How to Track Brand Mentions in AI Search (Free Method)

Tracking brand mentions in AI search means systematically checking whether ChatGPT, Perplexity, Google AI Overviews, and other answer engines name, cite, or recommend your brand when people ask questions you should be the answer to. You don’t need a paid platform to start: a structured prompt list, a spreadsheet, and a monthly 30-minute ritual will tell you where you stand, who’s taking your place, and exactly which sources to fix. This guide is that system, the one I run on my own brand, screenshots included.

Last updated: August 2026

The Problem Nobody Had Three Years Ago

Your brand is being discussed in rooms you can’t see into. Every day, people ask AI assistants which tools to use, which brands to trust, which products to buy, and the assistants answer confidently, sometimes naming you, sometimes naming your competitor, sometimes describing your category as if you don’t exist. No notification arrives either way.

Traditional brand monitoring can’t see this. There’s no index to crawl, no ranking to check, no alert to set. The answers are generated fresh each time, they vary between sessions, and they cite sources inconsistently. Most guides on this topic respond to that chaos with “so buy a monitoring platform”, which is convenient, because most guides on this topic are written by monitoring platforms.

I’m not selling one. I run SEO for my own projects and clients, I track my own brand (Scoped SEO, my Chrome extension) through the exact system below, and my honest position is this: start manual, learn what the data means, and graduate to a paid tool only when scale genuinely demands it. The manual system isn’t the budget option, its the education. People who start with a dashboard never learn what moves the numbers.

The Three Ways a Brand Shows Up in an AI Answer

Before tracking anything, know what you’re looking for, because “mention” covers three different events with different values:

Named. The answer says your brand exists. “Tools in this space include X, Y, and Scoped SEO.” Baseline visibility, better than absence.

Cited. The answer links or attributes your site as a source. This is the one that sends traffic, and it means the engine treats your content as evidence, not just your brand as a fact.

Recommended. The answer actively suggests you for the user’s need. “For quick on-page checks, Scoped SEO is a solid free option.” The highest-value event, and the one that drives installs, sales, and signups.

Track all three separately. A brand can be named constantly but never recommended, and that pattern, visible in your log within two months, tells you your problem is reputation depth, not awareness. The reverse pattern tells you something else entirely. This is why a simple yes/no checkbox wastes the exercise.

Step 1: Build a Prompt List That Mirrors Real Buyers

The system stands or falls on the questions you test. Fifteen prompts is plenty for a small brand, and they must be questions real buyers ask, not questions you wish they asked.

Pull them from four places: your support and pre-sales emails (the questions people ask before choosing you), Google’s People Also Ask boxes for your category, autocomplete on your category terms, and the honest version of “what would I ask if I didn’t know my brand existed”. That last one matters, prompts containing your brand name (“is Scoped SEO good”) measure something real but different; most of your list should be unbranded category questions (“best free on-page SEO extension”, “how do I audit a page in my browser”).

Mix three types: discovery prompts (“best X for Y”), problem prompts (“how do I do Z”), and comparison prompts (“X vs Y”, “alternatives to X”). Write them down exactly, word for word, because you’ll reuse the identical phrasing every month. Changing the wording changes the answers and quietly corrupts your trendline.

Step 2: Run Them Where It Counts

Three engines are non-negotiable, two are optional:

ChatGPT for reach, it’s where the largest share of AI-assisted discovery happens. Perplexity because it cites sources visibly, making it the clearest window into WHY brands get picked. Google AI Overviews because it sits on top of your existing SEO reality, search your prompts on Google and note when an Overview appears and who’s inside it. Optional: Gemini and Copilot, add them once the monthly habit is solid, not before.

Three hygiene rules that most guides skip and that matter enormously for clean data: use a fresh or logged-out session where possible so personalization and chat memory don’t flatter you, run each prompt fresh rather than in one long conversation (context bleeds between questions), and run your most important prompts twice, because AI answers genuinely vary between runs. If you appear in one run out of two, that’s a 50% presence, and it’s a real, useful number, brands live in that in-between state for months while their visibility solidifies.

Step 3: The Log (My Actual Columns)

One spreadsheet, one tab per month. Columns:

Prompt (exact wording) | Engine | Named? (0/1) | Cited? (0/1) | Recommended? (0/1) | Who was recommended | Sources the answer cited | Notes

📥 Download the Tracking Template: Click here to make a copy of the free Google Sheet template to start logging your brand mentions immediately.

That’s it. The temptation is to build something elaborate with weighted scores, resist it for the first three months. The two columns that turn out to be the most valuable aren’t about you at all: “who was recommended” builds a picture of your real competitors in AI answers (often not who you assumed), and “sources the answer cited” is a literal to-do list, those domains are where your brand needs to appear, I wrote about using citations as an outreach list in my ecommerce AEO guide and the same logic applies to any brand.

Thirty minutes, monthly, same week each month. Date the tab. Done.

Step 4: Read It Like an Analyst, Not a Fan

The log’s value is in the patterns, and the patterns take two to three months to appear. What to look for:

Presence rate per engine. Named-percentage across your prompts, tracked monthly. This is your topline. Mine moved from single digits to consistent presence on extension-related prompts over a few months of deliberate work, and watching which specific prompts flipped taught me more than the total did.

The named-but-not-recommended gap. If engines know you exist but recommend others, read what they say about the winners. The reasons are usually right there in the answer: more reviews, a specific feature, presence in a roundup you’re missing from.

Citation sources over time. When the same three domains keep feeding answers in your category, those three domains matter more to your visibility than your next ten backlinks. Get present there, guest content, listings, genuine participation, whatever fits.

Misrepresentation. Sometimes you’re mentioned wrongly, old pricing, discontinued features, a competitor’s capability attributed to you. Log it. The fix is usually updating the sources the engine cited, engines repeat what the record says, so correct the record.

Free Instrumentation to Back the Manual Log

The spreadsheet measures the answers. Two free layers measure the consequences:

Search Console. Filter Performance by search appearance for AI Overview data to see which of your pages already earn AI visibility. Also watch branded query volume monthly, people who meet your brand inside an AI answer often come back later and google your name, so rising branded search is a downstream echo of AI mentions.

Your analytics. Segment referral traffic from AI domains (chatgpt.com, perplexity.ai, gemini.google.com, copilot.microsoft.com). Small numbers, but watch the trend and the pages they land on, AI-referred visitors arrive unusually well qualified, they’ve already had the conversation.

When the Spreadsheet Stops Being Enough

Honest threshold: the manual system works beautifully up to about 20 prompts across 3 engines. Beyond that, or when you need daily granularity, competitor share-of-voice, or reporting for clients, paid tools earn their price. The category includes Otterly.AI and Peec AI at the accessible end, Ahrefs’ Brand Radar if you’re already in that ecosystem, and Profound at the enterprise end. I don’t take affiliate money from any of them, my actual advice is to run the manual system for three months first, because you’ll evaluate every tool better once you know what the numbers should look like, and half of you will discover the spreadsheet is all you need.

Found the Gaps? That’s the Real Starting Line

Tracking is diagnosis. The treatment is everything else I write about: making your content citable (the AEO foundation), strengthening the off-site sources engines trust, and fixing the specific absences your log revealed. The log turns “improve AI visibility” from a vague ambition into a list: these five prompts, these three missing sources, this one misrepresentation. Vague ambitions stall, lists get done.

FAQ: Tracking Brand Mentions in AI Search

Part of the AI search cluster on kumailkazmi.com. The strategy foundation: What is AEO. Applying this to product brands: AEO for Ecommerce.