We sell answer engine optimisation (AEO) and generative engine optimisation (GEO) to Singapore businesses. So before we optimised anything else this quarter, we did what we would tell any client to do first: we measured ourselves. In July 2026 we asked four AI engines fifteen real buyer questions about the services we offer. Kaizenaire was cited in the answers to exactly one of them — the question that asks about us by name.
Most agencies would not publish that number. We think publishing it is the point. If we are going to sell AI visibility, you should be able to see our own baseline, our method, and whether our numbers move — and you should demand the same from anyone else pitching you this service.
How we measured
We wrote fifteen questions a real Singapore SME owner might ask an AI assistant — covering offshore staffing, AI marketing, AEO/GEO and WhatsApp chatbots. Examples:
- Best offshore staffing agency Singapore
- How much does it cost to hire offshore staff from the Philippines for a Singapore company?
- How to legally hire remote staff from the Philippines as a Singapore business
- Best AEO agency in Singapore
- How do I get my Singapore business cited by ChatGPT?
- Best WhatsApp AI chatbot provider for a Singapore business
Each question was asked three times on each of four engines — Perplexity (Sonar, with live web search), ChatGPT (GPT-5.5), Claude (Sonnet 4.6) and Gemini (3.1 Pro) via API — 180 answers in total. We counted a question as a “citation” if Kaizenaire’s brand or domain appeared anywhere in the answer. One honest caveat: the API answers from ChatGPT, Claude and Gemini test what the model already knows, not the consumer app with browsing turned on. Perplexity’s live-search results are the closest proxy to what a real user sees today.
The results
| Engine | Questions where Kaizenaire was cited | Which ones |
|---|---|---|
| Perplexity (live search) | 1 of 15 | Only “What is Kaizenaire and is it legit?” |
| ChatGPT (GPT-5.5) | 1 of 15 | Only the branded question |
| Claude (Sonnet 4.6) | 1 of 15 | Only the branded question |
| Gemini (3.1 Pro) | 1 of 15 | Only the branded question |
On the one branded question, we were cited in all three runs on every engine. On all fourteen generic buyer questions — the ones that actually bring in new customers — we appeared in zero of three runs, on every engine, every time. There is no way to spin that, so we won’t.
Who gets cited instead
Because Perplexity shows its sources, we could see exactly who was answering our buyers’ questions in our place: job platforms, business directories, LinkedIn pages, Reddit threads, marketing agencies’ blogs — and, in several answers, competitors’ own websites. The pattern is consistent: AI engines cite sources that are visible and corroborated across the wider web, not just websites that describe themselves well.
What this actually tells us
Three things, and they match what we tell clients. First, AI visibility is mostly an authority problem, not a markup problem — schema and clean structure are necessary, but no amount of on-page polish makes an engine cite a business the rest of the web never mentions. Second, branded visibility and generic visibility are different games: an engine can describe you accurately when asked about you by name and still never recommend you to a stranger. Third, this takes months. Model knowledge lags; live-search engines move first. Anyone promising fast guaranteed citations is describing a service that does not exist.
What we are doing about it
We are running the same programme we sell: restructuring our service pages so each answers a real buyer question directly; fixing the technical basics; building the third-party presence AI engines actually cite; and re-measuring monthly against this exact fixed question set. We will publish the follow-up numbers whether they improve or not.
If you are evaluating AEO or GEO services
Ask any provider — including us — three questions: how do you measure citations, what is your own baseline, and will you show me the follow-up data? Be wary of anyone who cannot answer all three. And treat any guarantee of citations as the red flag it is: no agency controls what an AI engine says. Every business also starts from a different baseline, in a different niche, with different competition — which is why every engagement we take on is scoped and priced individually rather than sold as a fixed package.
If you want to know what your own baseline looks like, our AEO & GEO services page explains how we run this measurement for clients, and our AI marketing agency page shows how the SEO, AEO and GEO work fits together. No sales pressure — but if you ask an AI about your business today, it is worth knowing what it says.