Integration Without the Integration Costs

Why Integration Quotes Are So Alarming

You ask three vendors to connect your e-commerce store to your accounting system. The quotes come back at RM40,000, RM85,000, and RM210,000.

For what sounds like "copy orders into invoices".

The spread is real, and it isn't primarily about quality. It's about approach. Understanding the difference lets you buy the cheap version when the cheap version is sufficient — which is most of the time.

The Four Ways to Connect Two Systems

1. Native connector

Some systems already integrate with each other. Someone else built it, tested it, and maintains it.

Cost: the subscription, often RM50 - RM500 per month.

When it works: popular system to popular system, standard fields, no unusual business rules.

Always check this first. SMEs routinely commission custom builds for connections that already exist as a product. Spend an hour searching before you spend RM40,000.

2. Integration platform

Tools in the iPaaS category let you build connections by configuration rather than code. Trigger on an event in one system, transform the data, write it into another.

Cost: RM150 - RM1,500 per month for the platform, plus RM5,000 - RM20,000 to configure it properly.

When it works: both systems have decent APIs, the logic is moderate, and you want it running in weeks.

The catch: per-operation pricing. At high volume, the monthly bill can quietly exceed what a custom build would have cost. Model your actual transaction volume before committing.

3. Custom integration

Purpose-built code that moves data between the systems.

Cost: RM15,000 - RM60,000 to build, RM100 - RM600 per month to host.

When it works: complex business rules, high volume, or systems the platforms don't support well.

The advantage: no per-transaction fees, complete control, and it does exactly what your business needs rather than approximately.

4. Scheduled file exchange

Unglamorous and frequently correct. One system exports a file on a schedule, another imports it.

Cost: RM5,000 - RM15,000.

When it works: the data doesn't need to be current within minutes. Nightly is fine for a great deal of back-office work.

Why it's underrated: it's simple, it's easy to debug, and when it breaks the failure is obvious rather than subtle.

The Question That Saves the Most Money

"How fresh does this data actually need to be?"

Real-time integration typically costs three to five times more than scheduled batch, because you now need message queues, retry logic, ordering guarantees, and monitoring for a continuously running service.

Be honest about the requirement:

• Stock levels shown to customers on a website: genuinely needs to be near real time
• Orders into accounting: nightly is almost always fine
• Customer records into a CRM: hourly is generous
• Sales into a management dashboard: daily is fine
• Payroll data: monthly, obviously

Most SMEs specify real-time out of habit and pay several times over for a benefit nobody uses.

Where the Cost Actually Goes

When a vendor quotes RM85,000 for something that sounds simple, the money is usually here:

Error handling. The happy path is perhaps 20% of the work. What happens when the target system is down, a record is rejected, a duplicate arrives, or a field contains something unexpected? That's the other 80%.

Data mapping. Your store calls it "customer name". Accounting wants "first name" and "last name" separately. Someone has to define every one of those rules, including the messy ones.

Reconciliation. How do you prove nothing was lost? A serious integration includes a way to check that both sides agree.

The legacy end. If one system is modern and one is fifteen years old, essentially all the cost is at the old end.

If a quote seems high, ask which of these it includes. If a quote seems suspiciously low, ask the same question — you may be buying only the happy path.

Practical Ways to Reduce the Bill

1. Reduce scope to what you actually use. You don't need all forty fields. Sync the twelve that someone reads.

2. Accept one-way where one-way suffices. Bidirectional sync requires conflict resolution and is markedly more expensive. Ask whether data genuinely needs to flow both ways.

3. Batch it. As above. This is the single biggest lever.

4. Standardise your data first. If the same customer exists three times with different spellings, no integration will fix that — it will faithfully replicate the mess. Clean up first; it's cheaper than building around the problem.

5. Start with one connection. Not five. Prove the pattern, learn what you got wrong, then extend.

6. Own the code. If you commission custom work, make sure the contract gives you the source. Otherwise your cheap integration becomes an expensive dependency.

A Sensible Decision Sequence

1. Does a native connector exist? Use it.
2. Is the volume low and the logic simple? Use an integration platform.
3. Is nightly acceptable? Use scheduled file exchange.
4. Is it high volume, complex, or genuinely real time? Build it custom.

Working down that list in order will save most Malaysian SMEs a substantial amount of money, because most stop at step one or three.

The Ongoing Cost People Forget

Integrations aren't build-and-forget. APIs change, credentials expire, systems get upgraded.

Budget 10-15% of the build cost annually for maintenance, and make sure someone is alerted when a sync fails. An integration that silently stopped three weeks ago is worse than no integration, because everyone still trusts the numbers.

The Short Version

Check for an existing connector first. Ask honestly how fresh the data needs to be — the answer is usually "less fresh than you think". Sync only the fields you use. Start with one connection.

Do that and a problem that was quoted at RM85,000 frequently turns out to cost RM15,000.

Got an integration quote you aren't sure about? Send it to us — we'll tell you what's genuinely necessary and what's padding, whether or not you hire us.