The verdict in three sentences
Trade finance banks, counterparties and graduate recruits now ask AI assistants like ChatGPT to summarise a Hong Kong trading house before a first call, and the answer often shapes the tone of that call. Whether your firm appears, and whether what is said is accurate, depends on consistent public information, credible third-party mentions and clear compliance pages, far more than on marketing copy. The work is an emerging market, roughly 2,000 to 6,000 USD for a diagnostic and 1,500 to 5,000 USD a month for ongoing work (2026 ballpark), and it is measured with a dated record of answers to the same questions.
What counterparties, banks and recruits now ask
A commodity trading house has almost no consumer audience. The people who type questions about you are relationship managers, compliance officers, mine owners, shipowners, journalists and candidates, and they increasingly start with an AI assistant rather than a search engine. The questions below are realistic examples for a Hong Kong house; yours will differ by commodity and route.
| Who asks | Typical question in 2026 | What the answer tends to draw on |
|---|---|---|
| Trade finance banker | "Which Hong Kong commodity traders publish a sanctions compliance policy?" | Compliance page, annual or ESG report |
| Steel mill procurement head | "Iron ore traders with offices in Hong Kong and Singapore that sell on CFR terms" | Offices page, trade directories, press |
| Mine or smelter seller | "Who buys copper concentrate out of Hong Kong and how do they usually pay?" | Products page, published interviews |
| ESG analyst or NGO | "Does this trading house publish a responsible sourcing policy for cobalt?" | Responsible sourcing page, reports |
| Graduate candidate | "Commodity trading graduate programmes in Hong Kong that hire Cantonese speakers" | Careers page, recruitment profiles |
| Journalist | "Who owns this firm and where is it incorporated?" | About page, public registry data |
| Mainland buyer, in Traditional Chinese | The same supplier question, asked in Chinese | Chinese pages, if they exist and match |
The last row matters in Hong Kong. If your English and Traditional Chinese pages say different things about offices, products or ownership, the answers in each language drift apart.
Why some trading houses appear and others do not
AI assistants assemble answers from what is publicly and consistently written about a firm. For trading houses, a few patterns explain most of the gap:
- Consistency of basic facts: the same legal entity name, offices, commodities and leadership on your site, your professional network profiles, directories and press releases. Contradictions make an assistant hedge or skip you.
- Third-party mentions: trade press coverage, association memberships you really hold, conference speaker pages and bank or partner announcements that are already public. Never invent a counterparty or a volume to look bigger.
- Clear factual pages: what you trade and on which routes, governance, a summary of your sanctions and KYC approach, responsible sourcing, and ESG reporting exactly as published. Vague pages produce vague answers.
- Structured data and clean architecture: organisation details, addresses and leadership marked up so machines read them without guessing.
- Freshness: dated news and the latest report online. An assistant that only finds a 2021 brochure describes the firm of 2021.
Careers and onboarding forms collect personal data, so Hong Kong's Personal Data (Privacy) Ordinance, overseen by the PCPD, applies to how you collect and keep it; your counsel sets the wording. This is not legal advice.
What the work covers and what it costs
In broad terms the work means checking what assistants say today, correcting and enriching the public facts about the firm, aligning languages and tracking changes over time. It is not a recipe, and nobody controls what an assistant says. Prices below come from typical agency quotes and public agency rate pages; the market is young and ranges are wide.
| Work item | What it covers (broad terms) | 2026 ballpark (USD) |
|---|---|---|
| Initial diagnostic | A fixed set of 20 to 40 questions, answers recorded and assessed | 2,000 to 6,000 |
| Factual page improvements | Governance, compliance, offices and products pages clarified | 3,000 to 10,000 one-off |
| English and Traditional Chinese alignment | Matching facts across both languages, translation by third party | 1,500 to 5,000 |
| Third-party profile clean-up | Directories, profiles and outdated listings corrected | 1,000 to 4,000 |
| Ongoing monitoring and updates | Monthly re-runs of the question set, fixes as facts change | 1,500 to 5,000 per month |
| Local Hong Kong agency retainer | Broader digital retainer that may include this work | HKD 20,000 to 60,000 a month (about 2,600 to 7,700 USD) |
At Kolonell this work is priced on quote after a diagnostic, because the number of questions, languages and third-party profiles varies widely between a 15-person physical trader and a multi-office group.
How results are measured
Rankings do not exist in AI answers, so measurement relies on a fixed set of questions asked on a schedule and logged with the date, the language and the full answer.
Need a professional website?
Kolonell builds websites that attract clients, optimized for the Sénégalese market. Free quote in 2 minutes.
| Metric | How it is read | Example reading |
|---|---|---|
| Mention share | Share of questions where the firm is named | 6 of 40 questions, so 15 % |
| Accuracy | Share of mentions with correct facts (offices, commodities, ownership) | 4 of 6 mentions accurate |
| Language parity | Gap between English and Traditional Chinese answers | Named in English, absent in Chinese |
| Sources cited | Which pages or third parties the answer leans on | Old brochure cited twice |
| Freshness | Age of the most recent fact repeated in answers | Latest fact dates from 2023 |
| Business signal | Inbound inquiries mentioning an AI assistant as the source | "How did you hear of us" field |
The point of the log is comparison: the same questions, months apart, show whether the public picture of the firm became clearer.
Mini case study
Illustrative example: "Kowloon Bay Resources" is a fictional metals trader with 40 staff. A diagnostic of 30 English and 10 Traditional Chinese questions finds the firm named in 6 of 40 answers (15 %), with 2 of those 6 giving an office address it left years ago. The board approves a diagnostic at 4,000 USD, page improvements at 6,000 USD and six months of monitoring at 2,000 USD a month. Total: 4,000 + 6,000 + (6 x 2,000) = 22,000 USD, or about 3,700 USD a month over the six months. The success test is set in advance: accurate facts in every answer that names the firm, and a measured change in mention share between the first and sixth monthly log, with no outcome promised.
FAQ
Can a trading house be sure an AI assistant will recommend it?
No. Nobody controls these answers, and any provider promising a recommendation should be avoided. What can be improved is the accuracy and consistency of public facts, measured over 40 or so fixed questions.
Is this relevant if we do almost no marketing?
Yes, because the people asking are banks, sellers and candidates, not consumers. A single inaccurate answer about ownership or sanctions policy can cost a week of extra questions in onboarding.
How much should we budget in Hong Kong?
A diagnostic typically costs 2,000 to 6,000 USD and ongoing work 1,500 to 5,000 USD a month (2026 ballpark). Local retainers that bundle it run around HKD 20,000 to 60,000 a month.
Do we need Traditional Chinese pages?
If mainland or local Chinese-reading buyers and candidates matter to you, yes, and they must match the English facts exactly. A gap between languages shows up quickly when the same 10 questions are asked in both.
How long before we see a change?
Expect to compare logs over 3 to 6 months, since assistants refresh what they draw on at their own pace. The first gains are usually in accuracy rather than in the number of mentions.
Find out how AI assistants describe your firm. A dated diagnostic of 30 to 40 questions in English and Traditional Chinese shows what counterparties read about you today. 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.
