How to Track Your Brand in AI Search (Free Prompt Template)
A repeatable prompt-set method to see whether ChatGPT, Perplexity, and Gemini actually recommend you — and measure it over time.
If a model won't name you, a rank tracker can't save you. The only metric that matters in AI search is whether you're in the answer.
Your customers have quietly changed where they ask for recommendations. Instead of typing "best project management software" into Google and scanning ten blue links, a growing share of them now ask ChatGPT, Perplexity, or Google's AI Mode — and act on whatever two or three names the model volunteers. If you want to track your brand in AI search, you cannot rely on your old rank-tracker. There is no page two to climb to. Either the model names you, or you are invisible.
The problem is that most teams have no idea whether they are being named. They watch organic rankings that increasingly don't convert, while the decisions that matter happen inside a chat window they never see. This guide gives you a concrete, repeatable method to fix that: a prompt set, a scoring rubric, and a cadence you can run this week with nothing but a spreadsheet.
Why you need to track your brand in AI search
The short version: AI answers are a closed room, and you are either in it or you aren't. ChatGPT reportedly reached roughly 900 million weekly users by early 2026, and in categories from software to supplements, buyers increasingly treat its shortlist as the shortlist. Unlike a search results page, an AI answer rarely shows the runners-up. Position three on Google still gets clicks; "not mentioned" in an AI answer gets nothing.
It also matters because this traffic is unusually valuable. Across several 2026 B2B benchmarks, visitors referred by AI assistants converted far better than Google organic — one widely cited study put AI-referred conversion at 14.2% versus 2.8% for organic. People arrive pre-qualified because the model has already vouched for you. Missing from the answer means missing your highest-intent pipeline, not just some impressions.
If a model won't name you, a rank tracker can't save you. The only metric that matters in AI search is whether you're in the answer.
The catch is measurement. A large portion of AI referrals land in analytics as "direct" traffic — some estimates put it above 70% — so your existing dashboards undercount AI's influence by two to three times. That is exactly why you need a deliberate monitoring method instead of waiting for GA4 to tell you. (Once you've set this up, pair it with our guide to tracking AI search traffic in GA4 to connect visibility to sessions.)
What "tracking" actually means in AI search
Traditional SEO monitoring asks one question: what position do I rank for this keyword? AI search monitoring asks four:
- Presence: Does my brand appear in the answer at all?
- Citation: Is there a link or source pointing to my site?
- Sentiment and accuracy: How am I described — and is it correct?
- Competitive context: Who is named alongside me, and in what order?
Those four dimensions roll up into a single headline number marketers now call AI share of voice: the percentage of relevant prompts where your brand shows up. Tracking is simply the disciplined process of generating that number again and again so you can see it move.
How to track your brand in AI search: a 5-step method
Here is the whole workflow. It takes a couple of hours to set up and about 30 minutes a week to run.
Step 1: Build your prompt set
A prompt set is the fixed list of questions you test on every run. Think of it as your keyword list for the AI era. A good set spans the buyer's journey rather than just your brand name, because the valuable discovery moments happen before someone knows you exist. Aim for 20–40 prompts across four types:
| Prompt type | Example | What it reveals |
|---|---|---|
| Category / "best of" | "What are the best [category] tools in 2026?" | Whether you're on the default shortlist |
| Use-case / problem | "How do I [job your product does] without [common pain]?" | Whether you surface for intent, not name |
| Comparison | "[You] vs [competitor] — which is better for [segment]?" | How you're framed head-to-head |
| Branded | "Is [your brand] any good? What do people say?" | Sentiment and factual accuracy |
Write the prompts the way a real buyer would, in natural language. Keep the list stable — the point is to measure the same thing over time.
Step 2: Run the prompts across the major assistants
Test every prompt in ChatGPT, Google AI Mode / Gemini, Perplexity, and Microsoft Copilot at minimum; add Claude if your buyers use it. Two rules keep your data clean:
- Use fresh, logged-out sessions. Memory and personalization will bias results toward what you've clicked before.
- Run each prompt two or three times. AI answers vary between sessions, so a single run is noise. Record the pattern, not one lucky hit.
Step 3: Score each answer
Capture results in a simple rubric. Give each prompt a score so you can track a number, not a vibe:
| Dimension | 0 | 1 | 2 |
|---|---|---|---|
| Presence | Not mentioned | Mentioned in passing | Recommended by name |
| Citation | No link | Mentioned without link | Linked / cited |
| Sentiment | Negative or wrong | Neutral | Positive and accurate |
| Position | After competitors | Mid-list | Named first |
A prompt can score 0–8. Sum across your set and you have a baseline you can defend in a meeting.
