Connecting Systems That Don't Talk: A Plain Guide to Integrations

Somewhere in your business there is a ritual. An order lands in one system, and a person, probably a good one, retypes it into the accounting package, then updates the stock sheet, then maybe pastes the delivery address into the courier portal. Nobody designed this. It accreted, one tool at a time, and now it is just how Tuesday works. When people say "our systems don't talk to each other", this ritual is what they mean, and the person performing it is the integration.

The software industry's word for the fix, integration, manages to sound both technical and vague at once, which suits the people selling it. So here is the plain version: an integration is plumbing that moves data from one system to another so a person does not have to. That is all it is. What matters, and what this article is actually about, is that there are only four ways to get that plumbing, they have very different price tags, and the cheapest adequate one wins.

The four ways, cheapest first

The first option is the native integration: the connection one of your vendors already built. Your accounting package probably connects to your bank; your webshop probably connects to your payment provider. These are usually free or nearly free, maintained by somebody else, and boring in the best way. This is the honest starting point, and any consultant who does not begin by asking "have you checked what your tools already connect to" is skipping the step that might make them unnecessary. If a native integration exists and moves the fields you care about, use it and stop reading.

The second is a connector platform. Zapier, the biggest, claims more than 10,000 apps in its directory by its own count, and the pitch is real: when something happens in tool A, the platform does something in tool B, no code involved. For standard tools and standard fields, this works, and we have told plenty of prospects to just use it. The catches are the ones we covered when we priced this model in what automation actually costs: you pay by the task forever, so the meter follows your growth, and the connector covers the standard fields, which are not necessarily your fields. The moment your process needs the connector to understand your custom field, your approval quirk, your weird-but-profitable exception, you are working around the tool you pay for.

The third option almost never appears in articles like this, because nobody profits from it: the scheduled export-import. Once a week, a file goes out of one system and into another, possibly with a human pressing two buttons. If the data only needs to be right weekly, this is not a hack, it is the correct answer, and we have recommended it with a straight face to businesses that were being quoted five figures for real-time sync they did not need. Real-time is a requirement, not a virtue. Ask what actually breaks if the systems agree tomorrow morning instead of now; for a surprising amount of back-office data, the answer is nothing.

The fourth is a custom bridge: a small, purpose-built piece of software that reads from one of your systems and writes into another, shaped exactly around your fields and your rules. This is the connective tissue we argued for in the build-vs-buy piece, and it is most of what we build. It is the right answer when your combination of systems is not in anyone's directory, when the data needs transforming or validating on the way through, when volume makes per-task pricing silly, or when one end of the pipe is a scanner, a local database, or a machine on your own network that no cloud connector can see. It costs real money once: our builds start at €3,500 fixed, and most projects land between €8,000 and €30,000, fixed, with the code owned by you. In our projects, an integration bridge on its own usually sits at the small end of that band, because it is a small thing; it just has to be exactly your shape.

The question that actually matters

Buyers ask "is there an integration between X and Y", get a yes, and sign. The better question has four parts: does it move my fields, in my direction, at my volume, and how will I know when it stops?

The last part deserves its own paragraph, because it is the one nobody demos. The worst state for plumbing is not broken; it is silently broken. Picture it: a sync fails in March, quietly, because a password expired or a field was renamed, and nobody notices until the stocktake in June, when two systems that were supposed to agree turn out to have been drifting apart for a quarter. The cleanup costs more than the integration did. So when you evaluate any of the four options, the grown-up questions are the failure ones: what happens to a record that does not fit, where do errors land, who gets told, and would anyone notice within a day? A connector that fails loudly is worth more than a cleverer one that fails politely. This is the same review-queue logic we apply to document flows into accounting systems: the doubtful case must go somewhere a human looks.

How to choose without a consultant

Work the options in order. Check what your tools already connect to natively, and test whether the native option carries your actual fields, not just the demo fields. If that fails, try the connector platform against one real week of your data, watching for the fields and exceptions it cannot express. If you catch yourself building a spreadsheet on the side to patch what the connector misses, you have your answer: the standard plumbing does not fit your building, and the choice is now between a custom bridge and the humble weekly export. That choice is volume and urgency, nothing more mystical: high volume or same-day needs point to the bridge, low volume and patient data point to the export.

And if you do end up at the custom bridge, the de-risking rules from the build-vs-buy piece apply in miniature: fixed price, small scope, your name on the code, and loud failure built in from day one.

If you are staring at two systems that refuse to talk and you want to know which of the four options fits, describe the two systems and the retyping ritual between them and we will tell you, in plain terms, which shape of plumbing it needs, including the three shapes we do not sell. If it turns out a native integration or a weekly export solves it, that call cost you thirty minutes and saved you a project; how we scope and price the bridges we do build is on the custom software page.