This is our first AI visibility benchmark for DC-area consumer and B2B brands. AI visibility means how often AI assistants name your brand when buyers ask for recommendations.
In April 2026 we ran a group of DC-area brands through the same scan on four AI assistants. We asked each one a fixed set of buyer questions. The four were Perplexity, Claude, and ChatGPT with web search on and off. Model details are in the Method section below.
The headline finding: most of the brands never came up on a general category question. Every time they were named, the buyer had asked for them by name. When buyers asked "what are the best [category] options", the question that leads to discovery, most of these companies were not named at all.
Disclosure: Envoyra sells the AI visibility scanning tool used to collect this data. We have a financial interest in findings that show the value of AI visibility measurement. The method is set out in full below. Underlying scan data is available to clients and press on request.
Who AI named
We scanned these brands: Zevia, Better Times Market, URGO Beauty, Persica Skincare, Elodie's Naturals, KarmaStaff and Envoyra itself.
Zevia stood apart. It was named in most answers on every AI assistant we tested. Zevia is sold nationally, so we treat it as an outlier among the local brands.
Better Times Market came next, a clear step behind Zevia. The rest of the group, including Envoyra, were named only occasionally.
What Zevia's content looks like
Zevia's lead is not luck. Its content is specific in ways that line up with the questions buyers ask.
When a buyer asks Perplexity for the "best zero sugar carbonated drinks without aspartame," Zevia is named. Zevia's content spells out the sweetener (stevia). It names what is left out: no aspartame, no artificial colours, no synthetic dyes. It names who the drink is for: people with diabetes, health-conscious buyers, children avoiding sugar. And it names what to compare it with: LaCroix and regular diet soda. Every piece of the buyer's question has a matching fact on Zevia's pages.
Outside research points the same way. A Princeton University study presented at KDD 2024 introduced the idea of Generative Engine Optimization. It found that adding statistics, citing trusted sources and stating evidence plainly made content more visible in AI answers. Zevia was not writing for AI. But its content happens to have those same traits.
The category question problem
The most revealing finding sits below the headline. It is about which kinds of questions brands were named on.
For most brands, every AI mention came from a question that used the brand's name: "Is [brand] legit?", "[Brand] vs competitors," or a search for the company itself. Those questions only come up when a buyer already knows the brand. They add nothing to discovery.
Category questions ("best [product type] for [use case]," "top [industry] tools," "recommend a [category] solution") did not name most of these brands on Perplexity. Those are the questions that bring in new customers. Buyers who have never heard of your brand are asking them right now, and AI is naming your competitors.
The channel is already large. Gartner predicted in February 2024 that traditional search volume would fall by 2026 as people turn to AI assistants for answers. Perplexity alone handled 780 million questions in May 2025, up from 230 million in mid-2024. For most DC brands in this study, that channel returns nothing.
Each AI assistant answers differently
One of the clearest findings is how differently each AI assistant treats the same brand.
Better Times Market did better on ChatGPT with web search off than on Perplexity, which searches the live web. That is the reverse of what most brands expect. One possible reading: Better Times Market has been covered in enough food and drink publications to show up in what ChatGPT learned in training. Fewer of those publications show up among the sites Perplexity cites.
URGO Beauty showed the opposite pattern. It did better on Perplexity than on Claude. URGO's niche is very specific: makeup for equestrians, and non-pore-clogging formulas for athletes. Perplexity's answers, which cite many live sources, surfaced that niche more often than Claude's did.
The practical point: treating AI visibility as one number hides real differences. Each AI assistant draws on its own sources. Assistants that search the live web can reflect new content sooner. Assistants that answer from training change on a slower cycle. What moves one may not move another.
What Perplexity was citing instead
For the DC brands in this study that were not named, here is what Perplexity cited.
In AI visibility (Envoyra's own category): tryprofound.com, semrush.com, thedigitalelevator.com and minuttia.com, each cited more than once. These are the sites Envoyra and similar firms compete with for a place in the answer.
In wellness and beauty: comparison sites, ingredient-focused review databases and directories run by publications. These are sources outside the brands' own websites that vouch for the brands.
A 2025 Semrush study of AI citations found that Reddit, Wikipedia and YouTube were the three most-cited websites across AI assistants. The sites AI cites are mostly outside sources, not brand websites. A brand that only works on its own site is working on just one part of the picture.
Three things worth testing
Based on this scan and current outside research, here are three things emerging DC brands could test. Each one can be measured in later scans to see whether it helped.
1. Get listed on structured directories. Crunchbase, G2, Clutch, Futurepedia and similar sites are outside sources that confirm a brand exists in its category. A 2025 analysis from Nobori found that brands with active review-site profiles were about three times as likely to be cited by ChatGPT. It costs little and takes an afternoon.
2. Rewrite two pages with plain, specific use cases. The gap between Zevia and the rest of the group lines up with specificity, not content volume. Pick your two busiest product or service pages. Rewrite the opening to answer: Who exactly is this for? What exactly does it replace? What exact problem does it solve? Vague positioning gives an AI assistant nothing to match a buyer's question to.
3. Publish one piece of original research. The Princeton/KDD study linked statistics and cited sources in content with more AI visibility. The simplest version: write up what you see in your own industry, include real numbers and publish it on your site. Original research tends to be cited by other sites.
What this means for measurement
Most brands in this study were measuring their AI visibility for the first time. That is the most common situation we see. Brands active on Google and social for years have never checked whether AI recommends them to buyers.
In most cases, it doesn't, at least not on the questions that lead to discovery. Questions that name the brand register. Category questions do not.
This can change. AI assistants that search the live web can pick up new content. Assistants that answer from training change more slowly. Brands that start measuring now can see whether what they change shows up in the answers.
Brands that wait are not standing still. They can't see whether the gap between them and their competitors is growing.
Method
Scan date: April 2026. Brands: Better Times Market, Elodie's Naturals, Envoyra, KarmaStaff, Persica Skincare, URGO Beauty, Zevia. AI assistants: Perplexity sonar-pro (searches the live web and always cites sources), Claude Sonnet 4 (answers from training, no citations), ChatGPT gpt-4o with web search (sometimes cites sources), ChatGPT gpt-4o without web search (answers from training, no citations). Buyer questions: the same fixed number of buyer questions for every brand on every AI assistant, written by Claude from each brand's intake details (industry, buyer, keywords, problem solved). Measure: the number of buyer questions where the AI assistant's answer named the brand. Brand detection: we checked each answer's text for the brand name, ignoring case.
A single scan gives an early read, not a final answer. For brands named only occasionally, treat one scan as a floor; scanning again on a different day gives slightly different results. Track for four weeks or more before drawing conclusions.
Outside sources cited: Gartner (Feb 2024): forecast for traditional search volume; Princeton/KDD 2024: GEO paper (arXiv:2311.09735); Business of Apps (2026): Perplexity user statistics; Semrush (2025): study of websites cited by AI; Nobori (2025): AI visibility statistics.
Read the DC-area benchmark
Most brands assume they show up when buyers ask AI for recommendations. Many don't.
This benchmark shows which DC-area brands Perplexity, Claude and ChatGPT named, and how the named brands differ from the rest.