Follow
Follow The Observatory
Video hosts and directories recur in Singapore’s education results. Direct schools and a university also reach the front of the observed answers, while specific questions reveal another cast of sources.
Quick answer
Which sources and institutions appear in Singapore’s Google and AI results?
YouTube ranks first in 100% of Singapore’s 12 published Google AI Overview citation tables (12/12) from June 22–September 20, 2026. That repeated host-level position is only one part of the education record: SAIS and Nanyang Technological University each head separate AI Overview opening tables, and AIS appears directly in citation and query results. The wider picture combines recurring information sources, direct institutions and query-specific results, rather than one school leaderboard.
At a glance — key facts
- Cited hosts recur. YouTube heads every published citation table in the stated period, while Skoolopedia appears throughout that citation record.
- Institutions can open the answer. SAIS and NTU each take first place in different August opening tables.
- Organic recurrence has exceptions. Reddit usually heads the published organic tables; a school directory and Facebook head the other two.
- Specific questions change the view. An IB query names AIS first by median position; a school-fees query returns a school in Johor.
Repeated hosts do not make schools absent
The citation sequence runs from June 22–28 to September 14–20. YouTube is first in each of its 12 published top-five tables, with Skoolopedia included every time. The result establishes recurrence of the host and the discovery source, without identifying the creators behind every YouTube citation.
Direct institutions qualify any broader conclusion that education is represented only by platforms and directories. SAIS heads the August 3–9 opening table; NTU heads August 10–16. These are first positions in a different surface from citations, not exceptions to YouTube’s citation count. They show that a named institution can lead the opening even while a hosting domain leads cited support elsewhere in the record.
AIS also appears directly in the September 14–20 citation selection, present in all seven runs. A host-heavy citation table can therefore contain a school’s own identity. “Platforms or institutions” is too simple a choice for the evidence.
Three surfaces, three first names
One shared window makes the difference visible without mixing dates. In September 14–20, Reddit is first in organic search, Skoolopedia first in AI Overview openings and YouTube first in citations. AIS appears in all three tables, at positions 4, 4 and 5 respectively.
Singapore education, Google desktop, September 14–20, 2026: seven comparable runs and 105 queries per run. Each row preserves that table’s published order.
| Observed surface | First entry | What the ordered table measures |
|---|---|---|
| Organic search | reddit.com | Frequency of holding the top organic position |
| AI Overview opening | Skoolopedia | Cumulative opening volume |
| AI Overview citations | youtube.com | Cumulative citation volume |
The aligned dates make this a useful local contrast; the different measures prevent adding the positions or percentages into a single score. The Citation Surface reference explains why supplying cited support is a separate form of presence from opening an answer.
The following opening table, September 21–27, puts Facebook first and Skoolopedia second. Both displayed shares round to 11%, which does not establish an exact underlying tie. There is no citation table for that later week in this archive, so YouTube’s earlier citation position cannot be carried forward to it.
Reddit’s organic recurrence has two informative exceptions
Reddit heads 8 of the 10 published organic top-five tables from June 22–28 through September 21–27, 2026, and appears in all 10. The two exceptions are International Schools Database for August 17–23 and Facebook for August 31–September 6. The exceptions change which source is first, while keeping discovery and hosted information prominent.
That broad pattern does not answer every school question. The exact-query observations make the difference concrete:
| Google organic query | Repeated result in the published top-five readings | What becomes visible |
|---|---|---|
| “best international schools Singapore” | Good Schools Guide has median best position 1 in four readings, June 15–August 30 | A source helping readers discover and compare schools |
| “leading IB curriculum schools in Singapore for university placement” | AIS has median best position 1 in three readings, July 13–September 13; present in 7/7 runs in each | A direct institution in a programme-related question |
| “premium international school fees Singapore breakdown” | chis.edu.my has median best position 1 in three readings, July 6–September 6 | A school domain whose institution is based across the border |
The complete proof for these rows is the June, late-June, July and August discovery readings; the July, August and September IB readings; and the July, August and August–September fees readings. Position is the median when present, not a claim of being first in every run. CHIS appears in 4/7, 7/7 and 5/6 runs respectively; its official address places it in Johor, Malaysia.
