How to Prompt-Test Your Brand Across AI Assistants (Monthly Checklist)

0:00 Why AI Prompt Testing Matters Now
How do you prompt test your brand across AI assistants every month? What you do is run a set of scripted prompts through multiple AI assistants on the same day every month, and then you log what each one says about you. That’s the whole system. Everything else is just building the habit and knowing what exactly to look for. So let me walk you through the exact checklist that we use at push leads step by step so you can copy it and run it for yourself starting this month. Why this matters right now? Think about how your customers are searching for you today. A lot of them aren’t typing things into Google any longer. What they’re doing is asking ChatGPT, Perplexity, Google Gemini, or Microsoft co -pilot a question like who’s the best roofer in Asheville or which HVAC company should I call in Denver? If your name isn’t coming up in those answers or what comes up as wrong, you’re losing calls that you never knew you lost. The problem is that most businesses have no idea what AI assistants are saying about them. They’re not checking, they’re not logging in, they have no baseline, and that’s what this checklist fixes.

1:12 Pick Your Four AI Assistants
So what you want to do is pick your four AI assistants and stick with them because consistency is the whole point of monthly testing. You’ll need to test the same platforms every single time so that you can track change over time. So the four that we test every month are ChatGPT, Perplexity, Google Gemini, and Microsoft co -pilot. Those four cover the majority of where the real searches are landing right now. Pick those four, open accounts if you don’t have them, and don’t rotate platforms month to month. Changing platforms will break your comparison. What you want to do is build your core prompt library.

This is part of our AI search visibility guide.

1:52 Build Your Core Prompt Library
You’ll need a library of prompts, and every prompt needs to say the same thing word -for -word, month after month, which means you won’t know if your visibility has changed or your question has changed. Bucket one is category prompts. These are questions like who is the best plumber in Charlotte or what’s the top -rated landscaping company in Nashville. These show whether AI assistants are putting you in their short list at all. Bucket two is comparison prompts. Questions like how does this company compare to their competitors or what do customers say about this business? These reveal what reputation data the AI has absorbed. Bucket three is direct brand prompts. Questions like tell me about this company or is this company trustworthy. These tell you what the AI believes about you specifically. Start with 12 to 15 prompts total across those three buckets. That’s enough to give you a real signal without burning an hour every month.

2:56 Schedule and Protect Your Monthly Session
Pick a specific day every month and protect it. We run R’s on the first Tuesday of every month. It doesn’t matter really what day you pick. It matters that you pick one and don’t skip it. Set a reoccurring calendar block for 90 minutes. That’s enough time to run all four platforms and log everything. If you skip a month, your trend data breaks. One month of skipping won’t ruin you, but two months in a row starts to hurt your baseline.

3:23 Run Prompts and Capture Raw Output
Run the prompts and capture raw output. When you sit down for your testing session, open each AI assistant in its own browser tab and start with a fresh session for each platform. Don’t use the same chat thread that you used last month because some platforms pull context from previous conversations. You want to do a fresh session each time. Then run every prompt from your library through platform one. Copy the full response and paste it into your log. Then move on to platform two. Use the same prompts in the same order and repeat for all four platforms. Don’t edit the AI responses before you log them because you want the raw output exactly at the AI responses before you log them because you want the raw output exactly as it appears. Resist the urge to paraphrase paraphrasing is how data as it appears. Resist the urge to paraphrase paraphrasing is how data gets distorted over time.

4:23 Five Things to Look for in Output
What to look for in the output. Now you review what you’ve captured and you’re looking for five things. First is your business mentioned at all. If you’re not in the answer, that’s your biggest problem. Second is your business described accurately. Check your location, your services, your specialty. Wrong information is common and is fixable. Third, what sentiment does AI use about you? Is it positive, neutral, or negative? Log the actual words it uses. Fourth, who else is the AI recommending alongside you or instead of you? Log your competitors by name. That tells you who the AI trusts in your category. Fifth, does AI cite any other sources? Perplexity almost always does. ChatGPT sometimes does. Note the sources because they’re the context that AI is learning from.

