TL;DR: To check what AI says about your brand, ask the engines directly — ChatGPT, Perplexity, Gemini, and Google's AI Overviews — a spread of buyer questions, and record whether you're mentioned, how you're described, and who appears instead. Do it consistently, because single answers are noisy. This guide gives you a rigorous manual method first, then explains when to graduate to automated tracking for scale.
Why this is worth doing
Your buyers are asking AI assistants about your category right now, and the answers they get shape who makes their shortlist. If you've never checked what those answers say, you're flying blind on a channel that's growing fast — Bain found 42% of LLM users already ask AI tools for shopping recommendations. The good news: you can get a real read today, by hand, for free. Let's do it properly.
The manual method, step by step
You don't need a tool to start. You need a spreadsheet, thirty minutes, and discipline about consistency.
Step 1: Write your question set
List 8–12 questions a real buyer would ask, in three groups:
- Category questions (no brand name): "best [category] for [use case]", "affordable [category] for [buyer type]", "what should I look for in [category]". These are the most revealing — they show whether you appear when someone isn't already searching for you.
- Comparison questions: "[your brand] vs [competitor]", "[competitor] alternatives", "is [your brand] or [competitor] better for [need]".
- Reputation questions: "is [your brand] legit", "what do people say about [your brand]", "is [your brand] worth it".
Write them the way a person talks, in full sentences. Save them; you'll reuse this exact set every time, because a stable question set is what makes your checks comparable over time.
Step 2: Ask each engine
Open each engine and ask your questions, one per fresh conversation (so previous answers don't bias the next):
- ChatGPT — with web search available.
- Perplexity — which cites its sources by number, making citations easy to see.
- Gemini — Google's model.
- Google AI Overviews — search your questions on Google and note when an AI summary appears at the top.
A fresh chat per question matters — context from a prior answer can skew the next one and give you a falsely rosy read.
Step 3: Record what you see
For each question and engine, log four things in your spreadsheet:
- Mentioned? Are you named in the answer at all?
- How described? The exact phrasing — "budget pick", "best for beginners", "premium option". This is often more useful than the yes/no.
- Cited? Are any of your pages linked as sources (clearest on Perplexity and AI Overviews)?
- Who else? Which competitors appear, and how are they described?
That fourth column is where strategy comes from: the competitors who show up when you don't are the ones winning your category's AI visibility, and the sources cited are the pages doing the winning.
Step 4: Repeat and compare
Run the same set again in a few weeks. The variation you see between runs is itself the most important lesson — it teaches you not to overreact to any single answer. What matters is the pattern: consistent absence on category questions is a real problem; a one-time dip is probably noise. This is exactly why professionals track on a schedule rather than checking once.
Reading your results
A few interpretation guidelines that keep you honest:
- Being one of three named brands is good, not bad. AI answers name only a handful of options, so high "coverage" is structurally impossible. Judge yourself against competitors, not against a 100% ideal.
- How you're described matters as much as whether you're named. "The best for sensitive skin" beats "a cheaper option" even though both are mentions.
- Citations are the controllable lever. If a competitor's comparison page is cited and you have no equivalent, that's a specific, buildable fix. Our citation tracking guide goes deeper.
- Absence on category questions is the real alarm. Being invisible when a buyer describes their need — without naming you — is the gap that costs sales.
The limits of checking by hand
Manual checking is the right start, and it has honest limits you should know before you rely on it:
- It's noisy. A handful of prompts on a couple of engines, run occasionally, can't distinguish a trend from run-to-run variance — one study found under half of brands stay visible across a three-week window.
- It doesn't scale. Checking 10 prompts by hand is fine; checking 50 prompts across 5 engines every week is a part-time job.
- It's hard to keep consistent. Small wording changes and different people running the checks introduce drift that muddies your trends.
- You can't easily prove change. Showing "our AI visibility improved after we shipped this" needs systematic before-and-after data.
When those limits start to bite — usually once you're actively working to improve and want reliable trends — that's the signal to automate.
A note on checking product and ecommerce visibility
If you sell products, add a fifth column to your manual check that generic brand-monitoring guides skip: product-level questions. Alongside "best [category]," ask the specific buyer questions your products answer — "best [product type] for [use case]", "affordable alternative to [popular product]", "is [your product] good for [need]". These reveal whether AI understands your catalog, not just your brand, and they're where DTC revenue actually moves.
Watch for a specific failure pattern: your brand gets named in general answers but your products don't surface for specific product queries. That usually means AI knows your brand exists but can't parse your product pages well enough to recommend individual items — a catalog-readiness problem (offer data, schema, descriptions) rather than a brand-awareness one. It's worth diagnosing separately because the fix is different: brand-awareness gaps need reputation and content; catalog-readiness gaps need structured, parseable product pages. This distinction is why ecommerce-focused GEO tools exist as a category, and why a general brand monitor can miss the problem entirely.
When to move to a tool
Automated GEO tools run a fixed prompt set across engines on a schedule, smooth the noise, track citations and sentiment, and show trends over time. They're how you go from "I checked once and we looked absent" to "here's our visibility trend across five engines and which content moved it." The monitoring tools guide compares the options, and the main tools guide covers the whole category.
If you want a fast automated baseline without committing to a subscription, several tools offer a free one-time audit. Stride's free audit, for instance, runs a starter prompt set across multiple engines and reports your visibility, sentiment, and which competitors appear — a quick way to compare your manual findings against an automated read before deciding whether ongoing tracking is worth it.
Frequently asked questions
How do I find out what AI says about my brand?
Ask the AI engines directly. Open ChatGPT, Perplexity, Gemini, and Google (for AI Overviews), and ask category and brand questions a buyer would ask — "best [category] for [use case]", "is [your brand] any good", "[your brand] vs [competitor]". Record whether you're mentioned, how you're described, and which competitors appear. Repeat over time, because answers vary run to run.
Is checking AI mentions manually reliable?
It's a good start but noisy. A single answer isn't representative — AI responses vary between runs, engines, and phrasings. Manual checks are perfect for confirming whether you have a visibility problem and seeing how you're described. For reliable trends over time and across many prompts and engines, scheduled tracking with a tool is more dependable.
What questions should I ask the AI about my brand?
Mix three types — category questions that don't name you ("best [category] for [buyer]"), comparison questions ("[your brand] vs [competitor]"), and direct-reputation questions ("is [your brand] legit", "what do people say about [your brand]"). The category questions are the most revealing, because they show whether you appear when a buyer isn't already looking for you.
How often should I check what AI says about my brand?
For a manual check, monthly is a reasonable rhythm, with a fresh look after any major content or PR change. For anything you're actively working to improve, weekly or automated tracking is better, because AI answers are volatile and single checks can mislead. The point is a consistent cadence, not a one-time audit.
Can I check what AI says about my brand for free?
Yes. Asking the engines directly is free, and several GEO tools offer a free one-time audit that checks your visibility across engines automatically. The free approaches are great for a baseline; ongoing monitoring across many prompts and engines is where paid tools earn their cost.
Stride is a GEO platform for Shopify and DTC brands. Its free audit automates the check across multiple engines in about two minutes — no account required.