Nobody is scrolling to page two. They're reading one answer.
We get retail and service brands named inside ChatGPT, Gemini, Perplexity and AI Overviews. The moment where a customer decides where to go, before they ever see a search result.
Get a free visibility checkWatch how it worksThree real prompts, run live on this page, with buttons to verify each one in ChatGPT, Claude, Perplexity and Google. No call, no email required.
Published pricesMonth to monthOne client per category per metroYou own the work
9:41●●●
AI assistantToday
what's the best dispensary in Pacific Beach for a first-time buyer?
Here are the top options for a first visit in Pacific Beach:
Competitor Ahas the deepest review history and shows up in most local roundups.
Competitor Bis mentioned in a few forum threads and one local guide.
Competitor Canswers first-timer questions right on its website.
Want directions to any of these?
Your store has been open eleven years. It is not in this answer.
Illustrative. Not a real AI output.
Ask anything
The engines we track and the platforms we plug into
ChatGPTTrackedGeminiTrackedPerplexityTrackedGoogle AI OverviewsTrackedClaudeVerifyCopilotTrafficGoogle AI ModeVerifyYouTubeTranscriptsChatGPTTrackedGeminiTrackedPerplexityTrackedGoogle AI OverviewsTrackedClaudeVerifyCopilotTrafficGoogle AI ModeVerifyYouTubeTranscripts
Google Business ProfileYelpWeedmapsLeaflyDutchieShopifyWordPressSquarespaceWixGoogle AnalyticsRedditApple MapsGoogle Business ProfileYelpWeedmapsLeaflyDutchieShopifyWordPressSquarespaceWixGoogle AnalyticsRedditApple Maps
When a customer asks ChatGPT or Gemini where to go, the answer names three businesses. We find the questions your customers ask, fix what the models read, and earn the mentions that get you named first. Then we prove it, every week, in numbers you can check yourself.
Keep scrolling. The page is the pitch.
Three steps to get started
Twenty seconds to your first real answer.
No call, no card, no deck. You type, the engines answer, you see where you stand.
Business and city
Wellgreens, Pacific Beach
Run my check
1
Type your business
Your name and your city. We pick three questions a real customer would ask, or you write your own.
best dispensary in Pacific Beach
ChatGPT First
Gemini Named #2
Perplexity Absent
2
Read the answers
Word for word, with your name highlighted and a verdict on each: first, named, cited, or absent. Plus who got named instead.
Share of answer1824313947556066%
W1W2W3W4W5W6W7W8
3
Watch the share climb
Become a client and the same questions run every week, per location, in a dashboard you can verify against the real apps.
Ranking first stopped being the same thing as being recommended.
An answer engine names three businesses. No second page, no map pack, no scrolling. You're one of the three or you're invisible, and most owners have never checked which.
85%
of brand mentions in AI answers come from third-party pages, not from a company's own website.
AirOps citation research
62%
of the time an AI uses your page as a source, it never names your brand in the answer at all.
2026 study, 3,981 domains
2 to 6×
higher conversion reported on leads arriving from AI answers than from any other channel.
Reported by AI-referral tracking vendors
This is why bolting "AI" onto an SEO retainer does not work. Schema and FAQ blocks are the cheap half of the job. The half that changes the answer happens off your website. In the reviews, roundups, forums, transcripts and local coverage the models actually read.
Four engines. Four different ideas of who to recommend.
Tracked
ChatGPT
Leans on third-party coverage and review depth.
Reviews & roundups
Forum threads
Local press
Tracked
Gemini
Closest to Google's own index. Entity signals matter most.
Business profiles
Entity & schema
YouTube transcripts
Tracked
Perplexity
Cites its sources openly. Quotable pages get quoted. And linked.
Comparison pages
Question pages
Fresh coverage
Tracked
AI Overviews
Sits above every Google result. Video correlates strongest.
First-forty-words answers
Video mentions
Map & profile data
We track all four, per location, per question.02Next: Proof
02Proof
We ran it on our own stores first. Here's what came back.
Live numbers from the latest weekly run on Wellgreens, the San Diego chain we come from. Nothing is picked to look good: the weak locations are in here, and so is a prompt we lost.
Loading the latest run…
Where we're named, question by question.
Each pin is one question at one location. The number is where we're named in the answer.
1Named first2Named second3Named third or later×Not named in most answers
Pick your industry and city. This is the shape of the prompt set you get on day one, grouped by what the customer is trying to decide. Every prompt opens in the real apps.
20 prompts · Dispensaries in San Diego Re-run weekly across four engines once you're a client
Run three of these now. FreeCurated set. The AI button writes a fresh set for any industry, including ones not in the list.
The same question, ninety days apart.
Three things we score on every answer. Toggle the example to see what the shift looks like.
NamedYour brand appears in the answer.
