Ask it the way a customer would. Type the questions a buyer types into ChatGPT, Perplexity, and Google, then write down whether you are named, where you land, which sites the answer leans on, and whether the facts about you are right. Do that on a schedule and you have a measurement instead of a guess.
The problem
The owner who asks me this usually ranks fine on Google. They check Search Console, see clicks, and assume the job is done. Then a customer asks ChatGPT for a business like theirs and gets one answer, not ten links. Either the business is in that answer or it is not. The owner has no idea which.
Analytics cannot show this. Search Console reports clicks on Google, not what an assistant said. Ranking on Google does not, on its own, put you in the AI answer.
The branded side has its own problem. Ask an assistant about a business by name and it tends to quote prices, hours, programs, and an address from whatever page it read. Old pages tend to produce old facts. A directory can outrank your own site on your own name, so the assistant reads the directory instead of you.
The fix
The fix is a test you can repeat. Here is the short version you can run today.
1. Write down five questions a customer would type. Two or three unbranded ("[your service] in [your city]", "[your service] near me"). One or two branded ("[your business name] prices", "[your business name] hours"). Use the words a customer uses, not the words on your homepage.
2. Ask each one in three places. ChatGPT, Perplexity, and a Google search. Same wording in each.
3. For each answer, note four things. Are you named? Where are you listed: first, third, or not at all? What sources does the answer cite? Are the facts right (price, hours, address, programs)? Check them against your own pages, not your memory.
4. Write down who shows up instead of you. The names that appear when yours does not are your competition in AI search, whether or not they are your competition on the street.
5. Fix what you found, then run it again. Wrong hours on a directory listing. A program you dropped that still lives on an old page. Each one is a task with an owner. Rerun the same five questions in a month and compare.
That is the whole method. It is not sophisticated. Most owners I talk to have not done it.
What the AI Baseline adds
The manual version tells you where you stand today. The AI Baseline is what 412 Freelancers runs when an owner wants a number they can track. Same idea, more structure:
- A fixed question set. Branded and unbranded, written once, reused every run so results compare cleanly.
- Four lanes instead of three. ChatGPT, Perplexity, Google AI Overview, and Google organic. The AI Overview and the organic results are recorded separately.
- A score out of 100. Assistant positioning, 40 points, for unbranded ChatGPT and Perplexity answers where you are listed first. Brand citation, 20, for answers that cite your website. Google organic, 20, for questions where your own site holds the top result. Accuracy, 20, with deductions for every wrong or partly wrong fact. The method is public so you can check the math.
- Who is cited instead of you. Which sites the answers were built from, and which competitor is gaining authority run over run.
- A fact check against your own pages. Every claim about you marked right, partly wrong, or wrong.
- An action plan with owners. Each change assigned to someone on your side, with the points it is worth.
- Scheduled reruns. Every two weeks or monthly by package, reviewed in a monthly meeting.
- Per-location scoring. A multi-location brand or franchise gets a test, score, and plan for each location.
One thing to be clear about: 412 Freelancers does not implement the changes. Your website team or location manager does. The AI Baseline tells you what to fix and whether the fix worked.
The industry name for this is AEO, answer engine optimization. In plain English: being named, cited, and described correctly in the one answer an AI gives.