To check whether your business shows up in ChatGPT, ask it the five prompts below and note whether your name appears in the answers. The whole test takes ten minutes, costs nothing, and works the same way on Perplexity and Gemini. Most Singapore SMEs discover they are absent from all five.
AI visibility is the degree to which AI assistants such as ChatGPT, Perplexity and Gemini mention or cite a specific business when people ask them buying questions. It differs from search ranking: a company can sit on page one of Google yet never be named by an AI assistant, because assistants compose answers from sources they select and trust.
Published 8 July 2026 ยท By Ken Tan, founder of Kaizenaire, building Singapore digital marketing and offshore teams since 2015.
Why this test matters now
Buying behaviour has shifted upstream of the click. SparkToro’s 2026 research puts zero-click searches at 68 percent, and roughly half of Singapore consumers now use AI assistants when researching what to buy. Singapore also leads the world in generative AI adoption, at about 61 percent of the working-age population. When a customer asks an assistant for a recommendation and your business is not in the answer, you were never in the running. The uncomfortable part is that most owners have simply never looked. This self-test gives you a repeatable way to look.
The 5-prompt self-test
Open ChatGPT (search mode on, if available), then ask each prompt with your own category and area substituted. Save each answer somewhere you can compare against next quarter.
- Category prompt: “I’m looking for [your service] in Singapore. What are my options?”
- Recommendation prompt: “Which [your category] in [your area or MRT station] would you recommend, and why?”
- Comparison prompt: “Compare the well-known providers of [your service] in Singapore.”
- Brand prompt: “What is [your business name], and is it reputable?”
- Objection prompt: “What should I watch out for when choosing [your category] in Singapore?”
How to read your results
Score one point each time your business is named. Five points means assistants already treat you as an answer; protect that position. One to four points means you appear on brand or category prompts but lose the recommendation ones, which is where buying decisions happen. Zero is the common score, and it is not a verdict on your business. It usually means the sources assistants lean on, such as editorial roundups, comparison articles and community threads, do not yet mention you. Repeat the test monthly, because answers shift as models and their sources update.
Why zero citations is normal, and fixable
Assistants cite what they can retrieve and quote cleanly. Ahrefs’ analysis found brand mentions correlate with AI citations at 0.664, nearly three times the 0.218 correlation for backlinks, so being talked about in credible places matters more than link-building alone. “Most Singapore businesses we test score zero on their first run. The gap is not quality, it is absence from the places assistants read,” says Ken Tan, founder of Kaizenaire. The practical fix is presence in citable editorial and community sources, plus pages structured so an assistant can lift the answer directly.
What to do with your score
Run the test before spending anything on AI-focused marketing, and keep the dated screenshots as your baseline. If you would rather have it done rigorously, Kaizenaire runs a free AI-Visibility Check that covers 20 buyer prompts across ChatGPT, Perplexity and Gemini, logged and repeatable month to month, so improvement is measured rather than claimed. There are no guarantees in this field, and you should be wary of anyone who offers one; what can be promised is a clear picture of where you stand and a plan to raise the probability of being cited. Details are on the Kaizenaire.com service page.
FAQ
Does this test cost anything? No. All five prompts run on free tiers of ChatGPT, Perplexity and Gemini.
Should I test more than one assistant? Yes. Each assistant reads different sources, so scores often differ; Perplexity in particular leans on community discussion.
How often should I re-test? Monthly. Model updates and fresh sources change answers, and a dated log is what turns anecdotes into evidence.