Your AI visibility check is wrong by a factor of six hundred

The single number that changes how to think about AI visibility

Your AI visibility check is wrong by a factor of six hundred if you only ran it on one engine. The same brand can earn hundreds of citations on one AI engine and zero on another. That gap was 615 times in the most recent dataset. Superlines published the number in their March 2026 review of 34,234 AI answers across ten platforms. A citation, in this post, means an AI assistant linking to your site as the source for an answer. Grok came in highest at 27 percent of answers, while Claude sat at the bottom of the dataset and almost never quoted outside sources at all.

This post is for service-business operators who've already typed their own business name into ChatGPT, seen whether it showed up, and decided that was the answer. It was the answer for one engine out of five your customers use. The actual visibility is a row of five different counts, one per engine. The fix is a ten-minute spot check you can run on a Tuesday morning. Here's what the data shows, and what to do this week.

What the 615 times spread looks like on a real business

Superlines tracked 34,234 answers across ChatGPT, Claude, Perplexity, Grok, Google AI Mode, Gemini, and four other engines during early Q1 2026. They counted how often each engine quoted outside sources, and the rates ran across a huge range: Grok at 27 percent of answers, Perplexity at 13 percent, Google AI Mode at 9 percent, ChatGPT under 1 percent, Claude at near zero. Top to bottom, that's the 615 times number.

Translate that to a medspa. The same medspa, with the same Google Business Profile, the same reviews, the same services page, can get quoted in two dozen Grok answers in a month and quoted in zero ChatGPT answers in the same month. Each engine uses its own rule about when to quote outside sources, and the medspa is identical across the row. The customer asking Grok gets sent to the medspa. The customer asking ChatGPT hears a summary that names two competitors and never says the medspa exists. The same business and the same audit produce a very different day on the phone.

The Superlines team put the point in one line: "A brand can be thriving on one platform and invisible on another, and without multi-platform tracking, you would never know."

Why a single-engine audit gives you the wrong answer

Most operators I talk to ran one of these checks last quarter. They opened ChatGPT, typed their business name, and either felt good (named in the answer) or felt sick (not named). That one test told them something true about ChatGPT, and almost nothing about Perplexity, Grok, Claude, or Google AI Mode.

The engine you tested may quote outside sources rarely. A "yes" on a low-citation engine reflects the engine's citation rate first and your business second. ChatGPT quotes outside sources in fewer than 1 of every 100 answers. Most of its answers come from training data, which is a frozen snapshot of the websites it already read, months or years back. So a "yes I show up" on ChatGPT is a thin signal: the engine almost never quotes outside sources to begin with, so a positive result there only tells you you're in the small set of businesses ChatGPT happens to mention. When I run the same five-engine check on a medspa, the Grok column is usually a 3 or a 4 while the ChatGPT column sits at zero. The medspa is doing everything right, and ChatGPT just rarely quotes anyone in the medspa category in the first place.

Then there is the engine your customer actually used. The daughter looking for a dentist for her father is on Perplexity. A handyman pricing his next job runs the question through ChatGPT. The 22-year-old shopping for a medspa is on Grok or Gemini. Your phone rings or doesn't ring on the strength of whichever engine your buyer used. Most operators tested a different one.

The ten-minute, five-engine spot check

Pick a Tuesday morning. (Tuesday because the phone's quiet and you've got coffee.) Open five tabs. Go to ChatGPT, Claude, Perplexity, Grok, and Google AI Mode. In each tab, type the same customer question. Use the words a customer types into the search bar. Industry shorthand will not match what the buyer asked. A dental practice can try "best dentist for crowns under $1,200 in my area." An HVAC contractor might pick something like "honest furnace repair near me for an old furnace." For a medspa, a query like "where to get Botox near me with good reviews" usually returns the same set of competitors a real shopper sees.

Read the answer on each engine. Count how many times your business is named, and how many times your site is cited as a source. Write the five counts in a row on a piece of paper. That row, with five different numbers, is your real visibility today. Anyone giving you a single-number answer is reporting one column of a five-column table.

