First Page, OOm, Impossible Marketing, Clickr and MediaOne compared on one criterion you can check yourself

These five firms are the ones Singapore buyers keep asking about, and most comparisons of them rest on self reported summaries and star ratings. This page compares them on a single criterion that anybody can re run: how often ChatGPT and Gemini named each firm when a Singapore buyer asked who to hire, with no firm named in the question. Across 64 blind readings taken on 22 October 2026, First Page appeared in 17, MediaOne in 12, Impossible Marketing in 11, OOm in 10 and Clickr in 2.

We reproduce other vendors' prices here with the page and the date we read them. That changed on 26 October 2026, after a finding from 17 September 2026 that an engine discounted undated competitor prices. Their fees belong on their pages with their dates on them, and a quote you were given this week beats any figure we could print.

The one criterion, and its limits stated first

What the count measures is machine visibility for Singapore buying questions on two engines on one date. What it does not measure is delivery quality, retention, how they treat a small account, or whether any of them is right for you. Those are answered by references and a paid audit, not by a count. Read the table as a map of who your buyers are being told about.

Firm	Named in, of 64 readings	Its own site appeared in	What the answers put it next to
First Page	17	7 readings	Performance search and paid acquisition. The most consistent name across both engines.
MediaOne	12	6 readings	Search and content for small and mid sized Singapore businesses.
Impossible Marketing	11	4 readings	Search, with a narrow and visible service focus.
OOm	10	6 readings	Search and paid media, often beside enterprise ecommerce work.
Clickr	2	Not recorded separately	Occasional appearances alongside the four above.

The spread is narrower than it looks. First Page at 17 and OOm at 10 are not different leagues; they are seven readings apart out of 64, on one day, on two engines. The honest reading is that four of these five occupy roughly the same position and one appears much less often.

The method, so you can reproduce or refute it

33 batches of buying questions, put to ChatGPT and Gemini, with Singapore written into the text of every question. The market had to go in the prompt because the search interface we use accepts a country, a language and a location parameter, returns HTTP 200 for all of them, and ignores every one. The snapshot taken that evening holds 64 blind readings, 32 per engine, with the brand batch excluded because it names us in the question. Our own domain was cited zero times and named zero times in all 64.

How far apart these five really are

Seven readings out of 64 separate the most named firm from the fourth, which on a single days reading across two engines is a narrower gap than a ranked list makes it look. Treat the table as two groups rather than five positions. Four firms occupy roughly the same space in these answers and one appears much less often, and no reading on one date can order four firms that close together. If you want a tighter answer, run the same questions on five consecutive days and look at which names hold.

The nine questions to put to all five

1. Who specifically works on my account, at what seniority, for how many hours a month, and who covers them on leave.
2. What share of those hours sits outside Singapore, and in which country.
3. Is the quoted fee inclusive or exclusive of GST, and what is your GST registration number.
4. Whose name is the Google Ads account, the analytics property and the domain registered in, and what happens to each if I leave.
5. What is the notice period, and what is the minimum term.
6. Show me a dated before and after for a client in my sector, with the window stated.
7. If you sell AI search work, show me the question list you measure, the engines you cover and two readings of the same list on different dates.
8. What will you refuse to do, and what have you talked a client out of this year.
9. Which of my current pages would you leave alone.

Question seven is the one that separates genuine work from a relabelled retainer, and it is answerable in a sentence by anybody who actually does it. Question nine is the one that tends to produce the most honest conversation.

What the comparison cannot settle for you

Fit, which is mostly about account size. A firm that is excellent with a S$30,000 monthly media budget may staff a S$4,000 retainer with its most junior person, and that is not a criticism of the firm, it is a mismatch. Ask each one directly where your budget sits in its client book. The answer to that question predicts your experience better than any ranking on this page.

Questions about choosing between these five

Is the most named firm the best choice

Not necessarily, and the count is not built to answer that. It tells you which name a buyer is most likely to be handed by a machine, which is a fact about visibility.

Why is there no price comparison here

Because a price without a date and a source is a claim with no shelf life, and we hold ourselves to what we ask of everybody else. Ask all five for a quote in the same format on the same day and you will have a better table than any published one.

Should I shortlist all five

Three is usually enough, and the shortlist should include at least one firm that is not on this page. A specialist in your sector that the engines have not noticed yet is often the better value purchase.

Do these counts mean the other Singapore agencies are weak

No. It means they were not named in these 64 readings on this date. In 26 of those 64 readings the engines named nobody at all, and in the other 38 they named one of a small group, so absence from this particular list is not a verdict on a firm.

How do I check this myself

Ask both engines your own category question with Singapore in it, five times each, and write down who gets named and which sources appear. Half an hour gives you a reading you own, and our guide to running a repeatable check on whether a business appears in an AI answer sets out how to keep it consistent between runs.

The comparison none of the five currently publishes

Every firm on this page publishes about search. None of them publishes a dated reading of what the engines say about their own clients' categories before and after the work, which is the single piece of evidence a buyer in 2026 most needs and the hardest to fake. We publish ours, including the uncomfortable half: zero namings and zero citations for our own domain across these 64 Singapore readings. AI Knows Us is the measurement layer for exactly that, across ChatGPT, Gemini, Perplexity, Claude and Google AI Overviews, and it sits alongside an agency rather than instead of one. Ask whichever of the five you shortlist to bring their own reading to the first meeting.