First mentionYou're the first business the model names.
CitedYour page is used as a source, with a link.
Illustrative example. Not a client result.
The full prompt libraryEvery client gets 60 to 120 of these, per location, re-run weekly.04Next: Watch
04Watch
Three short videos on what we do and why it works.
Three explainers, each under two minutes, each answering one question.
1:25
What is answer engine optimization?
Why "ranking" stopped mattering the day customers started asking a chat window instead of scrolling a results page.
YouTube85 seconds
1:39
How we get a brand named in 90 days
The four stages, what happens off your website, and how you check every claim yourself in the same chat window.
YouTube99 seconds
1:34
Why YouTube is the channel AI reads most
Transcripts, chapters and long descriptions. Why a phone-shot store walkthrough gets quoted and a polished ad doesn't.
YouTube94 seconds
The most-cited channel in AI answers is the one you don't have.
In January 2026 YouTube passed Reddit as the most-cited social source in AI answers. Almost no local retail brand has noticed.
0.737
correlation between YouTube brand mentions and AI visibility. The strongest single signal measured across 75,000 brands. Backlinks scored 0.218.
Ahrefs AI Overviews study
16%
of all LLM answers now draw on YouTube, against 10% for Reddit. And 18× what Instagram earns.
Adweek, 6.1m citations across four research firms
41%
of AI-cited videos had under a thousand views. Audience size is not what gets a video quoted.
Otterly YouTube Citation Study 2026
All three videosEvery video we publish is transcribed, chaptered and described so the models can read it too.05Next: How it works
05How it works
Four stages. You see the prompt set on day one.
Everything we do is checkable by you, in the same chat window your customers use. If you can't verify it yourself, we didn't do it.
STAGE 01
Map the questions
Week 1
We build the 60 to 120 prompts your customers actually type, per location and per category, then run them across four engines to record who gets named, cited, and first.
STAGE 02
Fix what the models can read
Weeks 2 to 3
Entity and schema work, location pages that answer a question in the first forty words, consistent naming across every profile, and a clean crawl path for AI user-agents.
STAGE 03
Earn the off-site mentions
Ongoing, month 1 forward
The part that moves the answer: local roundups, review depth and recency, community threads, video coverage, and comparison pages where your name sits next to the competitors you're losing to.
STAGE 04
Report on share, not traffic
Every month
One number per location: how often you're named, and whether you're first. Plus the raw answers, so you can read what the model said about you word for word.
The full walkthroughAudit in 7 days. Build in 10 working days. Off-site work from month one.06Next: Your dashboard
06Your dashboard
How you'll know it's working.
Every client gets a live dashboard per location. One headline number, share of answer: how often you're named, first, and cited. AI-referred visits sit beside it.
Share of answer
.
First mention
.
Cited as source
.
AI-referred visits
.
Share of answer, weekly 12 weeks · all engines
Prompt
ChatGPT
Gemini
Perplexity
AI Overviews
First named firstNamed in the answerCited used as a sourceAbsent not mentioned
Illustrative sample data, not a client result. Client dashboards use weekly prompt runs and your own GA4 referral data.
Already a client? Your live dashboard is one click away.Client login
Google Analytics buries it under "referral". One line of code counts every visit from ChatGPT, Perplexity, Gemini, Copilot and Claude, and shows which pages they land on.
1Tell us your website and where to send the report link.
2Paste one script tag into your site, send it to whoever manages it, or ask us to do it for you. Works with Shopify, Squarespace, Wix, WordPress, Dutchie, anything.
3Watch the report fill in. AI visits, share of all visits, sources, landing pages. Updated live.
What it records, and what it doesn't. It records the referring site, the page visited, and the day. It sets no cookies and collects nothing about the visitor. No names, no emails, no device IDs, no IP addresses stored. We see counts, not people. Remove the tag and it stops.
Don't want to touch your site? We'll install it for you. Same tag, same rules. We never get access to your website, your analytics or your customer data; we place one line of code with whoever manages the site, and the counts push straight into your client dashboard next to your share-of-answer numbers.
Your tracker is ready
Paste this just before </head> on every page (most site builders have one "header code" box that does this):
Find out whether ChatGPT, Perplexity, Gemini or Copilot are already sending people to your website. One line of code, a live report, no account needed. We'll install it for you if you like; we never get access to your data, the counts just push to your dashboard.
If your share of answer hasn't risen 90 days after the build, the monthly retainer is free until it does. Measured on the prompt set we documented on day one, in your dashboard, on numbers you can verify yourself in ChatGPT and Gemini. No small print about traffic or rankings; the answer either names you more often or you don't pay.
Nothing to migrate. One line of code if you want the tracker, and only then. The engines, the site builders, the profiles and the languages are all on one page.
Plus Webflow, Jane, Leafly, Apple Business Connect, Bing Places, Google Analytics 4, Reddit and any site with a header box.