Step 4: Calculate share of voice and log competitors
Your headline metric is straightforward:
AI share of voice = (prompts where you're recommended ÷ total prompts) × 100
Log which competitors appear, too. If three rivals consistently show up for your best "category" prompts and you don't, that gap is your roadmap. Our walkthrough on GEO competitor analysis explains how to reverse-engineer why models cite them.
Step 5: Set a cadence and watch the trend
AI answers shift as models retrain and re-crawl, so one snapshot is almost useless. Run the full set monthly, and spot-check your five most important prompts weekly. Chart share of voice over time. When it moves, you'll know whether a new page, a Reddit thread, or a fixed fact moved it — and you can double down.
Tips and reminders
- Don't skip the un-branded prompts. Ranking for your own name is table stakes; the growth is in category and use-case prompts.
- Screenshot the wins and the losses. Verbatim answers are persuasive evidence for stakeholders and useful when you need to correct wrong information AI repeats about your brand.
- Automate once it hurts. A spreadsheet is perfect for one brand and 20 prompts. At 100+ prompts across five engines, run weekly, the manual version collapses — that's where a monitoring tool or an agent earns its keep.
- Treat it as a loop, not a report. Tracking only pays off if each reading triggers an action: a new page, a better answer block, a pitch to a source the model trusts.
DIY spreadsheet vs. automated monitoring
| Spreadsheet (DIY) | Automated monitoring | |
|---|---|---|
| Cost | Free | Paid tool or agent |
| Setup time | ~2 hours | Minutes |
| Best for | 1 brand, ≤30 prompts, monthly | Many prompts, daily/weekly, multiple engines |
| Weakness | Tedious, easy to abandon | Needs budget |
| Payoff | Proof of concept | Continuous, alertable trend |
Start with the spreadsheet. It forces you to understand the mechanics and gives you a baseline before you pay for anything. The day the manual run becomes the chore you keep postponing is the day to automate it. ivon's autonomous marketing platform runs this prompt-set monitoring continuously and turns the gaps into published fixes — the same five-step loop, without the Monday-morning copy-pasting.
Closing: measure the answer, not the ranking
Search didn't disappear; it moved into a conversation you can't see unless you look. Tracking your brand in AI search is how you look — deliberately, repeatably, and across the handful of models your buyers actually use. Build the prompt set, score the answers, calculate your share of voice, and run it on a cadence. Within a month you'll know exactly where you stand, which rivals own the shortlist, and what to fix first.
If you'd rather not run it by hand forever, see how ivon automates the whole monitoring-to-fix loop.
Frequently asked questions
How do I check if ChatGPT recommends my brand?
Open a fresh, logged-out ChatGPT session and run a short set of real buyer questions — a best-of category prompt, a problem-based prompt, and a branded prompt. Note whether you're named, whether you're linked, and which competitors appear. Run each prompt two or three times, because answers vary, and track the pattern rather than a single result.
What is AI share of voice?
AI share of voice is the percentage of relevant prompts where an AI assistant recommends your brand. Calculate it as the number of prompts where you're named divided by the total prompts tested, times 100. It's the closest AI-search equivalent to keyword rankings, and it's the single number to watch as you improve your visibility over time.
How often should I track my brand in AI search?
Run your full prompt set monthly and spot-check your most important five prompts weekly. AI models retrain and re-crawl on their own schedules, so visibility shifts without warning. A single snapshot is nearly useless; a trend line tells you whether your changes — new pages, citations, or fixed facts — are actually moving the needle.
Which AI assistants should I monitor?
At minimum, track ChatGPT, Google AI Mode / Gemini, Perplexity, and Microsoft Copilot, since they cover the vast majority of AI-assisted queries. Add Claude if your audience skews technical. Each engine builds answers differently, so a brand that dominates ChatGPT can be invisible in Perplexity — monitor them separately rather than assuming one result speaks for all.
Can I track AI search visibility for free?
Yes. A spreadsheet with a fixed prompt set and a simple 0–2 scoring rubric costs nothing and is the best way to start. The limit is scale: once you're testing 100-plus prompts across several engines every week, the manual process becomes unsustainable and a dedicated monitoring tool or agent pays for itself.