How NIXAR checks what AI says about your business

1. 01

   A fixed list of questions, written the way an owner types them

   NIXAR’s list has 25 questions: discovery (“who can help me with this”), problem (“why am I not showing up”), cost (“what drives the price”), brand and decision questions. The list stays stable from week to week. We change at most three questions a month and log each change, so a move in the answers is not just a move in the wording.

2. 02

   Four assistants, every week

   Each question goes to ChatGPT, Gemini, Perplexity and Claude. Each assistant reads a different mix of sources, so the same question can name different businesses in each.

3. 03

   A fresh chat for every question

   We open a new chat for each question, signed out or in a temporary chat. An earlier message in the same chat, or an account’s saved memory, can change what an assistant says, and then the answer no longer shows what a stranger would see. Answers from a long thread or a signed-in account are labeled as such and kept out of the headline number.

4. 04

   Brand questions in their own chat

   Questions that name NIXAR, such as “what is NIXAR Solutions”, are asked in a separate chat. Once a name appears in a chat, the assistant keeps using it, so a brand question asked first would push the name into the answers to every question after it.

5. 05

   Every answer recorded the same way

   For each answer we write down the assistant, the date, how the chat was run, and the four facts in the next section. Answers we cannot record fully, such as a partial export, are kept but not counted.

Examples of the questions

These come from the list. The first group are buyer questions. The second group name NIXAR and are asked in a separate chat.

• best AI SEO agency in Frisco TX
• my business doesn’t show up when people ask ChatGPT for recommendations, what do I do
• digital marketing agency in Allen TX for a small local business
• what drives the cost of an AI visibility audit

• what is NIXAR Solutions
• is NIXAR Solutions legit

• Is the business named

  Yes if the name appears anywhere in the answer, no if it does not.

• Its position in the list

  When the answer gives a list or a ranked shortlist, the business’s place in it. Blank when the business is not in the list.

• Which competitors are named

  The businesses the answer names, in order, so a change in who sits next to you shows up in the log.

• Which sources are cited

  The pages the answer points to. Cited sources tell us which pages and directories the assistant is reading, which is where the fixes start.

• Site fixes

  We check that assistants’ crawlers are allowed in, that every service has its own page with a plain, direct answer, and that the markup states who you are, where you work and what you do. A person edits every page before it goes live.

• Entity consistency

  An assistant trusts a business more when the same name, services, locations, phone number and founders appear everywhere it looks. We line up your site, your markup, your profiles and your directory listings so they say the same thing. When an assistant confuses a business with another one that has a similar name, this is the fix.

• Reviews

  We set up a review request on the platforms that allow reviews for your kind of business. Every customer is asked. We do not screen who gets asked and we do not steer unhappy customers away from the review page; low scores go to the owner separately.

• Third-party mentions

  Assistants also read pages that are not yours: directories, industry lists, local publications and partner sites. Where a real mention is missing, we help you earn one, for example a correct listing in the directory your customers use. We do not buy links or plant mentions.

The service behind these four jobs is on the AI SEO and GEO service page.

NIXAR was named in 0 of 6 fresh-chat Perplexity answers.

The last column counts the businesses each answer listed as a recommendation or option, checked by hand against the saved copy of the answer. We do not name them here.

What we did not count

• ChatGPT was checked from a signed-in account. Saved memory in a signed-in account can favor a business the account has talked about before, so those answers do not count toward the headline. ChatGPT needs a rerun in a temporary chat.
• The Gemini and Copilot exports were incomplete: the Gemini file held answers without the questions, and the Copilot file saved only its last screen. We did not count either one, and both need a rerun.
• Perplexity answers from longer threads, and any question that names NIXAR, are kept in the data but are also outside the headline.

How to read it

This is a baseline, not a verdict. Zero of six means that on the day we asked, in these six questions, Perplexity did not name NIXAR. AI answers can differ from one run to the next, which is why the check repeats every week and the comparison is week against week.

Four of the six answers told the buyer to ask an agency how it measures AI visibility, or to ask for a baseline audit first. That is the check on this page.

NIXAR’s weekly checks show whether the fixes to our own site and profiles change the answers.

1. Write five questions a customer would type, naming your service and your town. Leave your own business name out of them.
2. Open a new chat for each question in ChatGPT, Gemini, Perplexity and Claude. Sign out first, or use a temporary chat.
3. For each answer, note whether you are named, where you sit in the list, which businesses are named instead and which pages are cited.
4. In a separate chat, ask about your business by name and check that the answer gets your name, services and location right.
5. Do it again next week with the same five questions and compare.

If you would rather have it done for you, the visibility audit runs this check on your business and your competitors.