Knowing how to measure AI search visibility is one of the most practical things a small team can do right now, and you do not need an expensive platform to start. By combining Google Search Console, GA4, and a simple manual prompt log, you can track whether ChatGPT, Perplexity, Gemini, and Google AI Overviews mention your brand each month. This guide gives you a concrete, free-first workflow that most listicles skip entirely.
What AI search visibility metrics actually mean
AI search visibility is a measure of how often your brand, content, or site appears inside the answers that AI engines generate, covering mentions, citations, share of voice, and sentiment. Understanding these four metrics before you start collecting data saves a lot of confusion later.
- Mentions: your brand name appears in the generated answer, with or without a link.
- Citations: the AI engine links directly to your URL as a source.
- Share of voice: how often your brand appears compared with named competitors across the same set of prompts.
- Sentiment: whether the reference is positive, neutral, or negative in context.
These metrics behave differently from traditional organic search data. A citation in Perplexity does not always produce a measurable click, and a mention in ChatGPT may drive branded search days later rather than a direct referral. Tracking all four together gives you a fuller picture than any single signal alone. For a broader introduction to these concepts, the AI search visibility for small teams pillar guide is worth reading first.
How to measure AI search visibility with a manual prompt log
Manual prompt testing is the most accessible way to check whether your brand appears in AI-generated answers, and it costs nothing beyond time. The method works by running a defined set of prompts across ChatGPT, Perplexity, Gemini, and Google AI Overviews, then recording the results in a shared spreadsheet.
Building your prompt set
Choose 10 to 15 prompts that reflect real questions your customers type. Mix branded queries ("What is [your brand]?"), category queries ("best tools for [your use case]"), and problem-led queries ("how do I [solve the problem you solve]?"). Keep the list stable month to month so you can spot trends rather than noise.
Running each prompt reliably
A single run is not enough. According to research cited by Supermetrics, the prompt-level presence rate only becomes reliable when each prompt is sampled three to five times per engine, because one run can return two different answers. Run each prompt at least three times, clear the chat context between runs, and record every result, not just the ones where you appear.
A SparkToro study cited by Practical Ecommerce confirms the same point: AI platforms recommend different brands in different orders each time the same person asks the same question. Sampling three to five times per prompt, per engine, gives you a presence rate you can actually trust.
What to log
For each prompt run, record the engine, the date, whether your brand was mentioned, whether a link was included, which competitors appeared, and a brief note on sentiment. A simple Google Sheet with these six columns is enough to spot month-on-month movement without buying any software.
Using Google Search Console for AI Overviews data
Google Search Console now includes a Generative AI performance report that shows how your content performs inside generative AI features on Google Search and Discover. According to Google Search Central, site owners can use this report to measure impressions and performance inside AI-generated features, making it the most direct free signal available for Google AI Overviews specifically.
To find the report, open Search Console, go to Search results, and filter by Search type. The Generative AI tab shows impressions from AI features separately from standard web results. For a full walkthrough of what changed in the August 2026 update to this report, see the Generative AI Performance Report update guide.
One important caveat: the Search Console AI reports omit click data and query-level metrics, according to Search Engine Journal. You can see impressions but not which specific queries triggered the appearance. This limits how granular your analysis can get, so treat Search Console as a trend indicator rather than a query-level research tool.
Google has also introduced a generative AI content control in Search Console. SOMO reported that the toggle rolled out to UK properties on 17 June 2026 with Include, Exclude, and Inherit options, letting site owners manage whether their content appears in generative AI Search features at all. Click data is not yet broken out from traditional search in this view, but the impressions signal is real and trackable.
Setting up GA4 to isolate AI referral traffic
GA4 custom channel groups let you separate traffic arriving from AI platforms so you can measure referrals from Perplexity, ChatGPT, and similar engines without losing that data inside the "direct" or "referral" buckets.
To set this up, go to Admin in GA4, then Channel groups, and create a custom channel. Add rules for the referral sources you want to track. Common ones include perplexity.ai, chat.openai.com, and gemini.google.com. Label the channel "AI Referral" or similar. Once active, this channel appears in your acquisition reports so you can see sessions, engagement rate, and conversions by source.
- Monitor session volume from each AI engine month over month.
- Check engagement rate, AI-referred users often read more pages per session than social traffic.
- Watch for spikes in branded organic search after a prompt test period, a common indirect signal of AI mentions.
- Use the landing page report filtered to your AI Referral channel to see which content AI engines send traffic to.
Branded search volume is a useful proxy metric for measuring AI visibility. When your brand is mentioned frequently in AI answers, more people search for it directly in Google. Tracking branded impressions in Search Console alongside your AI referral sessions in GA4 gives you two independent signals pointing at the same thing.
