Sonar
September 18, 2026 2026-10-06 8:34Sonar
AI search intelligence for higher education
Find out what AI recommends when students don’t name you
SONAR measures how universities are discovered, evaluated and chosen across ChatGPT, Perplexity, Google AI Mode and Claude, then turns the evidence into a prioritised fix list your team can act on.
- ~70
- applicant questions per course
- 4
- AI engines measured
- Multi-day
- repeat runs for reliable results
Will Greeves, Services Director at Education Cubed, introduces SONAR
The discovery gap
Known when named. Missing when discovered.
Ask an AI engine about a university by name and it may describe the institution, course and fees accurately. Ask the open questions that create a shortlist, such as best courses with a placement year, good universities for BBB students or affordable options outside London, and the same institution can disappear.
“A known entity in an undiscovered category.”
In an anonymised UK undergraduate law audit, all four engines could answer direct questions about the university. Across repeated category questions, it never appeared for four in five prompts.
Anonymised UK provider · Undergraduate law · 70 prompts · Four engines · Discovery repeats across four days
A real read
The headline is useful.
The pattern behind it is the finding.
This is a short extract from one anonymised SONAR baseline, covering a single course. The full baseline goes much further, with the complete prompt set, every engine, repeat runs and a full prioritised fix plan.
Audit extract · Undergraduate law
Anonymised · Partial view
Discovery appearance by engine
Across 50 repeated discovery prompts
Never appearedSometimesConsistently
The missed questions clustered around value, career outcomes and broad category terms. That list, not the headline percentage, is where the fix plan begins.
The read
The university was accurately represented when found, but weakly connected to the category questions that create discovery. The immediate problem was not inaccurate brand content. It was findability, entity resolution and machine-readable course information.
Make course facts extractable
Add and validate structured course data so key facts can be lifted without inference.
Strengthen the entity connection
Improve the records connecting the institution, delivering school and subject.
Answer the missing questions
Create evidence-led content around the concerns revealed by the prompt set.
Baseline or ongoing measurement
Where are we, or has anything
actually changed?
Where are we and why?
A baseline measured over an approximately ten-day window.
- Repeated discovery questions
- Readiness assessment
- Evidence-backed diagnosis
- Prioritised action plan
Has anything changed?
Weekly capture as a rolling four-week view, with readiness reassessed quarterly.
- Movement against the previous period
- Or “no detectable change”
- Drift diagnosis before attribution
- A fixed comparable series
Start with one course
See what the engines say before a student knows your name
We will measure where the course appears across four AI engines, explain why, and give your team an evidence-backed order of work.
When you book a baseline, you receive
- Audit report and summary
- SONAR snapshot
- Owned and prioritised fix list
- Stakeholder presentation
- Full response workbook
- Per-engine capture logs
- Layer 2 readiness assessment
- Account-team playbook
- Frozen run manifest
Talk to us about a SONAR baseline
Tell us which course or recruitment priority you want to understand.