What we measured
- Test prompts
- 110
- AI models
- Gemini, ChatGPT
- Personas
- 5 personas across 4 languages: retail switcher, SME owner, cross-border sender, freelancer, security-conscious new user
- Competitor benchmark
- 6: two global neobanks, one transfer specialist, two incumbent banks, one local challenger
Results after 90 days
| Metric | Start | Day 90 | Change |
|---|---|---|---|
| AI Visibility Score | 29/100 | 52/100 | +23 pts |
| Prompt Impression Rate | 31% | 54% | +23 pp |
| Share of AI Answer vs competitors | 12% | 26% | +14 pp |
| Semantic Match Rate | 45% | 71% | +26 pp |
| Narrative Consistency | 33% | 69% | +36 pp |
| Brand Recommendation Rate | 22% | 38% | +16 pp |
| Content Gap Index | 67/100 | 36/100 | -31 pts |
Key findings
AI's description changed by language: a trusted challenger in one market, "limited / regional" in another.
The brand contradicted itself across markets it was trying to win.
The most critical-toned answers came in the Germanic-language prompts.
Strong markets read coolest, suppressing recommendation.
Recommendation share sat near the bottom of the peer set.
A solid product, but AI lacked confirming signals.
Safety, licensing and deposit-guarantee questions were answered vaguely.
AI hedged on trust, the gate to a financial recommendation.
The 90-day plan
Month 1: narrative foundation
Define one cross-market narrative with the same trust facts (licensing, deposit guarantee) in every language; remove the drift.
Month 2: content for AI intents
Per-market, per-persona decision content: fees, best-app-for, and safety/regulation explainers, localized for each language.
Month 3: external signals
Local fintech media in each market, comparison sites, trust references, a cross-border money report, monthly re-measurement across languages.
Before and after
| Area | Before | After 90 days |
|---|---|---|
| Visibility | Visible at home, thin elsewhere. | Surfaces across markets and personas. |
| Narrative | A different story in each language. | One consistent story: a licensed digital bank. |
| Competition | Global neobanks owned the shortlists. | Holds its place as a regional challenger. |
| Content | Marketing pages, vague on trust. | Content answers safety, fees and persona intents per market. |
| Evidence | Few local signals confirming trust. | Local media and comparison sites raise confirmation. |
Method. Anonymized case; the client is withheld under NDA, while the industry, scope and competitive set are real. Baseline figures are drawn from Rankfor's AI-visibility index measurements in fintech and digital banking (CEE and Nordic); the 90-day figures reflect the trajectory the Rankfor program is built to produce. We report AI visibility and understanding, kept separate from sales outcomes.
