Hong Kong

System integration in Hong Kong.

You already own the pieces — accounting, CRM, e-commerce, a warehouse system, three Google Sheets. The problem is that none of them agree. We make the tools you already pay for behave like one system.

The symptom

Somebody reconciles it by hand

Usually on a Friday. Usually the same person. Usually a full day, every month, that nobody has ever costed.

The tell is not that a system is missing. It is that the same fact — an order, a customer, a stock level — exists in four places with four slightly different values, and a human being is the integration layer holding them together.

The second tell is that nobody trusts a number until they have asked where it came from. Reports get rebuilt in a spreadsheet before anyone will act on them, which means the reporting is decorative and the real decisions are made from someone's memory.

The third is that a new tool now feels dangerous rather than useful, because everyone knows it will become a fifth thing to reconcile.

What the work actually is

Deciding is the hard part

The plumbing is the easy half. The half that takes judgement is agreeing which system is allowed to be right.

For every fact that matters — what a customer is called, what an order is worth, how much stock exists — exactly one system has to be authoritative, and every other copy has to be a follower. Most integration projects that fail were never wrong about the code; they were wrong because two systems were both allowed to be right, and the conflict surfaced six months later as a number nobody could explain.

So we start by mapping where each fact is created, who edits it, and what should happen when two sources disagree. That map is worth having even if you never build the integration, and you keep it either way.

System integration

APIs, webhooks and scheduled syncs between the tools you use, with retries, and a log you can actually read at the moment two systems disagree.

Data pipelines

Moving and reshaping data on a schedule, with validation at the edges, so bad records are caught on the way in rather than discovered inside a report.

A reporting layer

One place the numbers come from. Two people asking the same question get the same answer, and neither of them rebuilds it in Excel first.

Migrations

Getting off a system you are leaving, with the history intact and a rollback that exists in practice and not only in the plan.

Monitoring

Alerts for the sync that quietly stopped. A silent integration failure is the expensive kind, because the damage is measured in weeks of wrong data.

The manual edges, named

Where a system genuinely cannot be connected, we say so and design around it, rather than pretending a person copying values is automation.

A Hong Kong company's stack is rarely one vendor's ecosystem. It is assembled — some of it international, some regional, some inherited from whoever set up the office.

What we are usually connecting

  • Accounting and payroll, often cloud, occasionally a desktop package that predates the current finance team.
  • E-commerce and marketplaces, sometimes several storefronts with different SKUs for the same physical product.
  • Logistics and warehousing, including forwarders and third-party providers whose idea of an API is a scheduled CSV.
  • Messaging where business actually happens. In Hong Kong a genuine order frequently arrives on WhatsApp or WeChat, and a system that pretends otherwise is describing a company that does not exist.
  • Banking and payments across more than one currency, with the rate that applied on the day.
  • Spreadsheets, which are not going away and are often the only place a real rule is written down.

Naming these matters because the constraint is rarely technical. It is that the tool half your operation depends on was never designed to be integrated, and the honest options need to be on the table early.

The local stack

What is actually in the building

Assembled, not chosen. That is normal, and it is what the work has to fit.

Data protection

Integration is where data quietly spreads

Connecting two systems copies personal data into a third place. That is the moment to decide who may see it and how long it is kept.

Hong Kong's Personal Data (Privacy) Ordinance governs how personal data about customers and staff is collected, used, kept and disposed of. An integration project touches all four, usually without anyone framing it that way — a sync built to solve a reporting problem ends up putting customer records into a warehouse nobody has access rules for.

So access and retention are designed into the pipeline rather than written as a policy afterwards: which fields actually need to cross, who can read them at the destination, how long they persist, and what gets redacted in the logs. Doing this at design time costs almost nothing. Retrofitting it after an auditor asks costs a great deal.

None of this is legal advice, and where a question is genuinely legal we will tell you to ask someone qualified rather than guess on your behalf.

Questions

What people ask before starting

What does system integration actually involve?

Deciding which system is authoritative for each fact, then building the plumbing that keeps the others consistent with it — APIs, webhooks and scheduled syncs, with retries and a log you can read when two systems disagree. The deciding is the hard part; the plumbing is the easy part.

Do we have to replace our existing software?

No. Integration work exists precisely so you do not have to. The tools you already pay for usually stay; what changes is that they stop contradicting each other.

What if a system has no API?

Some do not, and that is a real constraint rather than an excuse. The options are a scheduled file exchange, a database-level read, or accepting that this system stays manual and designing around it honestly. We will tell you which one applies before quoting.

How do you handle personal data?

Hong Kong's Personal Data (Privacy) Ordinance governs how customer and staff data is collected, used and kept, and an integration is exactly where data quietly spreads into places nobody intended. Access rules and retention are designed into the pipeline rather than added afterwards.

Related

Where this connects

Custom software development in Hong Kong

When the tools disagree because the one you actually needed was never built.

Workflow automation in Hong Kong

Once the systems agree, the next question is why a person is still moving work between them.

Integrations & data

The same work in the context of everything else we do.

Systems consultancy

A systems audit first, if you are not yet sure integration is the right spend.

Which number does nobody trust?

Tell us what gets reconciled by hand each month and which systems disagree. We will tell you what is worth connecting and what should be left alone.