7 Questions to Ask Any AI Consultancy Before Hiring in Hong Kong

Under the AI boom, many more companies now call themselves AI consultancies. Uneven quality is a reality of the industry. For Hong Kong SMEs, hiring the wrong consultancy does not just waste money. It also drains the team's confidence in transformation. This article gives you 7 questions to ask before you meet a consultancy. You can then quickly tell who has genuinely done the work and who is simply talking.

Choosing an AI consultancy comes down to 7 questions: do they have proof from their own business, can they speak honestly about failure, do they advocate training or selling tools, what are the deliverables, what is the fee structure, how is effectiveness measured, and what remains with your team after the engagement ends. Only those who answer specifically are worth talking to further. DKH is a Hong Kong AI-first digital consultancy, founded in 2022, validating its methodology through its own brand.

TL;DR

• 7 questions covering: proof, honesty, philosophy, delivery, pricing, measurement, continuity
• Biggest red flag: the consultant's main income comes from reselling tools, or they cannot name a single failure case.
• What good consultancies have in common: they can spell out specific processes and numbers, and they will tell you what is not right for you.
• Conflict of interest declaration: DKH is itself a consultancy, and this checklist can be used to assess us too.

Question one: Have you used AI to run a real business yourselves?

Teaching people to swim and having swum yourself are two different things.

Ask a consultant whether they have run AI across a full workflow in their own business. Ask for real examples. DKH's answer is our own brand Skin Nods: brand, website and ecommerce, content and advertising, the whole chain handled in-house. We validated the AI workflow against real money, real stock and real customers before bringing it to clients. You do not need a consultancy to own a brand. But you do need evidence that they have encountered the pitfalls themselves.

Question two: Tell us about one case where you failed or could not deliver.

Be cautious of any consultant who cannot answer this.

Anyone who has genuinely done the work will have failure stories. They will also be willing to share them. What you want to hear is something specific: "we got this wrong in this situation, and here is how we addressed it." That is very different from "every project we run is a success." This question tests honesty. Honesty is the most valuable quality in a consultancy relationship.

Question three: Do you advocate training first or tools first?

This is a philosophy lie detector.

If a consultancy's proposal centres on "using the tool stack we recommend," the training may amount to little more than tool instruction. In that case, it is selling a product, not transformation. Ask this as well: do you take commission from tool vendors? Commission is not necessarily a problem. Failing to disclose it is. A genuine transformation consultancy focuses on team capability and workflow redesign. It selects tools only once the needs are clearly understood.

Question four: What do we actually receive on the day the project ends?

Be specific about the deliverables list. Is it a report, a documented workflow, a team training record, or a set of configured tool accounts?

The most valuable deliverables are the capability that stays with your team and the processes written down. These are what you can still use after the consultancy has gone. Reports are not without value. However, projects that produce only a report tend to revert to the original state within 3 months.

Question 5: What is the fee structure, and what is not included?

Common structures include project-based, monthly retainer, and phased billing. Which suits you depends on your needs. Whichever it is, ask clearly: how overtime is charged, how out-of-scope requests are handled, and who covers tool costs.

Be wary of vague quotes accompanied by pressure to sign quickly. A reasonable consultancy will be willing to begin with a small scope. This allows both sides to verify the fit.

Question 6: How do you measure results, and when can we see them?

A good consultancy will agree measurable metrics with you. These might include process time, adoption rate, and quality rejection rate, followed by business numbers. They will also manage expectations honestly. Ask which changes take weeks and which take months.

If someone promises "revenue will rise by X in 3 months," ask what that figure is based on. The responsible approach is to provide reference cases and ranges. For example, DKH has worked with Hair Again since 2023. The business expanded from 7 shops to 18 (internal data during the campaign, not independently verified). Results of this kind are the outcome of years of full transformation. They are not a templated promise.

Question seven: Once you have gone, can our team run it on their own?

The ultimate goal of a transformation consultancy should be to make you no longer dependent on them for day-to-day work.

Ask about their knowledge transfer plan. Find out how training will be delivered, how documentation will be maintained, and which member of your team will be developed into an internal champion. If their model leaves you reliant on them over the long term, that is not transformation. That is outsourcing. A long-term consultancy relationship can be valuable. But it should be something you actively choose, not something engineered so that leaving becomes difficult.

Next step: hand it to us. Free 15-minute consultation, leave your details on the contact page and we will WhatsApp you within one working day.