The AI answers are naming somebody to your buyers. Find out whether it's you.
This is a fixed-fee audit. It checks whether ChatGPT, Google's AI Overviews and Perplexity recommend you, and whether your own analytics registers the traffic when they do. You also get your standing against the competitors you name for me, and what to fix first.
Run by Eric, who builds and pays for these systems on client sites before selling the method to anyone.
Two blind spots, running at the same time
Your buyers ask ChatGPT, Google's AI Overviews, Perplexity and Gemini for recommendations before they ever reach a page of search results. Most owners have never once checked whether they get named in those answers. Nor which competitor gets named instead.
Meanwhile the analytics that's meant to catch that traffic, and the bookings and enquiries it produces, is quietly misconfigured on a large share of live sites. The booking gets recorded. Nobody ever ticked the box that marks it as a result worth counting, so the conversions report reads zero. A visitor sent by ChatGPT usually arrives with no label saying where they came from, so Google Analytics files them under "Direct". Most sites can see neither half.
Researchers from Princeton, Georgia Tech, IIT Delhi and the Allen Institute for AI tested nine tactics across 10,000 queries to see what actually moves a page into AI-generated answers. Adding specific statistics and citing sources were among the ones that worked, lifting visibility by up to 40%. Most sites aren't pulling that lever, because nobody has told them it exists.
Aggarwal et al., “GEO: Generative Engine Optimization”, KDD 2024. Read it yourself. That is rather the point.
Five failures I find on almost every site
Conversions read zero
The booking fires. The event was never toggled to a Key Event in GA4, so every "conversions" number downstream is wrong.
AI traffic hides in Direct
Perplexity, Claude and referrer-less ChatGPT sessions get bucketed as Direct. The fastest-growing channel is the one you can't see.
Rates invisible to AI
Answer engines are weak at running JS. If your pricing renders in JavaScript they often see an empty page, which on a booking site is the whole product going unread.
A crawler quietly blocked
One line in robots.txt naming Google-Extended or GPTBot can remove you from AI answers. Classic rankings don't move, so nobody notices.
No baseline exists
The business has never measured whether it appears in ChatGPT or Perplexity for its own core buyer questions. There's nothing to improve against, and no way to prove progress later.
The full audit, grouped so you can see what it touches
Every item is checked against your live site and your own analytics.
Tracking & analytics sanity
can you see it- GA4 Key Events configured and firing, verified character-for-character. A trailing space silently breaks the rule.
- Google Tag Manager feeding the right Google Analytics account rather than an old one or a developer's test account. I check it in Admin → Data Streams instead of taking it on trust.
- Whether the tracking code is installed twice, once through Tag Manager and once pasted straight into the page. That doubles every visit and every booking, so the traffic reads roughly double what it is.
- Consent Mode checked for an over-blocking config that suppresses your own tracking
Conversion & booking events
is the money counted- Booking-confirmed event fires once, with value + currency, marked as a Key Event
- Enquiry, quote, and lead-form submits fire and count as conversions
- Click-to-call and email clicks tracked as lead events
- UTM tags checked for consistency across every link, with the first source locked in. A lead that arrived from an AI answer is still credited to that answer in your CRM when the booking finally closes, rather than to whatever they clicked last.
AI answer visibility
can the engines read you- Whether the AI bots are allowed to read your site, checked in both places they get blocked: robots.txt, and your security service. A Cloudflare firewall rule or Bot Fight Mode can turn the bots away before they ever reach the site even when robots.txt says yes.
- Rates, availability & key answers confirmed present in the raw HTML, not injected after JavaScript
- A custom AI-referral channel so ChatGPT / Perplexity / Claude traffic stops landing in Direct
- Pages written so an AI can lift the answer straight off them: the answer in the first 40 words, each FAQ answer complete on its own, facts in tables, and numbers with a source next to them. The audit names the pages that fail and what each one needs. Rewriting them is the Buildout tier.
Schema & technical
does the web know you- Organization schema with legal name, founding date, founders, and a sameAs array to your real profiles
- Product / Offer (or LodgingBusiness for lodging) schema with name, price, aggregateRating
- FAQPage, BreadcrumbList & Person schema with named authorship
- Content freshness: visible last-updated dates, pricing current within ~90 days
Competitive share of voice
who gets named instead of you- A per-engine baseline across ChatGPT, AI Overviews, Perplexity, Gemini, Copilot & Claude, each measured separately. Cross-engine agreement is low.