Most operators find an uneven row: maybe a 4 on Grok and a 2 on Perplexity, but zeros on Google AI Mode, ChatGPT, and Claude. The fix is to start with the zero columns. A zero on Perplexity or Google AI Mode is almost always a gap on your services page or your Google Business Profile (GBP) that you can close in an afternoon. A zero on ChatGPT is harder, and it's slower, because ChatGPT cites outside sources so rarely; the fix there leans on earning mentions in third-party places that the model has already read.

What to fix once the spread is visible

The fix is the same set of small actions, sorted by which engine is failing. Google AI Mode, Gemini, and to a lesser extent Perplexity all read your Google Business Profile and your services page directly, so start with the page copy. List your services with the customer-side phrasing. "Same-day crowns" beats "advanced restorative dentistry" on every engine that reads your services page. Put your hours in plain text on the page. Publish prices in plain text on the page itself, visible to a reader. Contact-form-gated pricing is invisible to every engine that reads your page. AI engines that lean on Google for facts will pick those changes up within two to four weeks.

The Perplexity column moves with the same actions. BrightEdge's Perplexity citation research found that Perplexity offers about 5.28 citations per answer, and about 60 percent of those citations overlap with the top 10 Google results for the same question. So whatever lifts you on Google lifts you on Perplexity at about the same rate. Fresh reviews on your Google profile help here too, because Perplexity quotes review snippets when the answer is a recommendation.

There is a quieter version of this risk on Google itself. BrightEdge's March 2026 AI Overview impact report tracked organic click-through rates and found that when an AI Overview shows up at the top of the page, organic clicks drop from 1.76 percent to 0.61 percent. That is a 65 percent drop. The queries where you used to win on Google are now getting answered above your listing, so being inside the AI summary on those queries is the better piece of real estate.

ChatGPT and Claude are the slow ones. They quote outside sources rarely, so getting into their answers means earning mentions on the third-party pages they did read: a clear LinkedIn page for the owner, a Reddit presence where customers in your category are asking questions, a small set of trade-press mentions or interviews if you can earn them. Two short useful answers per week in the right Reddit community, for two months, has gotten more operators into the small pool of brands ChatGPT and Claude will quote than any paid SEO play I've seen this year.

What to do after you see your row of counts

Once you have the row of five counts, you have the real picture and you can start fixing the gaps the check exposed. The row will keep shifting as engines update their rules every few weeks. Plan to redo the check every month or so. When you run the row the second time, the real diagnostic is which columns lifted; the ones that moved tell you the work you did is reaching that engine, and the ones that stayed flat tell you the engine wants a different kind of signal you have yet to supply. Write the counts down on paper each month and pick next month's column off the row.

Common questions about multi-engine AI visibility

Why would the same business show up on one AI engine and not another?

Each AI engine has its own way of picking which sources to quote. Grok pulls heavily from public social posts. Perplexity leans on the same top results that Google ranks. ChatGPT and Claude quote outside sources far less often. So the same business can have a strong presence on a forum that Grok reads and almost nothing in the small set of pages ChatGPT will quote.

How do I run a five-engine spot check on my own in ten minutes?

Open five tabs. Go to ChatGPT, Claude, Perplexity, Grok, and Google AI Mode. Type the same customer question into each one. Read the answer and count how many times your business gets named or your site gets cited. Write down the five counts next to each other. That row is the real picture of your visibility today.

Which engine should I focus on first if my counts are uneven?

Start with the engine that has zero. A zero count is almost always a fixable gap on your Google Business Profile (GBP) or your services page. Engines like Perplexity and Google AI Mode read the same pages Google ranks, so cleaning up your service names, your hours, and your most recent reviews lifts both at once.

Is the 615 times gap going to shrink as the engines mature?

Not in the next six months. The gap is built into how each engine answers questions. Grok keeps quoting forums. Claude keeps almost never quoting outside sources. The gap between these engines might change a bit over time, but it's baked into how they're built and isn't going anywhere soon. Plan to keep monitoring all five engines as long as customers use more than one.