GEMS in AI search.
South African medical-aid scheme tracked in the SA-AEO-Bench citation pilot.
Brands AI groups GEMS with.
Each peer is a medical aid brand that surfaces in roughly the same AI search queries as GEMS. Edge density reflects how often AI engines mention them in the same response.
Net narrative control, own-domain share, complaint exposure — the live numbers are one scan away. Run the free Scorecard on GEMS for the full Q2 figures, or book a 30-minute walkthrough to discuss the strategy with our team.
The observation.
Across the AI search citations we measured for GEMS this quarter, the story splits between own-domain coverage, editorial pickups and complaint-site signal. GEMS’s net narrative control sits on the negative side — complaint-site citations outweigh own and editorial coverage. Run the live Scorecard scan for the exact figures, the per-engine breakdown and the brand-versus-industry gap — or book a 30-minute walkthrough to discuss the strategy with our team.
Where AI is looking.
The cited domains, grouped by category. Where a single source over-indexes, it disproportionately shapes how AI engines describe GEMS to users.
- complaintsboard.com
- africacomplaints.com
How the engines differ.
Each LLM retrieves and ranks sources differently. A remediation that lifts visibility in one model may produce no measurable change in another.
GEMS’s exact own / complaint / gap on each of GPT-5, Claude and Gemini — plus the moves that lift each engine differently — are in the live Scorecard scan.
Run the live scan →Competitive context.
GEMS measured against the other medical aid brands in our quarterly index. Click through for each brand's full report.
Discovery
Momentum
Medihelp
GEMS
What we’d do.
Three remediation actions ranked by projected lift to net narrative control. Projections from our remediation model, not commitments — actual lift depends on execution quality.
HelloPeter sub-1h response SLA on top 12 open GEMS complaints.
Action 1 is the move we’d open the conversation with. Actions 2 and 3 — and the projected lift on each — are part of the 30-minute brand-specific walkthrough. Bring your PR partner, we’ll work through the full plan together.
Common questions,
answered straight.
How is "net narrative control" calculated?+
Net narrative control = own-domain citation share minus complaint-site citation share, expressed as percentage points. Range is −100 to +100. It describes the citation pattern observed for a brand across measured AI search queries; it is not a judgement of the brand itself.
Where does the data for GEMS come from?+
Each quarter we run 10+ AI search queries about GEMS across ChatGPT, Claude, and Gemini. The citations the engines return are extracted, deduplicated, and categorised against a published source taxonomy. The full methodology is at /methodology.
How do I dispute or correct a number on this page?+
Use the "Request a correction" form in section §6 (or email legal@citedbrands.co.za for larger disputes). We review every correction within 5 business days and publish updates alongside the original record.
Is Cited Brands affiliated with GEMS?+
No. Cited Brands (Pty) Ltd is an independent research and SaaS company. We have no commercial relationship with GEMS unless explicitly stated. Trademark and brand names are used here for purposes of comment, criticism, research, and reporting in line with section 12 of the SA Copyright Act 98 of 1978 and nominative trademark fair use.
Request a correction.
We review every correction within 5 business days. Larger disputes: legal@citedbrands.co.za.
See your own brand's scorecard, then
ship the playbook to change it.
The Cited Brands platform runs this measurement for your brand every week, then executes the recommended remediation actions for you.