The verdict in three sentences
For a licensed fintech or digital asset firm in Tokyo, SEO remains the foundation: institutional clients and sophisticated users still check registration, custody and team on Google, in Japanese and in English. AEO/GEO (visibility in the answers of AI assistants like ChatGPT) becomes a rational second investment once the brand already exists, because buyers now put their due diligence questions to an assistant before they ever open your site. Either way, measurement has to be dated and verifiable: Search Console and qualified leads on one side, a dated record of AI answers to the same questions on the other.
What your buyers actually ask Google and AI assistants
A buyer in this sector rarely starts with your name. A corporate treasurer looking for a custody provider, a foreign family office seeking access to the Japanese market or an experienced user comparing platforms types very specific questions: 'which crypto asset exchanges are registered with the FSA', 'institutional digital asset custody Japan', 'yen stablecoin for B2B payments', 'settlement API for fintech in Tokyo'.
On Google, those queries lead to pages: your licence pages, your compliance FAQ, your product pages. In AI assistants like ChatGPT, they produce a synthesized answer that sometimes names two or three firms and leaves others out. If your registration number, custody model and security policy are not stated clearly and consistently on your site and on reliable third-party sources, the assistant has nothing solid to repeat.
Regulation shapes both channels. In Japan, crypto asset exchange service providers must be registered with the Financial Services Agency (FSA) under the Payment Services Act, and the industry body JVCEA also oversees how its members communicate. Data protection falls under the APPI, supervised by the Personal Information Protection Commission (PPC). In practice, every piece of content must show the registration reference exactly as held, carry risk warnings and never promise returns. This is not legal advice: your compliance team signs off every piece before it goes live.
SEO vs AEO/GEO: the comparison for a Tokyo fintech
The figures below are 2026 ballpark ranges based on typical agency quotes in Tokyo, converted at an approximate rate of 150 JPY to 1 USD.
| Criterion | SEO (Google, Japanese and English) | AEO/GEO (answers from AI assistants like ChatGPT) |
|---|---|---|
| What it targets | Rankings on licence, custody, API and comparison queries | Presence and accuracy of the firm in answers to buyer questions |
| Time to visible effect | 4 to 9 months on an established domain | 2 to 6 months to see a change in recorded answers |
| Monthly market budget | About 450,000 to 1,350,000 JPY (approx. 3,000 to 9,000 USD) | About 300,000 to 900,000 JPY (approx. 2,000 to 6,000 USD) |
| Main metrics | Impressions, clicks, qualified leads, demo requests | Mention rate, accuracy of cited facts, sources picked up |
| Main risk | Non-compliant content published too fast | Wrong licence details repeated by an assistant |
| Dependence on brand | Medium | High: a rarely cited brand rarely appears |
The point that often surprises executives: AEO/GEO does not replace SEO. AI assistants lean heavily on well-indexed pages and third-party sources. A fintech that is invisible on Google has little chance of being described accurately in an assistant's answer.
Where to put the budget, by situation
| Situation | Recommended priority | Indicative budget split | Why |
|---|---|---|---|
| Recently registered firm, unknown brand | SEO first | 80% SEO, 20% AEO/GEO | You first need to exist on licence and product queries |
| Established brand in Japan, weak in English | English SEO, then AEO/GEO | 60% SEO, 40% AEO/GEO | Foreign buyers ask AI assistants in English |
| Heavily regulated content, long sign-off | AEO/GEO focused on facts | 50% / 50% | Few pages, but accurate and consistent facts everywhere |
| Limited budget, under 3,000 USD a month | Technical SEO and licence pages | 90% SEO | Fix the basics before any work on AI answers |
| B2B product launch (custody, API) | Product SEO plus AEO/GEO record | 65% SEO, 35% AEO/GEO | Due diligence questions reach assistants quickly |
| Wrong information already repeated by an AI | Urgent AEO/GEO | 30% SEO, 70% AEO/GEO | A licence error costs more than a lost click |
How to measure without fooling yourself
On the SEO side, the toolkit is familiar: Search Console for impressions and clicks by query, your CRM for leads, and above all the rate of qualified leads, because a traffic spike on consumer queries is worthless for an institutional custody offer.
On the AEO/GEO side, the only honest measure is a dated record. You fix a list of 30 to 60 real buyer questions, in Japanese and English, put them to AI assistants like ChatGPT at regular intervals, and log for each answer: is the firm cited, are the facts accurate, which sources are used. No serious provider promises that an AI will recommend your firm. What you can track is how presence and accuracy change over time.
An initial diagnostic usually covers: the technical state and indexing of the site in both languages, the consistency of licence information across the site and third-party sources, a first dated record of AI answers, and a recommended budget split. At Kolonell, that diagnostic is on quote.
Need a professional website?
Kolonell builds websites that attract clients, optimized for the Sénégalese market. Free quote in 2 minutes.
Mini case study
Illustrative example. Hikari Custody Partners, a digital asset custody firm in Tokyo, has 6,000 USD a month. Its initial record covers 40 questions in English: the firm is cited in 4 answers, or 10%, and two answers use a former trading name.
It splits the budget 60% SEO (3,600 USD) and 40% AEO/GEO (2,400 USD). Six months later, a new record on the same 40 questions shows 9 mentions, or 22.5%, with no naming error. On the SEO side, the institutional custody page generates 14 qualified enquiries over the half year. On 36,000 USD invested, the cost per qualified enquiry comes to about 2,571 USD, to be weighed against the value of a single custody mandate.
FAQ
Should we stop SEO to fund AEO/GEO?
No. AI assistants rely heavily on well-indexed pages and third-party sources. In most cases AEO/GEO accounts for 20 to 40% of the visibility budget, not more at the start.
How long before we see an effect in AI answers?
Allow 2 to 6 months to see a measurable change in a dated record. Effects are gradual and vary from one question to another.
Can we mention returns in our content?
Any promise of returns should be avoided and risk warnings must be clearly displayed. Your compliance team approves each page, which often adds 1 to 3 weeks to production.
How do we know if an AI misdescribes our licence?
Only a dated record shows it: you ask the same 30 to 60 questions at fixed intervals and check every fact cited. An error on the registration reference should be handled first.
Should the diagnostic cover Japanese and English?
Yes if you target foreign buyers. Answers often differ between languages, and a record limited to one language can hide 50% of the problem.
Find out how AI assistants describe your firm. We record how AI assistants like ChatGPT describe your firm in Japanese and English, then size the right split between SEO and AEO/GEO. Request a diagnostic quote: reply within 48 hours.
Mohamed Bah
Fondateur, Kolonell
Passionate about digital and entrepreneurship in Africa, Mohamed has been helping Sénégalese businesses with their digital transformation since 2020. Founder of Kolonell, he believes every SME deserves a professional and accessible online présence.