Prompts in your customers' languages
70+
🇺🇸 🇲🇽 🇵🇭 🇻🇳 🇨🇳 🇰🇷 🇸🇦 🇧🇷 🇫🇷 🇩🇪 🇮🇳 🇯🇵
English runs today for San Diego; Spanish, Tagalog and Vietnamese are next. The engines answer in whatever language they're asked, and our scoring works the same in all of them.
Your business and city. Three real questions through ChatGPT and Gemini, web search on, actual answers. Nothing to book.
Your answers will appear here
Each question is asked the way a customer would ask it, in your city, with web search on. You'll see the engine's reply with your name highlighted, plus who got named instead.
FAQ
Questions we get before the first call.
What is answer engine optimization?
Getting your business named. Ideally first. When someone asks ChatGPT, Gemini, Perplexity or Google's AI Overviews a question you should be the answer to. It's measured in share of answers, not rankings or traffic.
How is this different from SEO?
SEO earns you a position on a results page. Answer engines don't show a page. They name a handful of businesses. Most of what decides that happens off your website, in reviews, roundups, forums, video transcripts and local coverage. Schema and FAQ blocks are the cheap half of the job.
How do I know the work is real?
You see the full prompt set on day one and can run any prompt yourself, any evening, in the same chat window your customers use. Every monthly report includes the raw answers so you can read what the model said about you word for word.
How do I know the numbers are real?
Five ways, all built in. Every prompt has one-tap buttons that open it in ChatGPT, Claude, Perplexity and Google AI Mode so you can compare the real app's answer to ours. Every stored answer lists the sources the engine read, with the model version and the time stamped on it. Each prompt is asked three times a week per engine, so you see a rate ("named 2 of 3"), not one lucky sample. You can re-run any prompt live from the dashboard while you watch. And you or your team can log what the real app showed; the dashboard keeps a running agreement score between the apps and our data. Full methodology.
How will I know it's working?
You get a live dashboard per location. It shows share of answer. How often you're named, named first, and cited across the prompts that matter. Updated from weekly prompt runs, next to AI-referred visits and form fills from your own analytics. Every raw answer is kept, so you can read what the model said.
Which engines do you track?
ChatGPT, Gemini, Perplexity and Google AI Overviews. Prompts are re-run weekly and reported monthly, per location.
Do I have to sign an annual contract?
No. The audit and the build are fixed-scope, one-time fees. The monthly authority retainer is month to month and can be cancelled with 30 days' notice. You own everything we build.
Will you work with my competitor too?
No. We take one client per category per metro. If we're working with a dispensary in San Diego, we won't take another one.
How long before the answer changes?
The audit ships in 7 days and the build in 10 working days. Off-site mention work starts in month one, and the before/after comparisons we show are framed around a 90-day window. Individual results vary by market and category.
Where do the prompts come from?
There's no public search-volume data for AI prompts, so we don't invent any. The set is our read on which questions actually produce a recommendation: discovery, first-timers, comparisons, price and deals, and logistics like hours or delivery, each tied to a location. You see the whole set on day one and every prompt opens in the real apps.
Why do you push video so hard?
Across 75,000 brands, YouTube mentions had a 0.737 correlation with AI visibility, the strongest single signal measured; backlinks scored 0.218. Subscribers made no difference, description length was the strongest metadata signal, only a third of cited videos had chapters, and 94% of citations went to long-form video. Foundational video datasets were built on YouTube transcripts, so models already treat it as a knowledge source. These are correlations, so we treat video as a strong bet with cheap downside, not a guarantee. For a multi-location retailer it's unusually winnable: a store walkthrough filmed on a phone, transcribed and chaptered, needs no audience.
Why should I trust an agency that came out of a dispensary?
Because we had the problem before we had the service. We watched customers arrive already holding a recommendation and realised it wasn't ours. Every tactic is tested across our own locations before a client is billed for it, and our own live numbers sit on this page, weak locations included. We also know what a review programme does to staff morale and why a deliverable that needs six hours of the owner's time never gets done, so the plan is written around the operation, not a reporting template.
What does the free visibility check include?
You run it yourself, on this page, in about twenty seconds: three real questions for your business and city, asked to ChatGPT and Gemini with web search on. You see the actual answers with your name highlighted. Leave an email and the same results land in your inbox within a minute, with a link you can send to a partner. Ask and we will run the deeper ten-question, four-engine version by hand. No call attached either way.
ChatGPTTracked
GeminiTracked
PerplexityTracked
Google AI OverviewsTracked
ClaudeVerify
CopilotTraffic
YouTubeTranscripts
Google Business Profile
Yelp
Weedmaps
Leafly
Dutchie
Shopify
WordPress
Squarespace
Wix
Google Analytics
Reddit
Apple Maps