Comparison: free vs paid AI visibility tracking options
Before committing to any paid tool, it helps to understand what each tier actually gives you. For tools covering Gemini specifically, the AI visibility metrics Gemini guide goes deeper on what to track and why.
| Method or tool | Cost | Engines covered | Data type | Best for |
|---|---|---|---|---|
| Manual prompt log | Free | All (manual) | Mentions, citations, sentiment | Any team, immediate start |
| Google Search Console | Free | Google AI Overviews | Impressions (no click data) | Google AI Overviews trend tracking |
| GA4 custom channels | Free | All (referral only) | Sessions, engagement, conversions | Measuring traffic from AI engines |
| Otterly (free tier) | Free / paid tiers | ChatGPT, Perplexity, Gemini | Automated prompt monitoring | Teams wanting partial automation |
| Supermetrics AI connector | Paid | Multiple | Aggregated AI referral data | Teams already using Supermetrics |
| Enterprise AI trackers | High cost | All major engines | Full share of voice, competitor data | Large brands with dedicated budgets |
If you want to compare lightweight paid options before committing, the roundup of Profound alternatives for small teams covers seven tools with honest notes on free tiers. For a head-to-head on one specific option, Otterly vs Remway breaks down which fits a one-person SaaS setup.
Building a monthly AI visibility report on a small team budget
A monthly AI visibility report pulls together your prompt log, Search Console impressions, GA4 referral data, and branded search trends into one place so you can spot what is working and what to fix. The goal is a repeatable process you can run in under two hours per month.
The reporting cadence
Run your full prompt set on the same week each month. Export your Search Console Generative AI impressions for the same period. Pull your AI Referral channel data from GA4. Then combine the four signals into a simple scorecard: presence rate (what percentage of prompt runs mentioned your brand), impression count, referral sessions, and branded search volume.
Reading the signals together
No single metric tells the full story. A high impression count with low referral sessions suggests your content appears in AI Overviews but does not earn the click. A rising presence rate in your manual log alongside flat branded search may mean you are getting AI mentions but not yet building recall. When all four signals rise together, that is the outcome worth working toward.
- Presence rate: target a consistent upward trend rather than a specific number.
- Impressions: compare month over month in Search Console, not absolute values.
- AI referral sessions: watch engagement rate to judge quality, not just volume.
- Branded search: rising branded impressions often lag AI mention increases by two to four weeks.
Keeping this report in a single Google Sheet, updated monthly, gives you a historical record that paid platforms often charge extra to retain. It also gives you the raw material for a brand visibility competitive analysis when you are ready to benchmark against named competitors.
Frequently Asked Questions
How do I know if my brand is mentioned in ChatGPT answers?
The most direct method is manual prompt testing. Run the questions your audience is likely to ask in ChatGPT, record whether your brand appears in each answer, and repeat each prompt three to five times to account for variation. There is no API or native dashboard that shows brand mentions inside ChatGPT answers at the time of writing, so a manual prompt log remains the most reliable free approach. Some third-party tools like Otterly can automate part of this process if you want to scale it.
Does Google Search Console show AI Overview data?
Yes. Google Search Console includes a Generative AI performance report that shows impressions from AI features including AI Overviews. According to Google Search Central, site owners can use this report to measure content performance inside generative AI features on Google Search and Discover. The report does not yet break out click data or query-level detail, but impression trends are trackable month over month.
How many times should I run each prompt to get reliable results?
Run each prompt at least three times per engine, and ideally five times for prompts that matter most to your business. Research cited by Supermetrics shows that a prompt-level presence rate only becomes reliable at three to five samples, because a single run can return different answers on consecutive tries. Clear the chat context between each run and log every result, including runs where your brand does not appear.
Which free tools can help track AI citations in Perplexity or ChatGPT?
The most practical free setup combines a manual prompt log in Google Sheets, Google Search Console for AI Overviews impressions, and a GA4 custom channel group to capture referral traffic from AI engines. Some tools like Otterly offer a free tier that automates prompt monitoring across multiple engines. For a comparison of lightweight paid options, the Profound alternatives roundup for small teams covers seven tools with notes on what each free tier actually includes.
Can I stop my site from appearing in Google AI Overviews?
Yes. Google rolled out a generative AI content control in Search Console that gives site owners Include, Exclude, and Inherit options for how their content appears in generative AI Search features. According to SOMO, this toggle became available to UK properties from 17 June 2026. Setting it to Exclude removes your content from AI feature appearances, though it does not affect standard organic rankings. The setting is available at a property level inside Search Console.