Ask what an integration costs and you usually hear "it depends". It does depend, but on a short list of things you can assess yourself before you speak to anyone.
This article does not give you a price. It gives you what sets the price, a way to check whether the work pays for itself using your own numbers, and the questions to ask before you commit.
Key PointThe number of exceptions, not the number of transactions, is what makes an integration complicated.
What drives the cost
- How many systems, and in which direction. Sending orders one way from a shop to the accounts is a smaller job than keeping two systems in agreement both ways. Two-way sync means deciding what wins when both sides change the same record.
- Whether the systems have a real API. A documented interface is quicker to build on and sturdier in use than file exports or screen automation. You can usually find out in an afternoon.
- Data volume and history. Moving today's transactions is one job. Bringing years of history, with its corrections and oddities, is another.
- How many exceptions you have. The trade customer invoiced monthly, the product sold by weight, the supplier who still sends a PDF: each is a branch to build, test and maintain.
- Who cleans the data. An integration copies disagreements faithfully. Someone has to decide which system is right and fix the rest first.
Clean data matters legally too when it is about people: the ICO's accuracy principle requires personal data to be accurate and, where necessary, kept up to date.
The cheap option that costs more
The lowest quote often treats the happy path as the whole job. It shows a record moving across and stops there.
What it leaves out is what happens when a record does not move: the rejected order, the duplicate customer, the blank field. Microsoft's guidance on the retry pattern describes how a well-built application handles short-lived failures. Ask any supplier what happens on failure, and who finds out.
Watch OutA connection with no error handling can fail quietly for weeks. Ask what happens when a record does not go through.
Work out whether it pays back
You need three numbers, all of which you already have:
- The hours. How long the re-keying, reconciling and chasing take each week, including the checking.
- What an hour costs you. The real cost of the person doing it, or what you would rather they did.
- What errors cost. How often two systems disagree, and what the last mistake cost in refunds, credits or goodwill.
Multiply the first two, add the third, and turn it into a yearly figure. That is what the problem costs whether you fix it or not, and any quote can be judged against it.
It also tells you when not to integrate. If a manual step takes twenty minutes a week and never goes wrong, leave it alone.
Pro TipWork out what the manual process costs you each year before you ask anyone for a quote.
What you are actually buying
- Numbers you can trust. The same figure means the same thing everywhere.
- Errors that surface at once. You hear from the system, not from a customer.
- Digital links for VAT. HMRC's VAT Notice 700/22 says it does not consider copy and paste to be a digital link.
- Something to build on. Once systems talk properly, reporting and automation get easier.
Questions to ask before you commit
- What happens when a record fails to transfer, and who finds out?
- Is it one-way or two-way, and what wins when both sides change?
- What history is included, and what is not?
- Who cleans the data first, you or them?
- What happens if a vendor changes its API?
- When it is finished, can we maintain it ourselves?
Our articles on what breaks when systems don't talk and API-first integration go further. See our integration service, or book a call: as our contact page says, the call is free and it is not a sales script.