- A fixed set of 20 to 50 real buyer-intent questions, held constant so every future report is directly comparable
- How often the engines name you, set against the competitors you name for me. Being named first counts for more than being named last, since the first name is the one people act on. One number per engine: your share of all the mentions.
- Citation share (does the engine link your page), sentiment/accuracy, and the source a competitor got cited from, which is your target to go earn
Mostly async, with no standing calls
You send what I need
For the Snapshot: your URL, three competitors, and your buyer questions. For the Audit, add read-only GA4, GTM and Search Console. About 20 minutes.
My software takes the first reading
I built a tool that asks each AI engine your buyer questions and logs which businesses it names. It runs on my side. The questions and their wording stay fixed from one month to the next, so this month's numbers compare directly with last month's rather than depending on whether a person typed the prompts the same way all week.
You get the report
One document: what's broken, what it's likely costing you, a fix list ranked by impact, and your first share-of-voice baseline against the competitors you named.
One 30-min walkthrough
Optional. After that everything runs async. The ongoing monitoring arrives as a generated report.
The person running your audit is the one who built the tooling.
I'm Eric. For close to a decade, from the mid 2000s into the 2010s, I took a crew into Whistler every November. We deep cleaned more than 40 rental properties over two to three weeks, from condos to large homes. Every one of them had been professionally cleaned between guests all year. We were there for what those cleans missed.
Alongside that, I walked properties before guest arrival to confirm they were ready. That was steady work through my years at EMR Vacation Rentals, the company I founded in 2004 and sold to my business partner Blake MacKenzie in 2012. He runs it today and we still work together. So I have spent a long time standing in rentals somebody else had already called clean, writing down what was wrong with them.
I would rather you download a free file and find out whether it fits than take my word for any of this.
Start small, diagnose once, then monitor monthly
The Snapshot is the visibility half of the Audit, cut down. It does not check your tracking. Its fee comes off the Audit if you go on to it within 30 days of the readout. Monitor is priced two ways. If you already pay an agency a monthly SEO or marketing fee, Monitor is 20 to 30% of that fee, kept inside the range below. Otherwise you pay the monthly price.
Visibility Snapshot
- 12 buyer questions across ChatGPT, Google AI Overviews and Perplexity, scored per engine
- Your standing against 3 competitors you name
- A one-page readout and the three visibility fixes I would start with
- Delivered 5 business days after you send your questions and competitors
Visibility Audit
- The full checklist, against your live site, across 6 engines
- Per-engine share-of-voice baseline vs. named competitors
- Tracking & booking-event verification
- Prioritized fix list + report + optional 30-min walkthrough
Visibility Monitor
- Automated monthly re-run, per engine
- Sentiment & accuracy watch + competitive-gap alerts
- Corrected AI-referral channel tracked over time
- Generated monthly report, no standing calls
Visibility Buildout
- Everything in the Audit, then I implement
- Correct Key Events + custom AI-referral channel
- Schema + server-rendered rates where feasible
- Pages rewritten so an AI can lift the answer, and first source locked in your CRM
A retainer is honest here for one reason: AI models change and citations decay month to month, so the measurement has to be continuous. The monthly fee buys that re-run (every engine, the same questions, every month) plus a written report of what moved. You get told when a competitor takes ground, or when an engine starts describing you wrongly. My software does the work, so the price isn't tied to my hours. Your exact number is in writing before anything starts.
The straightforward part
If the full Audit doesn't surface at least five things worth fixing, you don't pay for it. That promise covers the Audit, not the Snapshot. AI citations are volatile and no honest operator will promise you a ranking. You get your standing across every major engine. You get a yes or no on whether the tracking is firing. You get the fix list in priority order. All of it in writing, grounded in your own data.
Send your URL. I'll tell you what I need.
For the Snapshot I need your URL, three competitors and your buyer questions. Nothing else. For the Audit, I reply within 2 business days with the short list of read-only accounts to add me to. Each comes with how to do it. The report is in your hands before you decide anything about ongoing work.