In this article (16 sections)
Short answer: as a 2026 market reference, a simple one-way integration between two systems costs USD 800 to 2,500, a two-way sync of several records costs USD 2,500 to 6,000, and a complex one —a legacy system without an API, heavy business rules or high volume— costs USD 6,000 to 20,000. What moves the number is not how much data travels: it's how cooperative the system on the other side is.
"Make it talk to the other system" is the sentence that breaks more software budgets than any other. It sounds like plugging in a cable. It is, almost always, the hardest part of a project to estimate. This piece breaks down what's inside an integration —between your system and QuickBooks, Salesforce, HubSpot, Stripe or Shopify, for example— so you can read a quote and tell whether the number makes sense.
The ranges first
Type of integration | 2026 market reference | Typical example |
|---|---|---|
Simple | USD 800 to 2,500 | One direction, one record type, a modern documented API: every invoice you issue goes to QuickBooks |
Medium | USD 2,500 to 6,000 | Two directions, several record types, periodic sync: Shopify and your back office sharing products, stock, prices and orders |
Complex | USD 6,000 to 20,000 | A legacy system without an API, business rules in the middle or high volume: a fifteen-year-old desktop ERP talking to a new platform |
The jump between categories is not set by data volume. It's set by what the other system offers and by how many decisions sit in the middle.
If the integration is part of a larger build, it's priced inside that project. At Atomix a custom software project runs USD 5,000 to 10,000, and only exceptional, much larger projects go above USD 10,000; the integrations it includes are written into the scope before any code is written.
What actually gets priced
1. How you talk to the other system
This is the factor that moves the price the most, and neither you nor your vendor controls it.
- A modern, documented REST API. The best case. Days of work. QuickBooks Online, Stripe, Shopify and HubSpot all publish their APIs and their limits: QuickBooks Online API, Shopify API rate limits, HubSpot API usage limits.
- An old but working API. SOAP, odd formats, incomplete documentation. It can be done; it takes longer.
- No API, but the database is reachable. You read straight from the database. It works, and it's fragile: if the vendor changes a table in an update, it breaks without warning.
- No API and no access. Files are all that's left: exporting and importing CSVs through a shared folder. It's the worst case, and it's more common than people think with installed back-office systems.
Before you ask for a quote, find out what the other system offers. It's the single piece of information that changes the number the most, and your vendor can't know it without asking you.
2. Mapping the data
The boring part where half the time goes. Your system calls something a "customer"; the other one calls something similar, but not identical, an "account". You have a tax status field with five options and the other side has three.
Every difference is a business decision, not a technical one. And there are dozens of them.
3. Synchronization
When does the data travel? Real time is more expensive and more fragile. Every so often —every fifteen minutes, every night— is cheaper and sturdier, and for most cases it's more than enough.
Who wins if both sides changed? If a price was edited in both systems, which one stays? Someone has to decide, and that decision is yours.
4. Errors
This is the difference between an integration that works and one that works every time:
- What happens if the other system doesn't respond.
- What happens if it responds with an error.
- How you retry without creating duplicates. Payment providers build for this explicitly: Stripe's idempotent requests exist so a retried charge isn't charged twice.
- How you find out something failed.
That last point is the one most often skipped and the most expensive to lack. An integration without monitoring fails silently, and you find out three weeks later, when someone notices orders are missing.
The three questions that set the number
If you want an estimate before asking for a quote, answer these:
- How many record types travel? Customers, products, stock, orders, invoices, payments. Each one is its own work.
- In how many directions? A two-way integration costs well over double a one-way one, because conflicts show up.
- What does the other system offer? A documented API, a reachable database, or files.
With those three answers, any serious vendor can give you a range in one meeting.
The cost that comes after
An integration isn't a project that ends. It's a permanent dependency between two systems that evolve separately, on schedules you don't control.
Budget for periodic review. When the other system releases a new version, changes its API or modifies a field, someone has to follow. It's the same logic as maintaining any custom build, except here the trigger comes from someone else. For more on what that costs, see how much custom software maintenance costs.
When it's better not to integrate
It's worth saying, because it's the right answer more often than people think.
If the volume is low, exporting and importing by hand can be cheaper. Fifty records a month moved by one person in twenty minutes cost far less than an integration plus its maintenance. Do the math; don't assume it.
If you're replacing one of the two systems within the next year, wait. Integrating against something you're about to retire is paying twice.
If what you need is to see the data together, not move it, you may be looking for a report that reads from both sides, not a sync. That's considerably cheaper.
The sign a quote is badly put together
A round number with no breakdown for "integrating the system with the online store". That isn't a quote; it's a bet.
A serious quote says which records, in which direction, how often and what happens when it fails. If it doesn't say those four things, you can't compare it against another one, and the difference will most likely show up later, once the work has started.
If the integration is part of a system you're building, how we scope it —a free working demo in 24 to 48 hours, then fixed scope and price— is on the custom software development page.
Frequently asked questions
How much does it cost to integrate QuickBooks with another system?
It depends on direction and record types more than on QuickBooks itself, which has a documented API. As a 2026 market reference, sending invoices one way is a simple integration at USD 800 to 2,500; keeping customers, items, invoices and payments in sync both ways is a medium one at USD 2,500 to 6,000.
Is real-time sync worth the extra cost?
Usually not. Syncing every fifteen minutes or every night is cheaper and sturdier, and covers most business cases. Real time is worth it when a delay causes a concrete problem, such as selling stock you no longer have.
Why does a two-way integration cost more than double a one-way one?
Because conflicts appear: when the same record changes on both sides, you need rules for which version wins, and code that applies them and reports exceptions. A one-way integration never has that problem.
What should an integration quote include?
Which records travel, in which direction, how often, and what happens when it fails, including retries without duplicates and an alert when something breaks. A quote without those four items can't be compared.
When is a manual export cheaper than an integration?
When the volume is low or one of the two systems is about to be replaced. Fifty records a month handled in twenty minutes cost less than building and maintaining an integration.
If you want a number for your case, book a call.