An AI visibility score is a number that tells you how often tools like ChatGPT or Gemini mention your brand when someone asks for a product recommendation in your category, and right now more Australian shoppers are opening those tools before they ever type a word into Google. Think about the last time you asked an AI tool to suggest a gift, a supplement, or a pair of runners. It gave you three or four brand names, not a page of blue links. If your store wasn’t one of the names, you were invisible to that buyer, and they never even saw a search results page where you might have ranked.
The score itself is simple once you strip away the jargon. It measures two things: how often your brand gets mentioned across a batch of real buyer questions, and how well those mentions describe you, your products, your price point, your point of difference. A low score usually means the AI tools don’t have enough clear, specific information about your store to repeat back to a shopper. A high score means your brand shows up, gets described accurately, and sits next to names your customers already trust. You can work out a rough version of your own score this afternoon, using nothing but a notebook and the same AI tools your customers are already using.
Table of Contents
Why AI Search Changed What Visibility Means
AI Overviews, ChatGPT, Gemini and Perplexity now sit between your product page and the shopper. They answer questions like “best reusable water bottle under $40” or “where to buy natural skincare in Australia” with a direct answer, not a page of options to click through.
Old visibility was about where you sat on page one. A number three ranking meant steady clicks, and you could track it with any rank tracker. AI visibility works differently, because there’s no position to track, only a yes or no on whether your brand got named in the answer.
| Signal type | What it measures | Where it lives |
|---|---|---|
| Traditional SEO | Page rank for a keyword | Search results page |
| AI visibility | Brand mentions inside an answer | Chat or overview response |
| Traditional SEO | Click-through rate | Analytics dashboard |
| AI visibility | Accuracy of product detail repeated | The AI’s wording itself |
This shift matters for product-based businesses because an AI answer usually names three or four brands and stops there, according to AI Search Visibility in 2026: The Complete Guide. Ranking tenth on Google still gets you some clicks, missing an AI answer gets you none.
Work Out Your AI Visibility Score in 5 Steps
You work it out with ten prompts and a notebook, not a dashboard. Write ten buyer-style questions a real shopper would type, things like “best ceramic dinnerware brand Australia” or “where to buy ethical homewares Brisbane,” then run each one through ChatGPT, Gemini and Perplexity. For every answer, note whether your brand was named, where it sat in the list, and whether the details were right.
Score five categories out of 20 points each, for a total out of 100. A fictional store, Clay & Co, a Brisbane homewares retailer, ran this test and landed on 54 out of 100.
| Category | Clay & Co score (out of 20) |
|---|---|
| Mention rate | 10 |
| Position in answer | 8 |
| Accuracy of detail | 14 |
| Sentiment | 12 |
| Citation source | 10 |
That tally took one afternoon and a browser tab for each tool. An automated tool would run this across dozens of prompts in minutes, but a hand count today still gives you a real number to act on, which is more than most store owners have.
What Each Part of the Score Is Actually Measuring
Each category answers a different question about how an AI model sees your store, not just whether it noticed you. Mention rate is the blunt one, it just counts how many of your ten prompts returned your brand name at all. Position matters almost as much, because being named fourth after three competitors reads very differently to a shopper than being named first.
- Mention rate: appearances across all ten prompts, out of 20 points
- Position: named first versus buried after competitors
- Accuracy: correct trading hours, product range, delivery area or location
- Sentiment: whether the mention is favourable, neutral or mixed
- Citation source: whether the AI pulled from your own site, a review platform, or a directory
Accuracy catches stores out more than they expect, because an AI model often repeats an old address or a discontinued product line it found somewhere once. Sentiment and citation source together tell you whether the AI trusts your own words or someone else’s, which changes what you fix first.
What Your Score Range Actually Means
A score sits in one of three bands, and the band tells you what kind of problem you’ve got. Zero to 40 is low visibility, 41 to 70 is partial visibility, and 71 to 100 is strong visibility. Clay & Co’s 54 sits in the partial band, which means it shows up sometimes but doesn’t yet win the comparison.
