---
title: "The integration tax"
url: "https://hurukan.com/blog/the-integration-tax"
description: "Every company knows what its software costs. Almost none can say what the space between the systems costs, and it is usually the larger number."
---

# The integration tax

August 21, 2026·2 min read·hurukan

A finance director can tell you what each system costs to the euro. Licences, hosting, support, the implementation partner. What nobody can tell you is what the seams cost — the exports, the reconciliations, the person who knows which of two numbers is the real one.

That work has no line item because it is not anybody’s project. It is distributed across dozens of people in small amounts, and the total is almost always larger than the software it connects.

## Where the tax is paid

**Re-entry.** The same fact typed into a second system because the first cannot tell the second. Individually two minutes, collectively a role.

**Reconciliation.** Two systems disagree and somebody establishes which is right. This is routine, scheduled, and treated as normal rather than as a defect.

**Waiting.** A step that cannot start because the record has not propagated. Overnight batches are the classic case: work that could happen at four in the afternoon happens at nine the following morning.

**The meeting.** The weekly session that exists to assemble a picture no system holds. Count the people in it and multiply by the hours, then ask what it would cost if the picture simply existed.

**Connector maintenance.** Every integration is a small piece of software with no owner, which breaks whenever either side changes, and which is discovered to have broken by a downstream consequence rather than by an alert.

**Judgement about trust.** Experienced staff know which fields in which systems are reliable. That knowledge is real, valuable and undocumented, and it leaves with them.

## Why it never gets fixed

**It is invisible.** No single instance is worth mentioning, so no instance ever gets escalated.

**Fixing it is not a project.** Replacing a system is a project with a start and an end. Removing a seam is a structural change touching two owners who each think the other should move.

**The vendors are not incentivised.** Every product benefits from being the system of record and none benefits from making itself easy to leave. Integration standards exist and are implemented exactly as far as procurement demands.

**People adapt.** The organisation routes around the seams so effectively that the cost stops looking like a cost and starts looking like the job.

## Measuring it

Pick one process end to end — enquiry to cash, or order to delivery. Walk it with the people who do it, and write down every point where information moves between systems and a human is involved.

For each, record how long it takes and how often it happens. Multiply. The number that comes out is the integration tax on one process, and most companies find it exceeds the licence cost of the systems involved.

Then ask the harder question: how much of the delay in that process is decision time, and how much is transfer time. In most companies the second is larger, which means the organisation is slow for reasons that have nothing to do with how quickly anybody decides anything.

## What this argues for

Not better integration. Fewer things to integrate.

One record does not remove work because it is more advanced; it removes work because the transfer step stops existing. That is a less exciting claim than most enterprise software makes, and it is the whole of the argument.

-   integration
-   data silos
-   cost
-   enterprise systems

## Keep reading

-   [A system of record cannot act](https://hurukan.com/blog/a-system-of-record-cannot-act)
-   [Intelligence has a lawful meaning and an unlawful one](https://hurukan.com/blog/intelligence-has-a-lawful-meaning)