5:17 Log Everything in One Spreadsheet
Log everything in one place. You’ll need a single spreadsheet or document where every month’s results live. Column one is the prompt text. Column two is the platform name. Column three is the date. Column four is whether or not your business appeared. Column five is the sentiment score. And you can just use negative, positive, or neutral. And column six is your competitor mentions. Column seven is your source citations. So that’s seven columns. Keep it simple. Don’t add columns because it feels like progress. More columns usually means less consistent logging. And what you want to do is score your visibility month over month.

5:59 Score Visibility Month Over Month
Once you’ve logged the raw data, you need a simple score so that you can track trajectory. We use an appearance rate. Take the number of prompts where your business appeared and divide that by the total number of prompts you ran. If you ran 15 prompts across four platforms, that’s 64 total prompt runs. If your business appeared in 22 of them, your appearance rate is 37%. Write that number down. Do the same calculation next month. Is the number going up, flat, or down. That trend is your signal.

6:34 Run Tests in Incognito Mode
Run the same test in private or incognito mode. This is the one that most people miss. Some AI assistant’s personalized results based on your account history or your location data. Running a second past of your most important prompts in a private browsing window or incognito mode while logged out. Compare those results to your logged in results if they’re different you’re seeing personalization. The incognito results are closer to what a stranger sees and that’s what matters most for your business.

7:08 Check Competitor Prompts Too
Check your competitor prompts too. This is in our checklist every single month. Take your top two or three competitors by name and run them through the same category prompts. Ask the AI who is better, you or them. Ask it to compare your business to theirs. Log what it says. You want to know if an AI assistant is actively recommending a competitor over you before your potential customer finds out the hard way. This also shows which competitor the AI trust and points you towards the content or sources that you need to match.

7:43 Trace the Sources AI Is Pulling From
Trace the sources that AI is pulling from and when an AI sites a source go look at it. Open the URL and read the content. Ask yourself, why did the AI trust this page? In my experience, AI assistants tend to trust, review aggregators, local directories, industry associations and press coverage. If your name isn’t on those sources, that’s a content gap. If the information on those sources is outdated or wrong, that’s a correction you can actually make. Fix the source and the AI’s answer often changes within a few months. Wrong information is the most common problem businesses find when they start running this checklist. The AI says your hours are wrong or your address is old or your services are incomplete or your name is misspelled. Here is what actually moves the needle. First, update your Google Business Profile. Update your website with clear structured information. Get that correction published on the sources that AI is already citing. AI assistants don’t have a correction form for you to fill out, but you can change what they say by changing what they read.

8:52 What a Real Testing Session Looks Like
Let me walk you through what a session looks like in practice. Say you’re a residential HVAC company based in Raleigh. You’re going to open in ChatGPT in a fresh tab and type, who are the best HVAC companies in Raleigh for a residential service? You’re going to copy the full response and paste it into your log. Your company might appear in position 3. The AI described you accurately, but called you a commercial contractor, which is wrong. Now you open a Perplexity and run the same prompt. Your company doesn’t appear at all. Perplexities cited a local magazine article from two years ago that mentions four competitors, but not you. That article is now a target. You go to Gemini, run the prompt, and you go to Gemini and run the prompt. And you appear, but only in a generic list with no detail. Then co -pilot, run the same prompt. You appear with the correct description and a review snippet. Now you have a picture. ChatGPT has the wrong data to correct. Perplexity has the source gap to fill. Gemini needs more detail in your online profiles. And co -pilot is working. These are four specific actions from one morning session. What to do next? If you want to know exactly where your brand stands right now across AI systems and traditional search, you can get a free visibility teardown at SEO .PushLeads .com -backslash -audit. Or call us at 828 -348 -7686. And we’ll show you exactly what AI systems are currently saying about your business and where your biggest gaps are. Start running the checklist this month. The business that build this habit now are going to have the real advantage over the ones who figure it out in a year from now. I’m Jeremy Ashburn with PushLeads, helping small businesses get more leads.