Why Use a Tool When You Can Just Ask AI Yourself? The Free Method First, Then When It's Time to Switch

    By TriloAI Editorial TeamPublished Updated

    The short answer: if all you want to know is “does AI mention me right now?”, opening ChatGPT and asking a few questions is enough. You don't need to spend anything. This article lays out the manual method in full first, then is honest about where its limits are.

    The manual method: five steps

    1. List 10 to 15 questions your potential customers would actually ask. Don't just ask “do you know Brand X?” — that's prompting the AI, not measuring it. Ask questions that don't contain your brand name, such as “what are the best fitness studios in New York?”
    2. Cover different intents: recommendations, comparisons, pricing, concerns and alternatives. If you only ask recommendation questions, you'll overestimate yourself.
    3. Ask each question once in ChatGPT, Gemini and Perplexity, and record the results separately. Don't rely on just one.
    4. Record three things: whether you're mentioned, whether you're included as a recommendation, and which URLs the answer cites. The third matters most — those are the places you'll need to get into next.
    5. Use an incognito window and stay logged out. Otherwise your chat history shapes the answer, and you end up measuring “what the AI thinks of you” instead of “what a stranger would see.”

    Put the results in a spreadsheet and repeat once a month. It costs nothing, and it answers most of the basic questions.

    Three places manual tracking goes wrong

    These aren't sales arguments. They're problems you run into as soon as you actually do it:

    ProblemWhat happens manuallyConsequence
    Sample too smallEach question is usually asked only onceAI answers naturally vary, so a single result may be noise rather than a trend
    Questions driftNext month you re-ask from memory and the wording shifts slightlyOnce the questions change, month-to-month scores are no longer comparable
    Login contaminationIt's easy to forget to log out or to use an account with memoryYou're measuring a personalized result, not what a stranger sees

    The second one is the most underestimated. You think you're tracking month over month, but the test is actually different every month — so that trend line isn't real.

    When to switch to a tool

    The most practical way to decide is by time cost. One manual round of 15 questions × 3 platforms = 45 queries, plus recording and tidying up, takes roughly one to two hours. Here's how that adds up:

    • Tracking only your own brand, checking once a quarter — manual is completely worth it. No tool needed.
    • Your own brand, tracked monthly — about 12 to 24 hours a year. Still acceptable, depending on what your hour is worth.
    • An agency serving several clients at once — one to two hours per client per month means five clients take 5 to 10 hours of pure manual work every month, and the labor often costs more than a tool would (see how agencies can offer GEO).
    • You need to deliver reports to clients or managers — turning the results into something presentable often takes longer than the queries themselves.

    The dividing line usually isn't the number of brands — it's whether you're accountable for the number. If it's just for your own reference, manual is enough. If you need to use it to convince someone else, traceability and consistency become requirements.

    If you start manually, don't skip these three things

    1. Save the exact wording of each question and re-ask it word for word next time. It's the only way to make results comparable over time.
    2. Record the cited source URLs, not just whether you were mentioned. That list of URLs tells you exactly where you need to show up.
    3. Use an incognito window every time and note the date of each query. AI answers change over time, and a record without a date can't be interpreted.

    Will asking manually and asking through a tool give the same answers?

    At the same time, with the same question and the same logged-out state, the results will point in a similar direction, but they won't be identical — AI answers naturally vary. That's also why no single result should be treated as a conclusion, whether it comes from manual checks or a tool.

    Can I just ask the AI to analyze it for me?

    You can ask, but be careful. If you ask an AI directly “how visible is my brand to you?”, what you usually get is a statement generated from its impressions, not a measurement. To measure, you have to ask questions that don't contain your brand name and see whether it brings you up on its own.

    How do free tools compare with doing it manually?

    The value of a free tool is that it saves the time spent recording and organizing, and it locks your questions in place. If you're already doing it manually, moving to a free option is usually a painless upgrade. We also offer a free quick check that looks at ChatGPT only, takes about one to two minutes and doesn't require an account (one per email address), so you can see where you stand before deciding. Tool prices are compared in our pricing article.