- Low (0 to 40): thin product pages, no structured data, no reviews cited by name
- Partial (41 to 70): some mentions, but patchy accuracy or weak positioning
- Strong (71 to 100): named consistently, described correctly, cited from a trusted source
Traditional rankings can hold steady for months once you earn page one. AI visibility moves faster, because a model can shift which brands it favours after a single content update, and a score can swing ten or more points between monthly checks, according to AI Visibility Score: How to Measure and Interpret It.
Fixing a Low AI Visibility Score
A low or partial score points to specific gaps, not a vague content problem. Start with structured data and clear service or product pages, since AI tools lean on schema markup and plain product descriptions to work out exactly what you sell.
- Add structured data and specific product or service pages so models can parse your range
- Get listed and reviewed on directories and platforms AI tools actually cite
- Correct any wrong detail an AI keeps repeating, like an old address or a discontinued line
- Re-run the same ten prompts every month and track the shift in each category
Most of this overlaps with ordinary local SEO work, which is the starting point for stores before they add on AI SEO as a separate layer. The fourth step matters most, because a score with no retest is just a one-off guess.
Doing It Yourself Or Getting a Full Audit
Ten prompts in an afternoon gives you a real number, but it’s a sample of one afternoon, one person’s wording, one location. A model can answer differently for a shopper in Perth than one in Melbourne, and ten prompts won’t catch that spread.
A professional audit runs the same scoring logic across hundreds of prompts and checks your structured data, your citations and your factual details in one pass, instead of you clicking between browser tabs for an hour. That matters most for stores with more than a handful of product lines, where a manual check can miss which categories are weak and which are fine.
| Approach | Time cost | Coverage |
|---|---|---|
| Manual notebook scoring | About an afternoon | Ten prompts, one run |
| Spreadsheet tracking monthly | A few hours each month | Same ten prompts over time |
| Professional audit | Done for you | Hundreds of prompts, structured data and citations checked together |
If you’ve only got one product line and a few minutes to spare, the manual version is honestly enough to start. If you’re managing a catalogue with dozens of product categories and multiple store locations, the gaps in a hand count start to cost more than the time it saves, and that’s closer to the scale a local SEO audit is built to check.
Pick a Week to Run the Ten Prompts
Open a notebook this week and run the ten prompts before you touch a single product page. Write down the score, then put a date next to it, because that’s your baseline for next quarter, not just a one-off grade. Changing your descriptions or your schema before you’ve scored anything means you won’t know what actually moved the number. If you run the test monthly and it starts eating an afternoon you don’t have, email hello@scalecurrent.com.au and ask for a free AI check instead of letting the tracking lapse.
Want more customers from Google and AI search? Tell us about your business.
Common questions
Can two people get different scores testing the same business?
Yes, and that’s normal. AI tools don’t give the same answer twice in a row, so your mention rate or position can shift a point or two between two people running the same prompt. The gap matters less than the pattern. If both testers land in the same band, low, partial or strong, you’ve got a reliable read on where you stand.
Does a low AI visibility score mean my Google rankings are also bad?
Not necessarily. A store can rank on page one of Google and still get skipped in an AI answer, because the AI pulls from different signals, mainly structured data, reviews and how clearly your product pages describe what you sell. Plenty of well-ranked stores score low here simply because their pages were never written with an AI model in mind.
How often should I re-run the test to see if anything improved?
Monthly is the pace worth sticking to. AI answers can shift after a single content update on your site or a competitor’s, so a quarterly check misses the swing and leaves you guessing which fix worked. If the monthly habit starts eating time you don’t have, email hello@scalecurrent.com.au and ask about a free AI check instead.
What if AI tools mention my competitor instead of me even when I rank higher on Google?
That’s the most common reason stores go looking at this score in the first place. It usually means the competitor’s product pages or reviews give the AI clearer, more specific detail to repeat, even if your Google ranking is stronger. Check their citation source against yours, since that often shows exactly what the AI is copying from.
