market, Elodie's Naturals, Envoyra, KarmaStaff, Persica Skincare, URGO Beauty, Zevia). Engines: Perplexity sonar-pro (live web, 100% citation density), Claude Sonnet 4 (training data, 0% citation density), ChatGPT gpt-4o + web_search_preview (selective retrieval, ~21% citation density), ChatGPT gpt-4o base (training data, 0% citation density). Prompts: 12 buyer-intent prompts per brand per engine, generated via Claude using brand-specific intake (industry, buyer persona, keywords, problem solved). Scoring: Share of Model — % of prompts where brand name appeared in AI response. Brand detection: case-insensitive substring match on raw response text.
Single-shot scores are directional signals. For emerging-tier brands (14–20%), scores should be interpreted as floor estimates — re-scanning on different days produces variance of ±2–4pp. Trend measurement over 4+ weeks is recommended before drawing conclusions.
Industry sources cited: Gartner (Feb 2024) — traditional search volume forecast; Princeton/KDD 2024 — GEO paper (arXiv:2311.09735); Business of Apps (2026) — Perplexity user statistics; Semrush (2025) — AI citation domain study; Nobori (2025) — AI visibility statistics; Primary Position (2025) — Perplexity indexing speed.
Read the DC-area benchmark
Most brands assume they show up when buyers ask AI for recommendations. Most don't.
This benchmark documents which DC-area brands are appearing — across Perplexity and Claude — and what structural differences explain the gaps.