For an education team, the broad panel and an exact curriculum question can therefore give different accounts of the same institution’s visibility. A platform’s recurring aggregate position cannot settle whether a school appears for the particular question the team cares about.
Model accounts can change the subject—or narrow an official framework
The narrative record adds a different kind of evidence. Google AI Mode in web mode on June 29, 2026 frames much of the subject through AI adoption and technology, while also mentioning NUS, SMU and MOE. The August 10 web reading again emphasizes technology but includes Premium Tutors. Those education examples matter: the accounts shift emphasis without eliminating education altogether.
Claude Sonnet 4.6 in native mode on September 21 instead discusses established institutions and less publicly documented tutor, parent and admissions networks. This records the model’s framing of visibility and familiarity; it does not measure those networks or establish their educational value.
A narrower statement can be checked against an official source. The July 27 Claude Sonnet 4.6 native reading describes EduTrust in administrative and financial terms that omit part of its scope. The official scheme criteria also cover academic systems and student or graduate outcomes. That discrepancy concerns the model’s description of the framework, not the certification of a named school.
“Singapore’s education record combines recurring information sources with direct institutional appearances. The question and surface determine which part of that picture is visible.”
The Observatory
Together, these observations distinguish three issues an institution might otherwise conflate: appearing under its own name, having material cited through a host, and being described accurately in a model account. The local evidence identifies those observed states; it does not identify a school-specific cause or prescribe a response.
Questions about Singapore’s education visibility
Does YouTube’s first position mean schools were not cited?
No. AIS appears in the September 14–20 citation table alongside YouTube and other sources. The host-level count also leaves the underlying video creators unresolved.
Can institutions lead the AI Overview opening?
Yes. SAIS and NTU each head a separate opening table in August. These dated examples qualify the wider prominence of hosts and directories without changing the citation result.
Does the IB-query result establish university-placement performance?
It establishes AIS’s observed organic position for that exact wording. The query’s reference to placement is not a measured student outcome.
Do these observations show a June-to-September improvement?
They show recurrence, different first positions and distinct model accounts. Changes in aggregation, incomplete family coverage and rotating model/mode combinations prevent a normalized improvement or decline claim.
Evidence basis and limits
The Singapore pack, closed on September 29, 2026, contains 91 published readings: 10 organic, 13 opening, 12 citation, 14 exact-query, 29 model Lens and 13 Model Watch publications. This synthesis checks every numerical table across that record. Watch publications reuse model captures and are not extra independent experiments.
Organic and opening coverage runs from June 22–September 27; citations from June 22–September 20; exact-query readings from June 15–September 20. Not every family has a publication for every intervening week. Narrative coverage reaches September 28, but that final Gemini 2.5 Pro native capture has no same-date partner and does not refresh any measured surface. The narrative examples discussed above are June 29, July 27, August 10 and September 21.
Early and later tables use different aggregation rules. The percentages here count published tables, not a share of all answers, users or queries. School and domain labels have not been merged into group totals. Method’s comparability test governs any stronger historical comparison.
The Observatory does not claim to measure real-world revenue, market share, or business leadership unless explicitly stated. It publishes observed AI and search positioning within a defined scope and time window.
Intelligence by AAI Kernel. Published by The Observatory.
Continue the reading
Attributing these appearances requires resolving which school, group, place or source is represented; the Singapore identity context makes those local distinctions. The Premium Education context explains how institutions and the services around study become different subjects across destinations. Because the EduTrust example concerns a model’s account of an official framework, Official and Narrative Layers supplies the distinction needed to assess that claim.
Related reading
Singapore education readings · How to read an AI market ranking
